The present ENRAC discussion about ontology of time started with
a remark by Pat Hayes on 13.3 in the discussion about the paper by
Grünwald at the Commonsense workshop, [c-fcs-98-42]. It had
been preceded by a brief discussion on a similar topic in the
discussion about the paper by Knight, Peng and Ma at the same workshop,
[c-fcs-98-183]. Furthermore, Pat Hayes had touched on the same topic
in a contribution to the panel debate on theory evaluation in
October, 1997 (see ENRAC issue for October 1997, page 83). Beginning
in March of 1998 we had an animated discussion on this topic. The debate
contributions follow.
References:
Dear Erik,
Since, as you mentioned in the Newsletter ENRAC 12.3 (98026), Pat
Hayes' opinion about instantaneous changes has a close relation to our
previous work (In fact, Pat did raise the similar question at the
Commonsense workshop), I would like to make the following
claims/arguments:
(1) For general treatment, both intervals and points are needed.
(2) To overcome the so-called Dividing Instant Problem, that is the
problem in specifying whether intervals is "open" or "closed" at their
ending-points, both intervals and points should be treated as
primitive on the same footing. Neither intervals are constructed out
of points, nor points are defined as the "meeting place" of intervals.
Points have zero duration and are non-decomposable, while interval
have positive duration and are either decomposable or non-decomposable
(moments). Intervals Meets/Met-by points or other intervals.
Therefore, although conceptually there is no definition of the
ending-points for intervals, one may still say if an interval is open
or closed at a point when the corresponding knowledge is available.
E.g., if we know interval I Meets point P1, we may say I is (right)
open at P1; if we know that interval I1 Meets Interval I2 and point P2
Meets I2, then we may say I1 is (right) closed at P2. In fact, this
interpretation is consistent with the conventional definition about
the closed and open nature of intervals that are constructed out of
points such as reals or rationals. (For full details of the
axiomatization of such a time structure based on both intervals and
points as primitive, see [j-cj-37-114]).
(3) Now consider the classical example of switching on a light. The
arguments really depend on what knowledge is given/available for such
a case. First of all, one can image there is an interval I immediately
before the switching point P, and another interval J which is
immediately after P. That is: Meets(I, P) ^ Meets(P, J) , where
HOLDS(¬ LightOn, I) ^ HOLDS(LightOn, J) . Now, what about the
switching point P?
By Commonsense, at any time, the light is either on or off, and cannot
be both on and off. In other words, one should be able to express the
example in terms of two adjacent intervals, I1 and I2, where over I1
the light is off and over I2 the light is on, that is
Meets(I1, I2) ^ HOLDS(¬ LightOn, I1) ^ HOLDS(LightOn, I2) . This
is in fact the
intention of Allen's approach, which, by excluding the concept of
points, overcomes the Dividing Instant Problem, successfully. However,
with Allen's logic, one cannot talk about anything about the switching
point P, which is intuitively there anyway.
The question now is that, by taking both intervals and points as
primitive temporal objects, on the one hand, we can talk about time
points such as the switching point P. However, on the other hand, can
we still successfully express the Commonsense knowledge for the above
example, without bearing the DIP? The answer is YES, since it really
depends on what knowledge is given/available. In fact, there are three
possible cases:
Case a) We have no knowledge about the state of the Light at the
switching point P, though we may insist that there is a switching
point, but we don't know if the LightOff interval I1, or the LightOn
interval I2 is open or closed at the switching point P. What we know
is just that the light changes from state "Off" to state "On". Hence,
such a case can be simply expressed as (1):
|
Meets(I1, I2) ^ HOLDS(¬ LightOn, I1) ^ HOLDS(LightOn, I2)
| |
Case b) We do know, or we impose (by some reason for the specified
application) that the Light is on at the switching point P, that is,
HOLDS(LightOn, P) . In this case, we still can express it as (1), but
with the additional knowledge that I1 = I , I2 = P+J . Therefore, we may
say that the LightOff interval I1 is right-open at the switching point
P, and the LightOn interval I2 is left-closed at P.
Case c) As an alternative to b), we may know, or we may impose that
the Light is still off at the switching point P, that is,
HOLDS(¬ LightOn, P) . In this case, the additional knowledge becomes
I1 = I+P , I2 = J . Therefore, we may say that the LightOff interval I1
is right-closed at the switching point P, and the LightOn interval I2
is left-open at P.
Jixin
References:
j-cj-37-114 | Jixin Ma and Brian Knight.
A General Temporal Theory.
Computer Journal, vol. 37 (1994), pp. 114-123. |
Im largely in agreement with Jixin about points and intervals, although
I dont think its got anything to do with knowledge, and I think there's
a simpler way to say it all.
First, just forget about whether intervals are open or closed.
This issue arises only if we insist (as the standard mathematical
account of the continuum does) that an interval is a set of points.
But if we take points and intervals as basic, there is no need to do
this. Points, as
Allen suggested long ago, can be thought of as places where intervals
meet each other, not as the substance out of which intervals are
constructed. It took me a long time to see how powerful this idea is.
The question of which interval 'contains' the meeting point is
meaningless. This gives a very simple, elegant formulation in which
points are totally ordered, intervals are uniquely defined by their
endpoints (which are also the points they fit between) and two
intervals meet just when the endpoint of the first is the startpoint of
the second. That's all the structure one needs. Truths hold during
intervals. One can allow instantaneous intervals, whose endpoints are
identical and which have no duration. One can, if one wishes, identify
the interval <t,t> with the point t , since such an interval makes no
'space' when interposed between two others, ie if <a,b> meets
<b,b>
meets <b,c> , then <a,b> also meets
<b,c> ; but it is also consistent,
if one wishes, to distinguish t from <t,t> , or even to forbid
instantaneous intervals completely. It is also quite consistent to have
arbitrary amounts of density, discreteness, etc.; for example, one can
say that time is continuous except in a certain class of 'momentary'
intervals whose ends are distinct but have no interior points.
(Vladimir might find these more congenial that points as the intervals
which things like flashes of lightning must occupy.)
(Instantaneous intervals have the odd property of meeting themselves,
by the way; in fact this is a way to characterise them without
mentioning points explicitly. It is also perfectly consistent to have
'backward' intervals whose end is earlier than their beginning, and
which have negative durations. Axiomatic details can be found in a
rather long document available as two postscript files
http://www.coginst.uwf.edu/~phayes/TimeCatalog1.ps
http://www.coginst.uwf.edu/~phayes/TimeCatalog2.ps
One can cast the whole theory in terms of a single three-place relation
MEETS-AT between two intervals and a point, much as Allen's original
theory can be cast in terms of MEETS.)
In this theory, to talk of the set of points 'in' an interval requires
one to specify what it means for a point to be 'in' an interval. If a
point is later than the beginning and earlier than the end, its clearly
in the interval, but we have some freedom with the endpoints. One could
insist that interval endpoints are 'in' the interval. But this is now
OK, since truths hold not at points but during intervals, so the
apparent contradiction of the light being both on and off at the
splitting point simply doesnt arise. The light isn't either on or off
at a single point: if you want to know whether the light was on or not,
you have to say which interval you are talking about. P may be true
during <a,b> but false during <b,c> , even if b is considered
to be 'in' both the intervals.
Jixin says that "one cannot talk about anything about the switching
point P, which is intuitively there anyway." Well, the point is
certainly there, and we can talk about it (for example, its relation to
other points and intervals) but the question is whether it makes sense
to say that something is true at it. Some truths may be instantaneous,
ie true only at points; others make sense only when asserted to hold
during noninstantaneous intervals. Lights being on or off, for example,
might be enduring, while changes in illumination, or isolated flashes,
can be instantaneous. So for example suppose it is dark during
interval <s,t> and the switch is hit at t . If the light stays on, we
have two meeting intervals. If the light flashes and immediately burns
out (put a 120 V bulb in a 230 V socket), one could say that there is a
flash at t , surrounded on both sides by extended intervals of darkness.
Both stories are perfectly consistent. It follows, for example, that a
random timepoint during a period of extended illumination is not a
flash, in spite of its being a timepoint at which the light is on.
Pat Hayes
After reading Pat's answers to our claims/arguments about the ontology
for time, we would like to raise the following questions/arguments:
1. First of all, it is not clear what's the exact role that time
points play in Pat's formulation, although, according to Erik's
understanding, Pat Hayes "argues in favour of an ontology for time
where intervals are the only elementary concept and timepoints play a
secondary role". As Pat points out in his answers (in agreement with
our opinion as stated in our claims), "if we take points and intervals
as basic, there is no need to do this", i.e., deal with the question
of whether intervals are open or closed. However, it is not clear
what's the exact meaning of "taking points and intervals as basic".
Are they both taken as primitive temporal objects, or, as Allen
suggests, points are thought as places where intervals meet each
other?
2. Pat argues that "the question is whether it makes sense to say that
something is true at points". However, his argument is quite
confused: in the first place, he claims "truths hold not at points
but during intervals" (as for the case when one insists that interval
endpoints are "in" the interval). Later, he states "Some truths may
be instantaneous, ie true only at points; others make sense only when
asserted to hold during noninstantaneous intervals". So, what's the
answer to the question "whether it makes sense to say that something
is true at points"?
3. Pat's claims that one may identify interval <t, t> with point
t , or
distinguish <t, t> from t , or even forbid instantaneous intervals
completely. However, what's the choice? Do we need points
(instantaneous intervals) or not? Let's consider the case that we do
(in fact, for general treatments, we do need them). For this case,
Pat's states that if
meets( <a, b> , <b, b> ) ^ meets( <b, b> , <b, c> )
then meets( <a, b> , <b, c> ) .
(In fact, it seems in Pat's formulation, we
always have
meets( <a, b> , <b, c> ) ^ meets( <a, b> , <b, b> ) etc.,
since the intervals are uniquely defined by their endpoints). Below
are some problems with this formulation:
I. As noted by Pat himself, "an instantaneous interval meets itself",
though the "basic" points are totally ordered. How to characterise the
relation between them? Pat's gives a suggestion: to characterise them
without mentioning points explicitly. Then, what's the relationship
between points and intervals?
II. How to define other relationships between intervals like those
introduced by Allen? For instance, it is intuitive to say that
meets( <a, b> , <b, c> ) ^ meets( <b, c> , <c, d> ) ·-> before( <a, b> , <c, d> ) . However, in this case, one would
have both meets( <a, b> , <b, b> ) and
before( <a, b> , <b, b> ) , and hence
"meets" and "before" would not be exclusive to each other.
III. By saying
meets( <a, b> , <b, b> ) , meets( <b, b> , <b, c> ) , and
meets( <a, b> , <b, c> ) , one can only express the first case,
that is case a), but not the other two cases, that is case b) and case c), as
we demonstrated in our former arguments.
4. Pat argues that "I'm largely in agreement with Jixin about points
and intervals, although I dont think its got anything to do with
knowledge". But it does. In fact, as pointed out by Pat himself, "if
you want to know whether the light was on or not, you have to say
which interval you are talking about". In other words, if the
(additional) knowledge of "which interval you are talking about" is
given (e.g., in terms of which interval is open/closed at the
switching point, or in terms of the corresponding meets relations -
"knowledge"?), we can say whether the light was on or not.
5. Pat also argues that his formulation is simpler (and elegant). In
what aspects, compared with which formulation? It seems that it still
needs a lot of axioms to characterise the formal structure, especially
when issues such as density, linearity, boundness, etc, are to be
addressed.
Jixin & Brian
Sorry, I wasnt sufficiently clear, and my carelessness in using intuitive
phrasing led to misunderstanding.
First, in my view there is no single answer to many of the issues that
Jixin raises. One can make various choices, each internally consistent but
not consistent with the others. (That is why I called the cited paper a
'catalog' of time theories, rather than a single theory of time.) This
freedom means that one must be clear which alternative one is using, as
confusion follows when one tries to put together bits and pieces of
incompatible views. (For example, the critique of Allen's account
by Galton in [j-aij-42-159] in 1990
(wrongly) assumes that Allen's intervals are sets of points on
the real line.) Having said this, however, there does seem to be a simple,
basic, account which can be extended in various ways to produce all the
other alternatives, and this core theory is the one I was referring to.
Second, I dont agree with Erik's introduction of my note (14.3) as putting
intervals before points. As Allen and I showed some time ago, the choice is
arbitrary, since points can be transparently defined in an interval theory
and vice versa, so the choice of either one as somehow more basic is, er,
pointless; and one gets a more useful account simply by allowing them both
as primitive. (Actually, if anything, the simple theory I outlined seems
more to rely on points as basic, since an interval there is completely
defined by its two endpoints and has no other structure, and all the
temporal relations between intervals can be inferred from the total
ordering of points.)
Jixin asks:
|
However, it is not clear
what's the exact meaning of "taking points and intervals as basic".
Are they both taken as primitive temporal objects, or, as Allen
suggests, points are thought as places where intervals meet each
other?
|
Both. These arent incompatible alternatives. The basic idea in the 'simple'
theory is essentially Allen's, that points are meeting-places. Still,
there's no harm in being able to mention these meeting-places as real
objects, and doing so makes it easier to say quite a lot of things, such as
'when' some change happens. Clock times seem to be associated more
naturally with points than intervals, for example.
|
2. Pat argues that "the question is whether it makes sense to say that
something is true at points". However, his argument is quite
confused: in the first place, he claims "truths hold not at points
but during intervals" (as for the case when one insists that interval
endpoints are "in" the interval). Later, he states "Some truths may
be instantaneous, ie true only at points; others make sense only when
asserted to hold during noninstantaneous intervals".
|
(In the above I was careless at the place marked by boldface, sorry.
I should have
said 'pointlike interval'. It gets hard to speak about this stuff clearly
in English, since I need to distinguish our intuitive notion of 'point'
from the way that a particular theory encodes this intuition, and different
theories do it differently. I will use scare-quotes to refer to the
intuitive concept.)
|
So, what's the
answer to the question "whether it makes sense to say that something
is true at points"?
|
There is no (single) answer: one can craft the theory to suit various
different intuitions on matters like these. The way I prefer, myself, is to
say that propositions hold only during intervals, so that it is simply
ill-formed to assert a proposition of a single point; but to allow the
possibility of pointlike intervals, of the form <t, t> ,
to be the temporal
durations of propositions which are (intuitively) thought of as happening
at a single 'point'. Or, put another way, some intervals may consist of
just a single point, and some points may completely fill an interval. These
pointlike intervals are the way that (this version of) the theory encodes
the times when instantaneous truths hold.
This doesnt require us to say that every point fills an interval, notice:
since 'interval' is a basic predicate, it is perfectly consistent to say
¬ interval( <t, t> ) ; this would entail, for example, that nothing
changed at that particular time. But it allows us to consider the
proposition that a tossed ball's vertical velocity is zero, and assert that
it is true at a single 'point', ie formally, that its interval of truth was
pointlike. And since it is easy to characterise pointlike in the theory:
|
pointlike(i) <-> begin(i) = end(i)
| |
one can, for example, say something like
|
illuminated(i) v dark(i) ·-> ¬ pointlike(i)
| |
so that the light is neither on nor off AT the switching point. In this
theory, every proposition has a 'reference interval' during which it is
true, and a proposition might not be true of subintervals of its reference
interval. (Though some propositions might be. This kind of distinction has
often been made in the linguistic literature. Note however that this
intuition is basically incompatible with the idea that an interval is
identical to the set of the points it contains.)
|
3. Pat's claims that one may identify interval <t, t> with point
t , or
distinguish <t, t> from t , or even forbid instantaneous intervals
completely. However, what's the choice? Do we need points
(instantaneous intervals) or not?
|
We certainly need something corresponding to 'points', I agree. I meant
only that the formal theory can be crafted in the way Ive outlined above,
or alternatively by identifying the pointlike intervals with their
endpoints, and allowing a proposition to hold at a single point. This is in
many ways more intuitively transparent but it is formally a bit more
awkward, since pointlike isnt definable any more, and one has to put in
special axioms forbidding points to meet each other. The 'reference
interval' of a proposition could now be a single point in the theory. This
is essentially the theory that Allen and I described in our 1985 IJCAI
paper [c-ijcai-85-528], though it takes a little work to see it.
|
Let's consider the case that we do
(in fact, for general treatments, we do need them). For this case,
Pat's states that
if meets( <a, b> , <b, b> ) ^ meets( <b, b> , <b, c> )
then meets( <a, b> , <b, c> ) .
(In fact, it seems in Pat's formulation, we
always have
meets( <a, b> , <b, c> ) ^ meets( <a, b> , <b, b> ) etc.,
since the intervals are uniquely defined by their endpoints).
|
Yes, exactly. Interval relations are comletely determined by endpoint
orderings,and Allen's huge transitivity table can be painstakingly derived
from the assumption of total ordering. That's all it amounts to, in fact.
|
Below
are some problems with this formulation:
I. As noted by Pat himself, "an instantaneous interval meets itself",
though the "basic" points are totally ordered. How to characterise the
relation between them? Pat's gives a suggestion: to characterise them
without mentioning points explicitly. Then, what's the relationship
between points and intervals?
|
The relations are quite simple and transparent: intervals lie between
endpoints, and points have intervals extending between them. Self-meeting
is the interval-interval relation corresponding to equality in the point
ordering. Again, if one has an intuitive objection to self-meeting
intervals, then one can take the second alternative mentioned earlier. (All
these alternatives are got by extending the basic theory.)
|
II. How to define other relationships between intervals like those
introduced by Allen? For instance, it is intuitive to say
that meets( <a, b> , <b, c> ) ^ meets( <b, c> , <c, d> ) ·-> before( <a, b> , <c, d> ) . However, in this case, one would
have both meets( <a, b> , <b, b> ) and
before( <a, b> , <b, b> ) , and hence
"meets" and "before" would not be exclusive to each other.
|
True, and indeed the Allen relations only have their usual transitivity
properties when applied to intervals which are nonpointlike and
forward-oriented. Of course both these are properties expressible in the
theory, so that the Allen transitivity relationships can be stated there,
suitably qualified. (When the alternative extension axioms are added, the
qualifications become tautologous.)
BTW, the claim that "meets" and "before" being exclusive is "intuitive"
depends on how one's intuition is formed. Part of what I learned by having
to construct alternative formalisations is that intuition is very
malleable. Having gotten used to pointlike intervals, I dont find this
exclusivity condition at all intuitive.
|
III. By saying meets( <a, b> , <b, b> ) , meets( <b, b> , <b, c> ) , and meets( <a, b> , <b, c> ) , one can only express the first case,
that is case a), but not the other two cases, that is case b) and case c), as
we demonstrated in our former arguments.
|
But these cases only make sense if one thinks of interval and points in the
usual mathematical way, which is exactly what Im suggesting we don't need
to do. We can get almost everything we need just from the ordering
structure: we don't need to get all tied up in distinguishing cases which
can only be formally stated by using all the machinery of real analysis.
|
4. Pat argues that "I'm largely in agreement with Jixin about points
and intervals, although I dont think its got anything to do with
knowledge". But it does. In fact, as pointed out by Pat himself, "if
you want to know whether the light was on or not, you have to say
which interval you are talking about". In other words, if the
(additional) knowledge of "which interval you are talking about" is
given (e.g., in terms of which interval is open/closed at the
switching point, or in terms of the corresponding meets relations -
"knowledge"?), we can say whether the light was on or not.
|
Again I was careless in using the word "knowledge", sorry. I should have
said: in order to answer the question whether the light is on or off, one
has to specify the interval with respect to which this question is posed.
On this view, the truth or otherwise of a proposition is only meaningful
with respect to certain intervals. I dont mean that the facts are
determined by knowing more about the details of the interval, but that the
question is a different question when asked about one interval than when
asked about another, and for some intervals in may be simply meaningless.
Is the light on or off at (exactly) 3.00 pm? The only way to answer this is
to find a suitable non-pointlike interval of light or darkness completely
surrounding 3.00 pm, because 'being on' is the kind of proposition that
requires a nonpointlike reference interval.
This has nothing to do with whether an interval is open or closed: in fact,
there is no such distinction in this theory. It only arises in a much more
complicated extension which includes set theory and an extensionality axiom
for intervals.
| 5. Pat also argues that his formulation is simpler (and elegant). In
what aspects, compared with which formulation?
| Perhaps I should have said, of all the various formalized temporal theories
I have ever examined in detail, which amounts now to maybe 25 or so, this
seems to distill out the essence. The others can all be described as
extensions of this one (some a little artifically, but mos tof them quite
naturally.) The conventional picture of intervals as sets of points carries
with it a lot of excess conceptual baggage, and removing this gives a
theory which is simple and intuitive (once you get used to it :-), and is a
sound formal 'core' which can be extended to give many other theories.
| It seems that it still
needs a lot of axioms to characterise the formal structure, especially
when issues such as density, linearity, boundness, etc, are to be
addressed.
| Yes; any theory needs to be extended, of course, to deal with density,
boundedness, etc., but again a merit of this very simple framework is that
it can be transparently extended in these different ways more or less
orthogonally to each other. One can establish unboundedness with one very
obvious axiom (there's always a future and past to any timepoint) and
density is also very easy. Lack of density, ie discrete time, is harder; in
fact, theres a sense in which no first-order theory can describe this,
since it assumes the integers. But again, this is a matter of adding one
(rather complicated) induction axiom, in a way that is mathematically very
ordinary. Or, alternatively, one can just assume that the integers are
defined elsewhere, and declare that every point has an integer 'date',
which gives the theory implicitly used by most temporal databases. It can
even be extended into the standard real line, if you wish, by
distinguishing 'open' and 'closed' intervals as triplets of the form
<point, interval, point> .
The theory is basically linear in its nature, since it assumes timepoints
are totally ordered. One can easily weaken it to allow partial orders, but
then the extensions involving density, etc.,, get rather tricker. I think
the universe is deterministic in any case, so linearity doesnt bother me :-)
Pat Hayes
References:
c-ijcai-85-528 | James Allen and Pat Hayes.
A Common-Sense Theory of Time.
Proc. International Joint Conference on Artificial Intelligence, 1985, pp. 528-531.
|
j-aij-42-159 | Galton.
A critical examination of Allen's theory of action and time.
Artificial Intelligence, vol. 42 (1990), pp. 159-188. |
During all the years that the debate has raged about time points vs intervals,
we devotees of the sitcalc have never seen it as an issue. Here's why I think
this is so.
In the sitcalc, a fluent ( LightOn ) has a truth value only with respect to
a situation (= sequence of action occurrences). So, we might have
|
LightOn(do(switchOn, do(switchOff, S0)))
| |
and
|
¬ LightOn(do(switchOff, do(switchOn, S0)))
| |
In the sitcalc with explicit time,
the first might become
|
LightOn(do(switchOn(3.14), do(switchOff(1.41), S0)))
| |
meaning that as a result of the action history consisting of first
switching off the light at time 1.41, then switching on the light at time 3.14,
the light will be on. Notice that there is no way of expressing the claim that
the light is, or is not on at time 3.14 (or 3.5), independently of the
situation leading up to this time. On
the other hand, time based formalisms do allow one to write LightOn(3.14) ,
without expicitly referencing, in their notation, the history leading up
to the time 3.14 at which the fluent's truth value is to be determined.
This seems to be the source of all the problems about open vs closed vs
semi-open intervals and predicate truth values over these, and also why these
seem to be non-issues for the sitcalc.
Now, one could rightly object to the above account because it provides only for
fluent truth values at discrete time points, namely at the action occurrence
times. So we are tempted to understand
|
LightOn(do(switchOn(3.14), do(switchOff(1.41), S0)))
| |
to mean that the light is on at time 3.14, but it tells us nothing about time
3.5 say. This is particularly bad for (functional) fluents that vary
continuously with time, for example, the location of a falling object. To
handle this, introduce a time argument for fluents, in addition to their
situation argument. For the light, one can write:
|
LightOnT(t, s) <-> LightOn(s) ^ t > start(s).
| |
Here, start is defined by
start(do(a, s)) = time(a) ,
where time(a) is the time at which the action
a occurs in the history do(a, s) .
An instance of this would be
|
LightOnT(t, do(switchOn(3.14), do(switchOff(1.41), S0))) <-> t > 3.14
| |
Here we have committed to the light being on at exactly the time
of the switchOn action, and forever thereafter, relative to the history
|
do(switchOn(3.14), do(switchOff(1.41), S0))
| |
In other words, provided
switchOff(1.41) and switchOn(3.14) are the only actions to have
occurred, then the light will come on at time 3.14, and remain on forever.
Notice especially that we would have both
|
LightOnT(3.14, do(switchOn(3.14), do(switchOff(1.41), S0)))
| |
and
|
¬ LightOnT(3.14, do(switchOff(1.41), S0))
| |
without contradiction. This seems to be precisely the point at which purely
time-based formalisms run into difficulties, and the sitcalc version
of this problem illustrates the role that explicit situation arguments
play in resolving these difficulties.
Now, we can axiomatize falling objects:
|
positionT(t, s) = position(s)+
| |
|
velocity(s)*(t-start(s))+0.5*g*(t-start(s))2
| |
When my car accelerates, there is a time point at which it passes 65
miles per hour. It is awkward to describe this point in a language
not providing for time points.
John McCarthy wrote:
| When my car accelerates, there is a time point at which it passes 65
miles per hour. It is awkward to describe this point in a language
not providing for time points.
| Indeed so; but all this shows is that it's awkward to combine an
interval-based approach to time with a point-based approach to other
continua (such as, for example, velocity).
Graham White
McCarthy and Hayes
(1969) used time as a fluent
on situations, i.e. time(s). One motivation was that people, and
perhaps future robots, often do not know the time with sufficient
resolution to compare two situations, e.g. Ray Reiter's recent message
with times 1.41 and 3.14. A second motivation for making situations
primary was to make it correspond to human common sense. Many people
who can reason about the consequences of actions in situations
perfectly well do not know about real numbers, and some don't know
about numbers at all. The falling body example was also in that paper
with time as a fluent. Galileo did know about real numbers.
It's not clear that either of these considerations is of basic
importance for AI.
My previous message gave a reason for including time points in a
theory of events and actions. The theory could be founded so as to
regard them as degenerate intervals, but I don't see any advantage in
that, although I suppose the idea stems from the fact that people and
robots can't measure time precisely.
Responses to Ray Reiter and John McCarthy. Ray wrote:
|
In the sitcalc, a fluent ( LightOn ) has a truth value only with respect to
a situation (= sequence of action occurrences). So, we might have
|
LightOn(do(switchOn, do(switchOff, S0)))
| |
and
|
¬ LightOn(do(switchOff, do(switchOn, S0)))
| |
In the sitcalc with explicit time,
the first might become
|
LightOn(do(switchOn(3.14), do(switchOff(1.41), S0)))
| |
meaning that as a result of the action history consisting of first
switching off the light at time 1.41, then switching on the light at time 3.14,
the light will be on. Notice that there is no way of expressing the claim that
the light is, or is not on at time 3.14 (or 3.5), independently of the
situation leading up to this time.
|
I think of Ray's 'sequences of actions' as alternative ways the temporal
universe might be, ie possible timelines (or histories, as Ray sometimes
calls them.) The point/interval controversy is about reasoning within, or
with respect to, one of these possible timelines; sitcalc gets this muddled
up with reasoning about alternative futures for the partial timeline up to
the present. (Think of the tree of accessible situations in a state's
future: the distinction is between reasoning about a single branch, and
comparing two different branches.)
| On the other hand, time based formalisms do allow one to write
LightOn(3.14) ,
without expicitly referencing, in their notation, the history leading up
to the time 3.14 at which the fluent's truth value is to be determined.
This seems to be the source of all the problems about open vs closed vs
semi-open intervals and predicate truth values over these, and also why these
seem to be non-issues for the sitcalc.
| This isn't where the difficulties lie. Even if there is only one possible
future and only one thing that could happen at each situation, these
conceptual problems about points and intervals would still arise and some
solution for them would need to be found.
|
Now, one could rightly object to the above account because it provides only for
fluent truth values at discrete time points, namely at the action occurrence
times. So we are tempted to understand
|
LightOn(do(switchOn(3.14), do(switchOff(1.41), S0)))
| |
to mean that the light is on at time 3.14, but it tells us nothing about time
3.5 say. This is particularly bad for (functional) fluents that vary
continuously with time, for example, the location of a falling object. To
handle this, introduce a time argument for fluents, in addition to their
situation argument.
|
Hold on! What kinds of things are these 'times' supposed to be? They seem
to be something like clock-times, ie temporal coordinates (maybe understood
with respect to a global clock of some kind.) OK, but notice that this
isn't what I mean by a 'timepoint'. There are at least six distinct notions
of 'time' (physical dimension, time-plenum, time-interval, time-point,
time-coordinate and duration.) I think the nearest thing in Reiter's
ontology to what I call a time-point is something like the pairing of a
clock-time with a situation ('3.14 in situation s').
|
For the light, one can write:
|
LightOnT(t, s) <-> LightOn(s) ^ t > start(s).
| |
Here, start is defined by
start(do(a, s)) = time(a) ,
where time(a) is the time at which the action
a occurs in the history do(a, s) .
An instance of this would be
|
LightOnT(t, do(switchOn(3.14), do(switchOff(1.41), S0))) <-> t > 3.14
| |
Here we have committed to the light being on at exactly the time
of the switchOn action, and forever thereafter, relative to the history
do(switchOn(3.14), do(switchOff(1.41), S0)) . In other words, provided
switchOff(1.41) and switchOn(3.14) are the only actions to have
occurred, then the light will come on at time 3.14, and remain on forever.
Notice especially that we would have both
|
LightOnT(3.14, do(switchOn(3.14), do(switchOff(1.41), S0)))
| |
and
|
¬ LightOnT(3.14, do(switchOff(1.41), S0))
| |
without contradiction...
|
This seems to be the half-open-interval solution, where intervals contain
their endpoints but not their starting points. This makes sense for the
sitcalc, which focusses on the results of actions, but seems ad-hoc and
unintuitive in a broader context. (Also, BTW, the idea that one can ever
say that some finite list of actions is all the actions that have occurred
seems quite unrealistic. After all, people's fingers probably pushed the
switch and something somewhere was generating electricity. Surely one
should be able to actually infer this from a reasonably accurate
common-sense description of light-switching.)
| ...This seems to be precisely the point at which purely
time-based formalisms run into difficulties, and the sitcalc version
of this problem illustrates the role that explicit situation arguments
play in resolving these difficulties.
| This isnt where the difficulties lie. These alternatives are obviously
incompatible if they are asserted of the same timeline (the light can't
have been both switched on and switched off at the same timepoint) and
there is no contradiction is saying that p is true at time t in one
possible timeline but not in another, as these are different timepoints.
The problem is that even if we stick to talking about a single timeline (eg
the unique past, or one alternative future) there still seems to be an
intuitive difficulty about timepoints like the time when a light came on.
The solution I suggested - that is, truth at a point has to be defined
relative to a reference interval containing the point (which is not my
idea, let me add) - is similar in many ways to Reiter's , except it applies
not just across timelines but also within a single one.
John wrote:
| When my car accelerates, there is a time point at which it passes 65
miles per hour. It is awkward to describe this point in a language
not providing for time points.
| Yes, I agree. Examples like this are what motivate the inclusion of both
points and intervals as first-class objects. It is pretty awkward to do
without points in any case if one wants to refer to the places (...that is,
the times...) where (...that is, when...) intervals meet. However, points
can be defined in terms of intervals , in principle, so having them around
is essentially a matter of convenience more than a point of basic ontology.
There are there, in a sense, whether one wants them or not.
The 65mph example is logically similar to the point at the top of a
trajectory when the vertical velocity is zero. Examples like this appeal to
a basic intuition about continuous change, that it has no 'jumps', so if it
is < x at t1 and > x at t2 , then it must = x
somewhere between t1 and t2 .
One can state this quite directly in the basic theory (For strict
first-order syntax, replace (X...) by (value X ...) ):
(continuous X i) =df
(forall (y)(implies (between (X (begin i)) y (X (end y)))
(exists t) (and (in t i) (= y (X t))))
(strictlycontinuous X i) =df
(forall (j) (implies (subint j i)(continuous X j)))
where subint is the Allen union {begins, inside, ends}. (This assumes
that the timeline itself is dense; if not, strictlycontinuous is trivially
true everywhere.) Other conditions like monotonicity and so forth also
transcribe directly from their usual mathematical formulations.
Pat Hayes
What follows is our response to the arguments about the ontology of
time from Pat Hayes, Ray Reiter, and John McCarthy.
Response to John
The example of car accelerating demonstrates the need of time points
for time ontology.
A similar example is throwing a ball up into the air. The motion of
the ball can be modelled by a quantity space of three elements:
going-up, stationary, and going-down. Intuitively, there are
intervals for going up and going down. However, there is no interval,
no matter how small, over which the ball is neither going up nor going
down. The property of being stationary is naturally associated with a
point, rather than any interval (including Allen and Hayes' moment),
a "landmark" point which separates two other intervals.
Response to Ray
| During all the years that the debate has raged about time points vs
intervals, we devotees of the sitcalc have never seen it as an
issue.
| It has already been an issue of the sitcalc. For instance, in Pinto
and Reiter's 1995 paper, Reasoning about Time in the Situation Calculus,
[j-amai-14-251], the concept
of situation is extended to have a time span (an interval?) which is
characterised by a starting time and an ending time (two points?).
During the time span of a situation no fluents change truth value.
Also, an action with duration is modelled with two instantaneous
actions (start-action and end-action). So, all the debates about time
points vs intervals apply here, and the Dividing Instant Problem
still arises. In fact, all Ray's arguments show again, with his
revised formulation of the sitcalc to accommodate temporal reasoning,
either there is no way of expressing the claim that a proposition
(like Light is on) is true or false at some time points, or one has
to artificially take the unjustifiable semi-open interval solution.
This is not surprise at all and is what one would expect, since the
problem is actually there for the underlying time theory itself, and
therefore would be there for any formalism which wants to support
explicit time representation.
Response to Pat
| ...For example, the 1990 AIJ critique of Allen's account by Galton
(wrongly) assumes that Allen's intervals are sets of points on the
real line.
| After re-reading Galton's paper [j-aij-42-159], as we understand,
Galton's arguments are in general based on the assumption that Allen's
intervals are primitive, rather than sets of points on the real line.
In fact, the main revision Galton proposes to Allen's theory is a
diversification of the temporal ontology to include both intervals and
points. That is, in Galton's revised theory, intervals are still
taken as primitive. Having pointed out this, however, as shown in Ma,
Knight and Petrides' 1994 paper [j-cj-37-114], Galton's determination
to define
points in terms of the "meeting places" of intervals does not, as he
claims, axiomatise points on the same footing as intervals, and hence
that some problems still remain in these revisions.
| there does seem to be a simple, basic, account which can be
extended in various ways to produce all the other alternatives, and
this core theory is the one I was referring to.
| Does this core theory refer that one in which "intervals are uniquely
defined by their endpoints (which are also the points they fit
between) and two intervals meet just when the endpoint of the first is
the startpoint of the second"? Or Allen's one? - It seems the former
one.
Anyway, yes. There does seem to be such a simple, basic core theory.
For general treatments, in Ma and Knight's CJ 94 paper,[j-cj-37-114] a time theory
is proposed (as an extention to Allen and Hayes' interval-based
one) which takes both intervals and points as primitive on the same
footing - neither intervals have to be constructed out of points, nor
points have to be created as the places where intervals meet each
other, or as some limiting construction of intervals. The temporal
order is simply characterised in terms of a single relation "Meets"
between intervals/points. Some advantages of this time theory are:
(1) It retains Allen's appealing characteristics of treating
intervals as primitive which overcomes the Dividing Instant Problem.
(2) It includes time points into the temporal ontology and therefore
makes it possible to express some instantaneous phenomenon, and
adequate and convenient for reasoning correctly about continuous
change.
(3) It is so basic that it can be specified in various ways to
subsume others. For instance, one may simply take the set of points
as empty to get Allen's interval time theory, or specify each
interval, say T , as <T-left, T-right>
where T-left < T-right ,
to get that one Pat prefers.
|
The way I prefer, myself, is to say that propositions hold only
during intervals, so that it is simply ill-formed to assert a
proposition of a single point; but to allow the possibility of
pointlike intervals, of the form <t,t> , to be the temporal
durations of propositions which are (intuitively) thought of as
happening at a single 'point'. Or, put another way, some intervals
may consist of just a single point, and some points may completely
fill an interval. These pointlike intervals are the way that (this
version of) the theory encodes the times when instantaneous truths
hold. This doesnt require us to say that every point fills an
interval, notice: since 'interval' is a basic predicate, it is
perfectly consistent to say ¬ interval( <t, t> ) ; this would
entail, for example, that nothing changed at that particular time.
But it allows us to consider the proposition that a tossed ball's
vertical velocity is zero, and assert that it is true at a single
'point', ie formally, that its interval of truth was pointlike. And
since it is easy to characterise pointlike in the theory:
((pointlike i) iff ((begin i) = (end i)))
|
Yes, it's true. And, it seems that, all these can be reached
equivalently by simply taking pointlike interval <t,t> as identical
with point t in the case where both intervals and points are included
in the time ontology.
|
one can, for example, say something like
((illuminated i) or (dark i))) implies (not (pointlike i))
so that the light is neither on nor off at the switching point.
In this theory, every proposition has a 'reference interval' during
which it is true, and a proposition might not be true of
subintervals of its reference interval. (Though some propositions
might be. This kind of distinction has often been made in the
linguistic literature. Note however that this intuition is
basically incompatible with the idea that an interval is identical
to the set of the points it contains.)
|
This can be distinguished by applying Holds_In and
Holds_On (that is
Allen's Holds, see Galton's paper [j-aij-42-159].
Actually, as shown in Ma and Knight's 1996 paper [j-cj-37-114],
to characterise the intuitive relationship
between Holds_On and Holds_In ,
in the case where intervals are
allowed, some extra axiom is needed.
Also, it seems that, in Pat's formulation, for expressing that
interval <a,b> is a subinterval of interval <c,d> , one would have
c < a < b < d . In this case, we get that pointlike intervals
<a,a> and <b,b> (or equivalently, points a and b )
fall in interval <c,b> .
| We certainly need something corresponding to 'points', I agree. I
meant only that the formal theory can be crafted in the way Ive
outlined above, or alternatively by identifying the pointlike
intervals with their endpoints, and allowing a proposition to hold
at a single point. This is in many ways more intuitively
transparent but it is formally a bit more awkward, since pointlike
isnt definable any more, and one has to put in special axioms
forbidding points to meet each other. The 'reference interval' of a
proposition could now be a single point in the theory. This is
essentially the theory that Allen and I described in our 1985
paper [c-ijcai-85-528], though it takes a little work to see it.
| On the one hand, many cases suggest the need of allowing a proposition
to holds at a single point. For instance, see the example of throwing
a ball up into the air described earlier in the response to John
McCarthy.
On the other hand, allowing a proposition to holds at a single point
doesn't necessarily make pointlike un-definable. It depends on if one
would impose some extra constraints, such as
((illuminated i) or (dark i))) implies (not (pointlike i))
as introduced by Pat for the light switching example, which actually
leads to the assertion that the light is neither on nor off AT the
switching point.
Actually, in the later version of Allen and Hayes's theory that appears in
1989 [j-ci-5-225], an awkward axiom is proposed to
forbid moments to meet each other. It is interesting to note that,
although moments are quite like points (moments are non-decomposable),
they still have positive duration (they are not pointlike). Moments
are included in Allen and Hayes' time ontology, while points are not.
One of the reasons that such an axiom is awkward is that it doesn't
catch the intuition in common-sense usage of time. In fact, in many
applications, one would like to take some quantity as the basic unit
of time. E.g., we may take a second as the basic unit. In other words,
seconds are treated as moments - they cannot be decomposed into
smaller units. However, for a given second, we may still want to
express the next one, that is, a second can meet another second,
although they are both non-decomposable.
| True, and indeed the Allen relations only have their usual
transitivity properties when applied to intervals which are
nonpointlike and forward-oriented. Of course both these are
properties expressible in the theory, so that the Allen
transitivity relationships can be stated there, suitably qualified.
(When the alternative extension axioms are added, the
qualifications become tautologous.)
...BTW, the claim that "meets" and "before" being exclusive is
"intuitive" depends on how one's intuition is formed. Part of what
I learned by having to construct alternative formalisations is that
intuition is very malleable. Having gotten used to pointlike
intervals, I dont find this exclusivity condition at all intuitive.
| If the exclusivity condition is not intuitive at all, as Pat's claims,
then why is is proposed for Allen's relations applied to intervals? Is
it simply because they can be easily defined as exclusive when applied
to nonpointlike intervals? Why not when applied to pointlike ones?
(Simply because it cannot be conveniently defined?). Actually, in the
case where both intervals and points are treated as primitive on the
same footing, it is straightforward to extend Allen's 13 exclusive
temporal relations between intervals to govern both intervals and
points, while without losing the property of exclusivity. Vilain's
[c-aaai-82-197] and Ma and Knight's [j-cj-37-114] systems are
two examples.
| But these cases only make sense if one thinks of interval and
points in the usual mathematical way, which is exactly what Im
suggesting we don't need to do. We can get almost everything we
need just from the ordering structure: we don't need to get all
tied up in distinguishing cases which can only be formally stated
by using all the machinery of real analysis.
| The cases make sense not only if one thinks of intervals and points in
the usual mathematical way. In fact, all the three cases are
demonstrated under the assumption that both intervals and points are
treated as primitive, rather than in the usual mathematical way.
| Again I was careless in using the word "knowledge", sorry. I
should have said: in order to answer the question whether the light
is on or off, one has to specify the interval with respect to which
this question is posed. On this view, the truth or otherwise of a
proposition is only meaningful with respect to certain intervals. I
dont mean that the facts are determined by knowing more about the
details of the interval, but that the question is a different
question when asked about one interval than when asked about
another, and for some intervals in may be simply meaningless. Is
the light on or off at (exactly) 3.00 pm? The only way to answer
this is to find a suitable non-pointlike interval of light or
darkness completely surrounding 3.00 pm, because 'being on' is the
kind of proposition that requires a nonpointlike reference
interval.
| But it seems that there are also some other kind of proposition to
which one cannot assign any nonpointlike reference interval. For
instance, in the throwing ball up into the air example, proposition
"the ball is stationary" can only be true at points, and for any point
we cannot find any non-pointlike interval (completely) surrounding it
over which the ball is stationary.
| This has nothing to do with whether an interval is open or closed:
in fact, there is no such distinction in this theory. It only
arises in a much more complicated extension which includes set
theory and an extensionality axiom for intervals.
| In the time theory where both intervals and points are taken as
primitive, we can (if we like) talk about the open and closed nature
of intervals with some knowledge being available. This kind of knowledge
can be given in terms of the Meets relation, rather than some "much
more complicated extension which includes set theory and an
extensionality axiom for intervals". In fact, we can define that:
|
interval I is left-open at point P iff Meets(P, I)
interval I is right-open at point P iff Meets(I, P)
interval I is left-closed at point P iff there is
an interval I' such that Meets(I', I) ^ Meets(I', P)
interval I is right-closed at point P iff there is
an interval I' such that Meets(I, I') ^ Meets(P, I')
|
That's all, and it seems quite intuitive. For instance, with the
knowledge MEETS(P, I) which says that interval I is immediately after
point P , one can intuitively reach that point P is on the left of
interval I and P is not a part of I , and there is no other time
element standing between P and I . Therefore, we say interval I is
left-open at point P . Similarly, with knowledge
MEETS(I, I') ^ MEETS(P, I') ,
one can reach that point P is a part and the "finishing"
part of interval I , that is interval I is "right"-closed at P .
It is important to note that the above definition about the open and
closed nature of intervals is given in terms of only the knowledge of
the Meets relation. However, if one would like to specify intervals as
point-based ones, such a definition will be in agreement with the
conventional definition about the open and closed intervals.
Jixin
References:
c-aaai-82-197 | Marc Vilain.
A System for Reasoning about Time.
Proc. AAAI National Conference on Artificial Intelligence, 1982, pp. 197-201.
|
c-ijcai-85-528 | James Allen and Pat Hayes.
A Common-Sense Theory of Time.
Proc. International Joint Conference on Artificial Intelligence, 1985, pp. 528-531.
|
j-aij-42-159 | Anthony Galton.
A critical examination of Allen's theory of action and time.
Artificial Intelligence Journal, vol. 42 (1990), pp. 159-188. |
j-amai-14-251 | Javier Pinto and Ray Reiter.
Reasoning about Time in the Situation Calculus.
Annals of Mathematics and Artificial Intelligence, vol. 14 (1995), pp. 251-268. |
j-ci-5-225 | James F. Allen and Patrick J. Hayes.
Moments and points in an interval-based temporal logic.
Computational Intelligence, vol. 5, pp. 225-238. |
j-cj-37-114 | Jixin Ma and Brian Knight.
A General Temporal Theory.
Computer Journal, vol. 37 (1994), pp. 114-123. |
John McCarthy wrote:
| McCarthy and Hayes 1969 paper [n-mi-4-463] used time as a fluent
on situations, i.e. time(s). ...
| My chief motivation for this was the fact that the mapping from situations
to times is many-one, since the sitcalc can distinguish different
situations with the same clocktime. The differences between how precisely
times are known, etc., can be handled by making the set of "times" obey
different axioms. For example, in R-sitcalc the appropriate kind of time
for a situation would be an interval of clocktimes, presumably; something
like 'from 4.12 to 5.15'.
| My previous message gave a reason for including time points in a
theory of events and actions. The theory could be founded so as to
regard them as degenerate intervals, but I don't see any advantage in
that, although I suppose the idea stems from the fact that people and
robots can't measure time precisely.
| Actually no. The motivation for introducing 'pointlike intervals' was just
to maintain a certain expressive neatness, where all propositions, even
instantaneous ones, are asserted w.r.t. a reference interval.
That idea - that intervals are approximations to points, and the length of
an interval represents a degree of ignorance about the location of a point
- gives a rather different ontology. In that case, for example, it doesnt
really make sense to be able to refer to the precise endpoints or
meeting-points of intervals (since if one can, then absolute precision
about timepoints comes for free.) The Allen set of thirteen relations
reduces to just six (before, overlap, inside, and inverses) since those
that require endpoints to be exactly identified (meets, starts, ends,
equal, endby, startby, meetby) are undefinable (except in an infinite
limit.) This is the theory called 'approximate-point' in my time catalog.
There arent any points in this theory, of course, though they could be
defined if one added enough mathematical machinery to be able to talk about
limits of infinite sequences.
Pat Hayes
References:
n-mi-4-463 | John McCarthy and Pat Hayes.
Some Philosophical Problems from the Standpoint of Artificial Intelligence. [postscript]
Machine Intelligence, vol. 4 (1969), pp. 463-502. |
Answer to Jixin's contribution to this discussion on 1.4:
There is really little point in arguing about which theories are more
'intuitive' unless one is more precise about what one's intuitions are.
There are two fundamental problems with arguments like this. First,
intuitions are malleable, and one can get used to various ways of thinking
about time (and no doubt many other topics) so that they seem 'intuitive'.
Second, our untutored intuition seems to be quite able to work with
different pictures of time which are in fact incompatible with one another.
Jixin's own intuition, for example, seems to agree with McCarthy's that
time is continuous, and yet also finds the idea of contiguous atomic
'moments' (intervals with no interior points) quite acceptable. But you
can't have it both ways: if moments can meet each other, then there might
not be a point where the speed is exactly 60mph, or the ball is exactly at
the top of the trajectory. If time itself is discrete, then the idea of
continuous change is meaningless. Appealing to a kind of raw intuition to
decide what axioms 'feel' right lands one in contradictions. (That was one
motivation for the axiom in Allen's and my theory, which Jixin found
"awkward", that moments could not meet. The other was wanting to be able to
treat moments as being pointlike. That was a mistake, I'll happily
concede.)
Jixin wrote:
| What follows is our response to the arguments about the ontology of
time from Pat Hayes, Ray Reiter, and John McCarthy.
Response To John:
The example of car accelerating demonstrates the need of time points
for time ontology.
A similar example is throwing a ball up into the air. The motion of
the ball can be modelled by a quantity space of three elements:
going-up, stationary, and going-down. Intuitively, there are
intervals for going up and going down. However, there is no interval,
no matter how small, over which the ball is neither going up nor going
down. The property of being stationary is naturally associated with a
point, rather than any interval (including Allen and Hayes' moment),
a "landmark" point which separates two other intervals.
| Yes, I agree. However, notice that there is a coherent frame of mind which
would deny this. According to this intuition, which is similar to Newton's
old idea of the infinitesimal, one would say that there are no points, but
some intervals are so small that they can be treated like points at a
sufficiently larger scale. In this perspective, it would be false to claim
that there was no interval at which the velocity is zero; rather, one
would say that the interval was infinitesimal. (If you want to deny the
reasonableness of this perspective, first reflect on the fact that it is
nearer to physical reality than any model based on the real line.)
I agree with Jixin here.
| ...For example, the 1990 AIJ critique of Allen's account by Galton
(wrongly) assumes that Allen's intervals are sets of points on the
real line.
|
|
After re-reading Galton's paper [j-aij-42-159],
as we understand, Galton's
arguments are in general based on the assumption that Allen's
intervals are primitive, rather than sets of points on the real line.
In fact, the main revision Galton proposes to Allen's theory is a
diversification of the temporal ontology to include both intervals and
points. That is, in Galton's revised theory, intervals are still
taken as primitive.
| Galton's intuitions are clearly based on thinking of intervals as sets of
points. He takes it as simply obvious, for example, that there is a
distinction between open and closed intervals.
| Having pointed out this, however, as shown in Ma,
Knight and Petrides' 1994 paper [j-cj-37-114], Galton's
determination to define
points in terms of the "meeting places" of intervals does not, as he
claims, axiomatise points on the same footing as intervals, and hence
that some problems still remain in these revisions.
|
| there does seem to be a simple, basic, account which can be
extended in various ways to produce all the other alternatives, and
this core theory is the one I was referring to.
|
|
Does this core theory refer that one in which "intervals are uniquely
defined by their endpoints (which are also the points they fit
between) and two intervals meet just when the endpoint of the first is
the startpoint of the second"? Or Allen's one? - It seems the former
one.
Anyway, yes. There does seem to be such a simple, basic core theory.
For general treatments, in Ma and Knight's CJ94 paper
[j-cj-37-114], a time theory
is proposed (as an extention to Allen and Hayes' interval-based
one) which takes both intervals and points as primitive on the same
footing - neither intervals have to be constructed out of points, nor
points have to be created as the places where intervals meet each
other, or as some limiting construction of intervals. The temporal
order is simply characterised in terms of a single relation "Meets"
between intervals/points.
| This theory seems to be similar to that outlined in my 1990 paper
with Allen, [j-ci-5-225] (and given at greater length in a U of
Rochester tech report of the same date.)
But there is little point in bickering about who said what
first, as almost all this discussion (including for example Allens 13
relations) can be found in publications written in the last century, if one
looks hard enough. All AI work in this area (including my own) is like
children playing in a sandbox. The theories and idea themselves are a much
more interesting topic.
One technical point, about 'primitive'. One of the things I realised when
working with James on this stuff was that if ones axioms about points were
minimally adequate it was trivial to define interval in terms of points;
and one can also define points in terms of intervals, although that
construction is less obvious. (I was immensely pleased with it until being
told that it was well-known in algebra, and first described by A. N.
Whitehead around 1910.) Moreover, these definitions are mutually
transparent, in the sense that if one starts with points, defines
intervals, then redefines points, one gets an isomorphic model; and vice
versa. So to argue about which of points or intervals are 'primitive' seems
rather pointless. We need them both in our ontology. If one likes
conceptual sparseness, one can make either one rest on the other as a
foundation; or one can declare that they are both 'primitive'. It makes no
real difference to anything.
| Some advantages of this time theory are:
(1) It retains Allen's appealing characteristics of treating
intervals as primitive which overcomes the Dividing Instant Problem.
| See above. But in any case this doesnt overcome the problem. Allen's
treatment allows lights to just come on, but it doesnt provide anywhere for
the ball to be motionless.
| (2) It includes time points into the temporal ontology and therefore
makes it possible to express some instantaneous phenomenon, and
adequate and convenient for reasoning correctly about continuous
change. (3) It is so basic that it can be specified in various ways to
subsume others. For instance, one may simply take the set of points
as empty to get Allen's interval time theory, or specify each
interval, say T, as <T-left, T-right>
where T-left < T-right , to get
that one Pat prefers.
| Not quite right. In my simple theory, T-left isnt
before T-right , it
equals it.
....
|
Yes, it's true. And, it seems that, all these can be reached
equivalently by simply taking pointlike interval <t,t> as identical
with point t in the case where both intervals and points are included
in the time ontology.
| Yes, that is another alternative way to formalise things.
|
one can, for example, say something like
((illuminated i) or (dark i))) implies
(not (pointlike i))
so that the light is neither on nor off at the switching point.
In this theory, every proposition has a 'reference interval' during
which it is true, and a proposition might not be true of
subintervals of its reference interval. (Though some propositions
might be. This kind of distinction has often been made in the
linguistic literature. Note however that this intuition is
basically incompatible with the idea that an interval is identical
to the set of the points it contains.)
|
| This can be distinguished by applying Holds_In
and Holds_On (that is
Allen's Holds, see Galton's 1990 paper [j-aij-42-159]).
| Yes, exactly, although there is no need to use this formal strategy, as I
explain in the time catalog section 1. Briefly, HoldsIn P i is true just
when i is a subinterval of a reference interval j
where HoldsOn P j .
Again, it is largely an aesthetic judgement, but I find Galton's
hold-on
vs. hold-in
distinction awkward and unintuitive. (It suggests that there
are two different 'ways to be true'.)
| ....Actually, as shown in Ma and
Knight's 1996 paper [j-cj-37-114], to characterise the intuitive relationship
between Holds-On and Holds-In , in the case where intervals are
allowed, some extra axiom is needed.
| I will check this paper to see what you mean in detail, thanks.
| Also, it seems that, in Pat's formulation, for expressing that
interval <a,b> is a subinterval of interval <c,d> , one would have
c < a < b < d . In this case, we get that pointlike intervals <a,a>
and <b,b> (or equivalently, points a and b ) fall in
interval <c,b> .
| Yes; but note that if the theory uses reference intervals, that fact that P
holds for an interval I doesnt imply that it holds for every point (still
less every interval) in I . So this is quite consistent:
- Ball is rising for interval <a,b>
- Ball is falling for <b,c>
- Ball is stationary for <b,b>
You can consistently add that rising and falling are true for all
nonpointlike subintervals and every properly contained subinterval of the
reference interval.
|
On the one hand, many cases suggest the need of allowing a proposition
to holds at a single point. For instance, see the example of throwing
a ball up into the air described earlier in the response to John
MaCarthy.
On the other hand, allowing a proposition to holds at a single point
doesn't necessarily make pointlike un-definable. It depends on if one
would impose some extra constraints, such as
((illuminated i) or (dark i))) implies
(not (pointlike i))
as introduced by Pat for the light switching example, which actually
leads to the assertion that the light is neither on nor off AT the
switching point.
|
In my theory it leads to the conclusion that <a,a> does not exist (or, is
not a reference interval for 'illumination'), where a is the switching
point. All the distinctions between kinds of point - ones where something
is true and ones where something is switching - can be cast into a typology
of intervals. (This example illustrates why I like to distinguish between
the point a - which undoubtedly exists, is where the intervals meet, has a
clock time, etc. - and the interval <a,a> , which, if it existed, might be
an embarrassment.)
| Actually, in the later version of Allen and Hayes's theory that appears in
1989 [j-ci-5-225], an awkward axiom is proposed to
forbid moments to meet each other. It is interesting to note that,
although moments are quite like points (moments are non-decomposable),
they still have positive duration (they are not pointlike). Moments
are included in Allen and Hayes' time ontology, while points are not.
One of the reason that such an axiom is awkward is that it doesn't
catch the intuition in common-sense using of time.
| I agree. This was awkward in our old paper, and stemmed from our reluctance
to accept the idea of intervals which could meet themselves. I'm now
reconciled to that idea: in fact, it seems inevitable, much as the
existence of zero seems inevitable once one allows subtraction.
| But these cases only make sense if one thinks of interval and
points in the usual mathematical way, which is exactly what Im
suggesting we don't need to do. We can get almost everything we
need just from the ordering structure: we don't need to get all
tied up in distinguishing cases which can only be formally stated
by using all the machinery of real analysis.
|
|
The cases make sense not only if one thinks of intervals and points in
the usual mathematical way. In fact, all the three cases are
demonstrated under the assumption that both intervals and points are
treated as primitive, rather than in the usual mathematical way.
| Well, it depends on what axioms one assumes! Perhaps I have been speaking
too carelessly about the 'usual mathematical way'. Heres my intuition: the
standard account of the continuum seems forced to resolve the dividing
point problem by deciding which interval contains the point, distinguishing
open from closed intervals, because it identifies an interval with a set
of points. (So if both intervals 'contain' the point, the intervals must
intersect.) One can take points as basic or intervals as basic or both as
primitive; that's irrelevant, but the crucial step is that (set-of-points =
interval) identification. Thats exactly what I want to avoid. My point is
only that if we abandon that idea (which is only needed for the formal
development of analysis within set theory, a rather arcane matter for us),
then there is a way to formalise time (using both intervals and points as
primitive, if you like) which neatly avoids the problem.
| ... Is the light on or off at (exactly) 3.00 pm? The only way to answer
this is to find a suitable non-pointlike interval of light or
darkness completely surrounding 3.00 pm, because 'being on' is the
kind of proposition that requires a nonpointlike reference
interval.
|
|
But it seems that there are also some other kind of proposition to
which one cannot assign any nonpointlike reference interval. For
instance, in the throwing ball up into the air example, proposition
"the ball is stationary" can only be true at points, and for any point
we cannot find any non-pointlike interval (completely) surrounding it
over which the ball is stationary.
| Yes, exactly. Some properties can accept pointlike reference intervals,
some can't. Like Galton's distinction between 'at rest' and 'motionless'
(former requires nonpointlike, latter doesnt.) Thats the point. Notice the
distinctions are now all about intervals. They arent between different ways
of being true, but are bread-and-butter distinctions between intervals,
expressible within the theory. The machinery of truth wrt an interval is
the same in both cases (and in others, eg 'intermittently true' and other
exotic variations.)
| This has nothing to do with whether an interval is open or closed:
in fact, there is no such distinction in this theory. It only
arises in a much more complicated extension which includes set
theory and an extensionality axiom for intervals.
|
|
In the time theory where both intervals and points are taken as
primitive, we can (if we like) talk about the open and closed nature
of intervals with some knowledge being available. This kind of knowledge
can be given in terms of the Meets relation, rather than some "much
more complicated extension which includes set theory and an
extensionality axiom for intervals". In fact, we can define that:
I is left-open at point P iff Meets(P, I)
interval I is right-open at point P iff Meets(I, P)
interval I is left-closed at point P iff there is an interval I' such
that Meets(I', I) ^ Meets(I', P)
interval I is right-closed at point P iff there is an interval I'
such that Meets(I, I') ^ Meets(P, I')
That's all, and it seems quite intuitive......
It is important to note that the above definition about the open and
closed nature of intervals is given in terms of only the knowledge of
the Meets relation. However, if one would like to specify intervals as
point-based ones, such a definition will be in agreement with the
conventional definition about the open and closed intervals.
|
That certainly seems to be an elegant device. (Though the definitions have
nothing to do with knowledge; all Jixin is saying is that the definitions
of open and closed can be given in terms of MEETS . As Allen and I showed in
our old paper, the entire theory can be reduced to MEETS .) However, in
order to be nontrivial, it must be that points 'separate' meetings, ie if
meets(I, P) and meets(P, J) then ¬ meets(I, J) , right? For if not, all
left-open intervals are also left-closed, etc. This seems to make 'points'
similar to our old 'moments': in fact, if Jixin's theory predicts
meets(P, Q) ·-> P = Q for points P and Q , then I'll lay odds it is
isomorphic to our moments theory. One of the main observations in our paper
was that with the no-meets axiom, one can map moments to points with no
change to the theorems provable.
On the other hand, if the theory allows distinct points to MEET , I'd be
interested to know how it is able to map smoothly to a conventional account
of the continuum, since that is provably impossible. One-point closed
intervals exist everywhere on the real line, but no two of them are
adjacent. Atomic adjacent times (whatever we call them) are pretty much a
definition of discrete time models, and are incompatible with density, let
alone continuity.
Pat Hayes
PS. Maybe the most useful thing would be to put all these axiomatic
theories into some common place with a common syntax - we could use
vanilla-KIF - so people can compare and contrast them. I dont have enough,
er, time to offer to do this right now, im afraid, but will cooperate with
anyone who will volunteer.
References:
j-aij-42-159 | Anthony Galton.
A critical examination of Allen's theory of action and time.
Artificial Intelligence Journal, vol. 42 (1990), pp. 159-188. |
j-ci-5-225 | James F. Allen and Patrick J. Hayes.
Moments and points in an interval-based temporal logic.
Computational Intelligence, vol. 5, pp. 225-238. |
j-cj-37-114 | Jixin Ma and Brian Knight.
A General Temporal Theory.
Computer Journal, vol. 37 (1994), pp. 114-123. |
Pat wrote:
| Jixin's own intuition, for example, seems to agree with McCarthy's
that time is continuous, and yet also finds the idea of contiguous
atomic 'moments' (intervals with no interior points) quite
acceptable. But you can't have it both ways: if moments can meet
each other, then there might not be a point where the speed is
exactly 60mph, or the ball is exactly at the top of the trajectory.
If time itself is discrete, then the idea of continuous change is
meaningless. Appealing to a kind of raw intuition to decide what
axioms 'feel' right lands one in contradictions. (That was one
motivation for the axiom in Allen's and my theory, which Jixin found
"awkward", that moments could not meet. The other was wanting to be
able to treat moments as being pointlike. That was a mistake, I'll
happily concede.)
| First of all, as Pat noted (see below), our theory is similar (in
fact, it is specially stated to be an extension/revision) to that of
Allen and Hayes ([c-ijcai-85-528], [j-ci-5-225]).
As an extension, it allows not only
intervals and moments, but also points (however, it dose not
necessarily have to include points or moments). As a revision, Allen
and Hayes' constraint that moments could not meet each other is
replaced by "a point cannot meet another point", that is there must
be an interval (or a moment) standing between any two points (if
these two point are explicitly expressed). However, basically, the
theory doesn't commit the time structure as (left-, right-,
left-and-right) linear, (left-, right-, left-and-right) branching,
continuous or discrete, etc., although all these can be characterised
by means of some corresponding extra axioms.
Yes, in this time theory, atomic moments (AND points) are acceptable.
However, they are just ACCEPTABLE, but not necessarily to be
everywhere over the time. The theory only claims that a time element
is either an interval (or specially, a moment) or a point. If one
insists on using moments/points somewhere over the time, they can be
explicitly expressed there. For somewhere else over the time, it may
be the case that each time element is a decomposable interval. That
is, there may be no moments/points at all. It is also consistent to
have a time structure where each time element is either an
decomposable interval or a point, or even a time structure where each
time element is a decomposable interval. In fact, generally speaking,
the basic time structure may be neither dense nor discrete anywhere,
or may be continuous over some parts and discrete over other parts.
This depends on what you want to express and what extra axiom you
would impose.
Pat wrote
| Galton's intuitions are clearly based on thinking of intervals as
sets of points. He takes it as simply obvious, for example, that
there is a distinction between open and closed intervals.
| I still cannot find where Galton claimed this. Actually, before I
answered Pat's last contribution (Newsletter ENRAC 27.3 (98030), I had
re-read Galton's paper and phoned Galton to confirm about this. Yes, a
distinction between open and closed intervals did appear in Galton's
paper, but I found it was only used to demonstrate the corresponding
problem in thinking of intervals as sets of points, and hence as the
reason for him (and for James and Pat?) to take intervals as
primitive. Anyway, this issue doesn't really affect this discussion.
Pat wrote
| Some advantages of this time theory are:
(1) It retains Allen's appealing characteristics of treating
intervals as primitive which overcomes the Dividing Instant
Problem.
|
| See above. But in any case this doesn't overcome the problem.
Allen's treatment allows lights to just come on, but it doesnt
provide anywhere for the ball to be motionless.
| So, that is why we need to extend Allen and Hayes' treatment by
allowing time points as primitive as well. On the one hand,
treating time intervals as primitive avoids the question of whether
intervals are open or closed; on the other hand, allowing time points
(as primitive) provides means for expressing instantaneous phenomena
such as "the ball is motionless at a point".
Pat wrote
| (2) It includes time points into the temporal ontology and therefore
makes it possible to express some instantaneous phenomenon, and
adequate and convenient for reasoning correctly about continuous
change. (3) It is so basic that it can be specified in various ways to
subsume others. For instance, one may simply take the set of points
as empty to get Allen's interval time theory, or specify each
interval, say T, as <T-left, T-right>
where T-left < T-right ,
to get that one Pat prefers.
|
| Not quite right. In my simple theory, T-left isnt before
T-right , it equals it.
| But, the relation T-left < T-right does already
include T-left = T-right , that is T-left equals T-right . In
the case T-left < T-right ,
T is an interval; in the case T-left = T-right ,
T is a pointlike
interval.
Pat wrote
| Yes, exactly, although there is no need to use this formal strategy,
as I explain in the time catalog section 1. Briefly,
Holds-%In P i
is true just when i is a subinterval of a reference interval j
where Holds-%On P j .
Again, it is largely an aesthetic judgement, but I find Galton's
hold-on vs. hold-in distinction awkward and unintuitive. (It
suggests that there are two different 'ways to be true'.)
| Yes. And in Ma et al's paper [j-cj-37-114],
some examples are given to show the
problem with Galton's distinction between hold-on
and hold-in . Also,
it is claimed that the fundamental reason for the problem is that
Galton wanted to characterise the fact that a proposition holds for an
interval in terms of that the proposition holds for every point within
the interval.
Pat wrote
| Well, it depends on what axioms one assumes! Perhaps I have been
speaking too carelessly about the 'usual mathematical way'. Heres my
intuition: the standard account of the continuum seems forced to
resolve the dividing point problem by deciding which interval
contains the point, distinguishing open from closed intervals,
because it identifies an interval with a set of points. (So if
both intervals 'contain' the point, the intervals must intersect.)
One can take points as basic or intervals as basic or both
as primitive; that's irrelevant, but the crucial step is that
(set-of-points = interval) identification. Thats exactly what I want
to avoid. My point is only that if we abandon that idea (which is
only needed for the formal development of analysis within set
theory, a rather arcane matter for us), then there is a way to
formalise time (using both intervals and points as primitive, if you
like) which neatly avoids the problem.
| It is also exactly what the approach of treating both intervals and
points as primitive wants to avoid - it avoids to "identify" an
interval with a set of points. In addition, such an approach allows
expressions of all the three cases shown in my former discussion
(Newsletter ENRAC 13.3 (98027)), without thinking of intervals and
points in the usual mathematical way, or "identify an interval with a
set of points"
Pat wrote
| In the time theory where both intervals and points are taken as
primitive, we can (if we like) talk about the open and closed nature
of intervals with some knowledge being available. This kind
knowledge can be given in terms of the Meets relation, rather than
some "much more complicated extension which includes set theory and
an extensionality axiom for intervals". In fact, we can define that:
|
|
- interval I is left-open at point P iff
Meets(P, I)
interval I is right-open at point P iff Meets(I, P)
interval I is left-closed at point P iff there is an interval I'
such that Meets(I', I) ^ Meets(I', P)
interval I is right-closed at point P iff there is an interval I'
such that Meets(I, I') ^ Meets(P, I') ...
|
| That certainly seems to be an elegant device. (Though the
definitions have nothing to do with knowledge; all Jixin is saying
is that the definitions of open and closed can be given in terms of
MEETS . As Allen and I showed in our old paper, the entire theory
can be reduced to MEETS .) However, in order to be nontrivial, it
must be that points 'separate' meetings, ie if meets(I, P) and
meets(P, J) then ¬ meets(I, J) , right? For if not, all left-open
intervals are also left-closed, etc. This seems to make 'points'
similar to our old 'moments': in fact, if Jixin's theory predicts
meets(P, Q) ·-> P = Q for points P and Q , then I'll lay odds
it is isomorphic to our moments theory. One of the main observations
in our paper was that with the no-meets axiom, one can map moments
to points with no change to the theorems provable.
On the other hand, if the theory allows distinct points to MEET ,
I'd be interested to know how it is able to map smoothly to a
conventional account of the continuum, since that is provably
impossible. One-point closed intervals exist everywhere on the real
line, but no two of them are adjacent. Atomic adjacent times
(whatever we call them) are pretty much a definition of discrete
time models, and are incompatible with density, let alone
continuity.
| Yes. If Meets(I, P) ^ Meets(P, J) then ¬ Meets(I, J) . Actually, in
our theory, If Meets(I, P) ^ Meets(P, J) then Before(I, J) , where
Before(I, J) ·-> ¬ Meets(I, J) , since all the 13 relations are
exclusive to each other. Therefore, it is impossible for an interval
to be both left-open and left-closed.
Also, in our theory, Meets(P, Q) implies that at least one of P and Q
is an interval (or a moment) - they cannot be both points.
In addition, it is important to note that the constraint that a point
cannot meet another point makes it is possible to establish a
consistency checker for temporal database systems (see Knight and
Ma's 1992 paper [j-aicom-5-75]).
Yes. As claimed earlier in this dicussion and actually pointed out in
our published paper, our theory is in fact an extension to that of
Allen and Hayes. Our points are quite like Allen and Hayes' moments -
they cannot meet each other. However, on the one hand, points are
fundamentally different from moments - points have no duration while
moments do have, no matter how small they are. Therefore, it is more
convenient to use points than moments in modelling some instantaneous
phenomenon, especially in the case where duration reasoning is
involved. On the other hand, if moments are simply mapped
to points, how to express the real moments, i.e., non-decomposable
intervals with positives duration (like the "seconds" example given
in my last discussion)?
Jixin
References:
c-ijcai-85-528 | James Allen and Pat Hayes.
A Common-Sense Theory of Time.
Proc. International Joint Conference on Artificial Intelligence, 1985, pp. 528-531.
|
j-aicom-5-75 | Brian Knight and Jixin Ma.
A General Temporal Model Supporting Duration Reasoning.
AI Communications, vol. 5 (1992), pp. 75-84. |
j-ci-5-225 | James F. Allen and Patrick J. Hayes.
Moments and points in an interval-based temporal logic.
Computational Intelligence, vol. 5, pp. 225-238. |
j-cj-37-114 | Jixin Ma and Brian Knight.
A General Temporal Theory.
Computer Journal, vol. 37 (1994), pp. 114-123. |
Salve.
The following are some fragments from the current discussion:
From Pat Hayes - ENRAC 14.3.1998
| instantaneous intervals completely. It is also quite consistent to have
arbitrary amounts of density, discreteness, etc.; for example, one can
say that time is continuous except in a certain class of 'momentary'
intervals whose ends are distinct but have no interior points.
| From Jixin Ma - ENRAC 15.4.1998
| time element is a decomposable interval. In fact, generally speaking,
the basic time structure may be neither dense nor discrete anywhere,
or may be continuous over some parts and discrete over other parts.
| Pat and Jixin, what do you mean when you write ``continuous''?
Here in Pisa, we write ``continuity'' and we read ``axiom of
completeness'', which is what everyone commonly means when speaking
about (the founding notion of) continuity. I really find it difficult to
believe that you like to make an exception in this sense, also because the
hat here is ``formal (temporal) reasoning''. It also seems to me that any
temporal structure must necessarily fail to be persuasive if on one hand
it includes the notion of continuity and on the other it refuses it;
how can time be continuous ... with some exception? Either it is
continuous, or it is not! That is, either the Basic Time Structure
assumes the axiom of completeness, or it does not!!
In fact, in this discussion I have not yet seen any explanation why
an alternative notion of continuous structure is needed at all? I am not
asking you to argue about your own notion, I just ask you to give a
convincing argument on the need of a notion which is an alternative to
the classical one, such as: ``the problem P of temporal reasoning about
actions and change can not be solved adopting the axiom of completeness'',
or ``the axiom of completeness is too strong an assumption for our purposes;
axiom A is better suited, because...''.
Sergio
To Pat
As an addition to my response (ENRAC 15.4 (98035)) to Pat's
suggestion of "simply map Allen and Hayes' moments to Ma and
Knight's points":
The constraint that "moments cannot meet each" will lead to the
conclusion that we can have neither a completely discrete nor a
completely dense system which contains both moments and
decomposable intervals. However, if we revise Allen and Hayes'
system to include both points and intervals (including moments), and
impose the "not-meet-each-other" constraint on points only, rather
than on moments, this objection does not apply.
To Sergio
| Here in Pisa, we write ``continuity'' and we read ``axiom of
completeness'', which is what everyone commonly means when speaking
about (the founding notion of) continuity. I really find it difficult to
believe that you like to make an exception in this sense, also because the
hat here is ``formal (temporal) reasoning''. It also seems to me that any
temporal structure must necessarily fail to be persuasive if on one hand
it includes the notion of continuity and on the other it refuses it;
how can time be continuous ... with some exception? Either it is
continuous, or it is not! That is, either the Basic Time Structure
assumes the axiom of completeness, or it does not!!
In fact, in this discussion I have not yet seen any explanation why
an alternative notion of continuous structure is needed at all? I am not
asking you to argue about your own notion, I just ask you to give a
convincing argument on the need of a notion which is an alternative to
the classical one, such as: ``the problem P of temporal reasoning about
actions and change can not be solved adopting the axiom of completeness'',
or ``the axiom of completeness is too strong an assumption for our purposes;
axiom A is better suited, because...''.
| First of all, what do you mean "the classical one"? (the classical
continuous time structure)? Does it refer to the classical physical
model of time, where the structure is a set of points which is
isomorphic to the real line?
| in Pisa, we write ``continuity'' and we read ``axiom of completeness'',
which is what everyone commonly means when speaking'
about (the founding notion of) continuity.
| At the ontological level, the notion of continuous time vi
discrete time is closely related to questions "Is the set of
time elements dense or not?", and "Are there really time atoms?".
For a point-based model, the continuity is usually characterised as
"Between any two points, there is a third"; while for an
interval-based model (like that of Allen), it is characterised as
"Every interval can be decomposed into two adjacent sub-intervals".
In addition, as for a model which takes both intervals and
points as primitive, one may characterise two levels of density. At
the weak level, it is only required that each interval can be divided
into two adjacent sub-intervals. At the strong level, it is required
that there is always a point within any interval. It is easy to
infer that the latter can imply the former.
As for general treatments, the Basic Time Structure does not
have to impose the axiom of density or discreteness (Similar
arguments apply to issues such as linear/non-linear,
bounded/un-bounded). Therefore, the time structure as a whole may be
continuous or discrete, or neither continuous nor discrete.
Now, "why an alternative notion of continuous structure is needed at
all"? It has been noted that, temporal knowledge in the domain of
artifical intelligence, including "temporal reasoning about actions
and change", is usually imcomplete, and using time intervals in
many cases is more convenient and more in-keeping with common
sense of temporal concepts than to use the classical abstraction of
points. In fact, the notion of time intervals (or periods) has been
introduced for a long time in the literature. In addition, in order
to overcome/bypass the annoying question of whether various
intervals are open or
closed, various approached have been proposed. An example is Allen's
interval-based time theory. As for these time theories, the old
(classical?) notion of continuity no longer simply applies.
Jixin
Sergio Brandano wrote:
| The following are some fragments from the current discussion:
From Pat Hayes - ENRAC 14.3.1998
|
| instantaneous intervals completely. It is also quite consistent to have
arbitrary amounts of density, discreteness, etc.; for example, one can
say that time is continuous except in a certain class of 'momentary'
intervals whose ends are distinct but have no interior points.
|
|
From Jixin Ma - ENRAC 15.4.1998
|
| time element is a decomposable interval. In fact, generally speaking,
the basic time structure may be neither dense nor discrete anywhere,
or may be continuous over some parts and discrete over other parts.
|
|
Pat and Jixin, what do you mean when you write ``continuous''?
Here in Pisa, we write ``continuity'' and we read ``axiom of
completeness'', which is what everyone commonly means when speaking
about (the founding notion of) continuity. I really find it difficult to
believe that you like to make an exception in this sense, also because the
hat here is ``formal (temporal) reasoning''. It also seems to me that any
temporal structure must necessarily fail to be persuasive if on one hand
it includes the notion of continuity and on the other it refuses it;
how can time be continuous ... with some exception? Either it is
continuous, or it is not! That is, either the Basic Time Structure
assumes the axiom of completeness, or it does not!!
| Why cannot time be continuous in some places but discontinuous at others?
There is no mathematical objection to such a structure, and it has been
argued that a continuum punctuated by a sparse collection of points of
discontinuity might be a plausible mathematical picture of time which seems
to 'flow smoothly' except when things happen suddenly. (Similar arguments
can be made for describing spatial boundaries, by the way; and elementary
physics makes similar assumptions, where velocity is supposed to change
smoothly except when 'impact' occurs.)
You talk about a 'founding notion' of continuity as being that captured by
the axiom of completeness. Here, in my view, you commit a philosophical
error (especially in Pisa!) There are intuitions about continuity which
one can try to capture in various formal ways, but there is no 'founding
notion' of continuity other than those intuitions. In the late 19th
century, famous mathematicians objected strongly to the view of the
continuum as consisting of a set of points, for example. This modern
perspective, now taught in high schools, is a modern invention, not a
'founding' notion. It is more recent than the gasoline engine, yet people
have had intuitions about smoothness, instantaneity and continuity for
eons. (Whether or not one agrees with me on this admittedly controversial
point, it seems unwise to identify a mathematical property such as
continuity with any kind of axiom until one has verified that no other
axiom will do as well; and as I am sure Sergio knows, there are many
alternative ways to axiomatize continuity.)
In my view, axioms are tools which we can manipulate at will; they are not
set in stone or somehow inevitable. Different formal accounts of time might
be appropriate for different purposes or to capture different intuitions.
(I agree with Jixin that it is useful to seek a common 'core' theory which
can be extended in various ways to describe various possible more complex
temporal structures; and that this theory will have to be rather weak.)
| In fact, in this discussion I have not yet seen any explanation why
an alternative notion of continuous structure is needed at all? I am not
asking you to argue about your own notion, I just ask you to give a
convincing argument on the need of a notion which is an alternative to
the classical one, such as: ``the problem P of temporal reasoning about
actions and change can not be solved adopting the axiom of completeness'',
or ``the axiom of completeness is too strong an assumption for our purposes;
axiom A is better suited, because...''.
| The 'dividing point' problem which gave rise to this discussion would do.
According to the modern account of the continuum, this point must exist,
and since all intervals consist of points, the light is therefore either on
or off at it. But it seems more natural, as well as formally simpler, to
just say that the question is meaningless; perhaps (though this is no
longer my own preference) because that point doesnt exist.
A more mundane example is given by temporal databases, which usually assume
in their basic ontology that time is discrete: for example, they routinely
describe times as integers representing the number of milliseconds since
the birth of Christ. (Of course, one can always insist that these are to be
understood as being embedded in a continuum, but then what use is an axiom
whose sole purpose is to insist that times exist which have no name and
about which nothing can be asserted, other than that they exist?)
Pat Hayes
Pat Hayes wrote
| In my view, axioms are tools which we can manipulate at will; they are not
set in stone or somehow inevitable. Different formal accounts of time might
be appropriate for different purposes or to capture different intuitions.
(I agree with Jixin that it is useful to seek a common 'core' theory which
can be extended in various ways to describe various possible more complex
temporal structures; and that this theory will have to be rather weak.)
| and
| A more mundane example is given by temporal databases, which usually assume
in their basic ontology that time is discrete: for example, they routinely
describe times as integers representing the number of milliseconds since
the birth of Christ. (Of course, one can always insist that these are to be
understood as being embedded in a continuum, but then what use is an axiom
whose sole purpose is to insist that times exist which have no name and
about which nothing can be asserted, other than that they exist?)
| If axioms are guaranteed to be used only in a particular program or
set of programs, they need be no stronger than necessary.
As to the rhetorical "what use", suppose the theory is to tolerate the
elaboration that two successive events, shooting Pat and his falling
to the ground, occurred between successive ticks of the clock. If you
guarantee that no such elaborations will be required or that you are
willing to do major surgery on your theory should elaboration be
required, then you are ok with a weak theory even if it is
unextendable.
In reply to Pat and Jixin.
I apologize for the length of this message, although it mainly
consists of quoted text. As ``skin perception'', it seems to me
my critics hits the target. The arguments of reply I received, in
fact, are not as convincing as they were supposed to be.
The details follow.
To Jixin
| First of all, what do you mean "the classical one"? (the classical
continuous time structure)? Does it refer to the classical physical
model of time, where the structure is a set of points which is
isomorphic to the real line?
| I can just quote myself ...
| Here in Pisa, we write ``continuity'' and we read ``axiom of
completeness'', which is what everyone commonly means when speaking
about (the founding notion of) continuity.
At the ontological level, the notion of continuous time vi
discrete time is closely related to questions "Is the set of
time elements dense or not?", and " Are there really time atoms?".
| The word "continuity", even at the ontological level, can not be read
as "continuous with some exception".
| For a point-based model, the continuity is usually characterized as
"Between any two points, there is a third"; while for an
interval-based model (like that of Allen), it is characterized as
"Every interval can be decomposed into two adjacent sub-intervals".
| The axiom of completeness states:
Let be A and B non empty subsets of S such that a < b
for all a in A and b in B . Then there
exists xi in S such that
a < xi < b for all a in A and b in B .
Now, the set S , that is your domain, may consist as well either of
time-points or time-intervals; S holds real numbers on the former
case, intervals from the real line on the latter case.
| As for general treatments, the Basic Time Structure does not
have to impose the axiom of density or discreteness (Similar
arquements apply to issues such as linear/non-linear,
bounded/un-bounded). Therefore, the time structure as a whole may be
continuous or discrete, or neither continuous nor discrete.
| I agree with your premise: the Basic Time Structure does not have to
impose the choice, in fact it leaves you free in that sense. As soon as
you make the choice, then you obtain either a continuous structure or
a discrete structure, just depending on this choice. I do not agree,
instead, with your conclusion. If I leave you the freedom to choose,
it does not mean the Structure is neither continuous nor discrete; it
simply means you still have to make the choice.
| Now, "why an alternative notion of continuous structure is needed at
all"? It has been noted that, temporal knowledge in the domain of
aritifical intelligence, including "temporal reasoning about actions
and change", is usually imcomplete, and using time intervals in
many cases is more convenient and more in-keeping with common
sense of temporal concepts than to use the classical abstraction of
points. In fact, the notion of time intervals (or periods) has been
introduced for a long time in the literature. In addition, in order
to overcome/bypass the annoying question of if intervals are open or
closed, various approached have been proposed. An example is Allen's
interval-based time theory. As for these time theories, the old
(classical?) notion of continuity no longer simply applies.
| My question referred to what is needed rather than convenient.
I understand it may be convenient, in some cases, to use intervals, but
this is not pertinent with my criticism, which still holds.
Let me ask you a more stringent question.
Premise: It is evident that if you assume the axiom of completeness,
the domain S can just be continuous, while if you do not assume the
axiom of completeness then S is necessarily discrete.
Question: Suppose that you define your neither continuous nor discrete
Temporal Structure. What is your domain S ? What is your
replacement for the axiom of completeness? Does this structure
(provably) solve for at least one problem what can not be (provably)
solved via the axiom of completeness? Can you give an example?
To Pat
| Why cannot time be continuous in some places but discontinuous at others?
| Places? If we shall understand time like physicians understands the
space, then things become considerably simpler: time can just be discrete,
since space itself is fully discrete (the observable one). The use of
real lines, in physics, is just a theoretical convenience. If by
``place'' we shall mean, instead, some point in a lattice (and we
shall provide a convenient reference system for the branching-time
case), then the case may hold. In part my question was, in fact, to
insert in the actual discussion explicit and convincing arguments
about the case (examples, counterexamples and axioms, are welcome).
| There is no mathematical objection to such a structure, and it has been
| If a Temporal Structure exists in this sense, may I have a look at
its domain (that is at the S domain, as stated above)? Can you post
the reference together with the explicit case? But the primary
question still remains whether is it needed at all.
| You talk about a 'founding notion' of continuity as being that captured by
the axiom of completeness. Here, in my view, you commit a philosophical
error (especially in Pisa!) There are intuitions about continuity which
one can try to capture in various formal ways, but there is no 'founding
notion' of continuity other than those intuitions. ...
| What properly formalizes the notion of continuity is the axiom of
completeness. Alternative notions are equivalent, until we speak
about continuous domains. The point was whether one can have a
continuous domain (that is the S I stated above) ... with exceptions.
Concerning the intuition, let me remind that the student who
discovered the square root of 2 was killed (down the cliff), and
no one was allowed to speak about ... ``the fault of the god'' for
long time. Humans' common sense, to me, is something we shall not
call too much.
| century, famous mathematicians objected strongly to the view of the
continuum as consisting of a set of points, for example. This modern
perspective, now taught in high schools, is a modern invention, not a
'founding' notion. It is more recent than the gasoline engine, yet people
have had intuitions about smoothness, instantaneity and continuity for
eons. (Whether or not one agrees with me on this admittedly controversial
point, it seems unwise to identify a mathematical property such as
continuity with any kind of axiom until one has verified that no other
axiom will do as well; and as I am sure Sergio knows, there are many
alternative ways to axiomatize continuity.)
| ... unwise ?
If another axiom exists, which does as well, then it is surely equivalent
to the axiom of completeness, just because it does as well. Alternative
notions are clearly equivalent, until we speak about continuous domains.
The point here, instead, was whether one can have a continuous domain
with exceptions, that is the claim I originally criticized.
| In my view, axioms are tools which we can manipulate at will; they are not
set in stone or somehow inevitable. Different formal accounts of time might
| The point is what would you like to make out of it. If you do
something I can do with a simpler approach and without arising
criticism, then my approach will have much more impact than yours,
I think you may agree at least on this point.
| be appropriate for different purposes or to capture different intuitions.
(I agree with Jixin that it is useful to seek a common 'core' theory which
can be extended in various ways to describe various possible more complex
temporal structures; and that this theory will have to be rather weak.)
| Concerning the core theory that you and Jixin are willing to obtain,
I already developed a Basic Time Structure which may be of interest.
It is as simple as I managed to design it, without un-useful
complications. The structure works well in my case. you are welcome
to read and comment my contribution, which may be found in my ETAI's
reference.
Finally, concerning your examples:
| The 'dividing point' problem which gave rise to this discussion would do.
According to the modern account of the continuum, this point must exist,
and since all intervals consist of points, the light is therefore either on
or off at it. But it seems more natural, as well as formally simpler, to
just say that the question is meaningless; perhaps (though this is no
longer my own preference) because that point doesn't exist.
| Why is it a convincing argument?
| A more mundane example is given by temporal databases, which usually assume
in their basic ontology that time is discrete: for example, they routinely
describe times as integers representing the number of milliseconds since
the birth of Christ. (Of course, one can always insist that these are to be
understood as being embedded in a continuum, but then what use is an axiom
whose sole purpose is to insist that times exist which have no name and
about which nothing can be asserted, other than that they exist?)
| Why is it a convincing argument ?
Best Regards
Sergio
Pat,
In answer to Sergio, you wrote
| Why cannot time be continuous in some places but discontinuous in
others?
| (Jixin answered along similar lines). I have no problems accepting that
a function of time may be piecewise continuous, or that it may be
undefined for some points along the time axis. However, it seems to
me that there are several problems with saying that time itself
is piecewise continuous (btw - do you mean piecewise dense?).
The first problem is with respect to motivation. For what reasons would
Time suddenly skip over potential timepoints? If the reason is, as you
wrote, that
| The 'dividing point' problem which gave rise to this discussion would do.
According to the modern account of the continuum, this point must exist,
and since all intervals consist of points, the light is therefore either on
or off at it. But it seems more natural, as well as formally simpler, to
just say that the question is meaningless; perhaps (though this is no
longer my own preference) because that point doesn't exist.
| then exactly what events in the world would be allowed to contribute to
the continuity faults? Does the next time I hit a key on my keyboard
qualify? And what about the midpoint halfway between two continuity faults,
is it also a continuity fault, recursively?
The other problem is with respect to the axiomatizations. Since your
article "A catalog of temporal theories" characterizes the various
theories through axiomatizations, I thought I'd go back to that article
and check how you had done this formally. However I was not able to
find it; the closest I got was the denseness axiom on page 15. If the
intuitive notion is that time itself is continuous in some places but
not in others, wouldn't it be natural to start with an axiomatization
of continuous time (such as the real numbers) and then to proceed from
there? For example, a domain of piecewise continuous time could be
represented as a twotuple <R,D> where R is the real numbers and
D is a "small" subset of it; the intention being that R-D is the
modified time domain in question. The notions of non-standard intervals
could then be constructed as the natural next step.
Maybe I'm missing something - are constructs of this kind subsumed by
the axioms in your report, or can they be inferred as theorems? Or why
is this not the natural way of doing things?
Erik
To Sergio,
| First of all, what do you mean "the classical one"? (the classical
continuous time structure)? Does it refer to the classical physical
model of time, where the structure is a set of points which is
isomorphic to the real line?
|
|
I can just quote myself ...
|
| Here in Pisa, we write ``continuity'' and we read ``axiom of
completeness'', which is what everyone commonly means when speaking
about (the founding notion of) continuity.
|
|
Concerning the core theory that you and Jixin are willing to obtain,
I already developed a Basic Time Structure which may be of interest.
It is as simple as I managed to design it, without un-useful
complications. The structure works well in my case. you are welcome
to read and comment my contribution, which may be found in my ETAI's
reference.
| So, you didn't refer "the classical one" to "the Basic Time
Structure" you developed, did you? If No, why did you develop it?
What is your convincing argument(s) on the need of such a structure?
Is it also an alternative to the classical one? (Sorry, I am here
using the similar question raised by youself to ask you, though I
don't have to, see below). If Yes, I shouldn't ask this question.
| At the ontological level, the notion of continuous time vi
discrete time is closely related to questions "Is the set of
time elements dense or not?", and " Are there really time atoms?".
|
|
The word "continuity", even at the ontological level, can not be read
as "continuous with some exception".
| What I actually said is very clear as you quoted above. Does it imply
that "the word continuity can be read as continuous with some exception"?
In fact, even when Pat talked about "continuous with some exception", he
didn't really mean that it is as same as the word "continuity". What he
means, as I understand, is just that, with the exception of time moments,
each time interval can be decomposed into (at least two) sub-intervals.
| The axiom of completeness states:
Let be A and B non empty subsets of S such that a < b
for all a in A and b in B . Then exists xi in S such that
a < xi < b for all a in A and b in B .
Now, the set S , that is your domain, may consists as well either of
time-points or time-intervals; S holds real numbers on the former
case, intervals from the real line on the latter case.
| Firstly, you said here, "the former case" and "the latter case". Can
these two cases be mixed together? In other words, can the domain
contain both time-points and time-intervals. I suppose it should.
Otherwise, you will meet some problem in satisfying the so-called
completeness axiom (see below).
Secondly, you take time-points as real numbers, and intervals "from"
the real line. Are your intervals sets of real numbers limited by
their end-points (real numbers)? If no, what are they? If yes, have
you considered the dividing instant problem? This problem would be more
obvious with your time structure when you try to impose the axiom of
completeness (see below).
Thirdly, if the domain S consists of time-intervals, you need to
re-define (or revise, or, at least, explain) the
relation < between elements of the domain S. After you have done
this properly (You didn't show how to do it, you just
claimed that the domain "may" contain either time-points or
time-intervals), you have to show, for the case that interval a in A
is immediately before interval b in B (that is, there is no other time
elements between a and b ) what is the required xi
such that a < xi < b .
Obviously, xi cannot be an interval (non-pointlike),
otherwise, it will overlap with a and b . Therefore, if you can define
what it is, it has to be a point (This is why I said earlier in the
above that if your domain contains intervals, it needs to contain
points as well). Now, you meet the dividing instant problem, as I
expected.
By the way, may I take this as one of the "un-useful" complications
with your time structure?
| As for general treatments, the Basic Time Structure does not
have to impose the axiom of density or discreteness (Similar
arquements apply to issues such as linear/non-linear,
bounded/un-bounded). Therefore, the time structure as a whole may be
continuous or discrete, or neither continuous nor discrete.
|
|
I agree with your premise: the Basic Time Structure does not have to
impose the choice, in fact it leaves you free in that sense. As soon as
you make the choice, then you obtain either a continuous structure or
a discrete structure, just depending on this choice. I do not agree,
instead, with your conclusion. If I leave you the freedom to choose,
it does not mean the Structure is neither continuous nor discrete; it
simply means you still have to make the choice.
| If you don't impose the continuous axiom (!!! as argued by Pat, it
does not have to be the so-called axiom of completeness !!!) or
discrete axiom, the structure can be neither continuous nor discrete.
I think it is very easy to form a structure which satifies the basic
axiomatisation, but does not satisfy the continuous requirement, and
does not satisfy the discrete requirement. In fact, you can write
down any extra constraint as long as it is consistent with the basic
theory.
| Now, "why an alternative notion of continuous structure is needed at
all"? It has been noted that, temporal knowledge in the domain of
artifical intelligence, including "temporal reasoning about actions
and change", is usually imcomplete, and using time intervals in
many cases is more convenient and more in-keeping with common
sense of temporal concepts than to use the classical abstraction of
points. In fact, the notion of time intervals (or periods) has been
introduced for a long time in the literature. In addition, in order
to overcome/bypass the annoying question of if intervals are open or
closed, various approached have been proposed. An example is Allen's
interval-based time theory. As for these time theories, the old
(classical?) notion of continuity no longer simply applies.
|
|
My question referred to what is needed rather than convenient.
I understand it may be convenient, in some cases, to use intervals, but
this is not pertinent with my criticism, which still holds.
| So, you think intervals are not needed? Anyway, our arguements
about the convenience of using intervals are based on the belief of
the need of them.
| Let me ask you a more stringent question.
Premise: It is evident that if you assume the axiom of completeness,
the domain S can just be continuous, while if you do not assume the
axiom of completeness then S is necessarily discrete.
| Wrong! Even if you do not asssume the axiom of completeness, it is
still not necessarily discrete.
| Question: Suppose that you define your neither continuous nor discrete
Temporal Structure. What is your domain S ? What is your
replacement for the axiom of completeness? Does this structure
(provably) solve for at least one problem what can not be (provably)
solved via the axiom of completeness? Can you give an example?
| The domain is just a collection of time elements each of which
is either an interval (in a particular case, a moment) or a point.
The basic core theory doesn't commit itself to whether
the time stucture is
continuous or discrete. So, if you would like one which is neither
continuous nor discrete, you don't need the axiom of completeness.
Why do I need a replacement for it, anyway, if it is not supposed?
Extra axioms regarding dense/discrete, linear/non-linear,
bounded/non-bounded time structure, etc. can be given (e.g., see Ma
and Knight's 1994 paper [j-cj-37-114]). Specially, the characterisation
of continuity does not have to be in the form of axiom of completeness.
In addition, as shown above, in the case where time intervals are
addressed, it becomes very complicated (if not impossible) to
simply apply such an axiom.
As for example you would like to see, the DIP is a typical one, as I
have shown in the above.
Also, in your reply to Pat you wrote:
| What properly formalizes the notion of continuity is the axiom of
completeness. Alternative notions are equivalent, until we speak
about continuous domains. The point was whether one can have a
continuous domain (that is the S I stated above) ... with
exceptions.
If another axiom exists, which does as well, then it is surely
equivalent to the axiom of completeness, just because it does as
well. Alternative notions are clearly equivalent, until we speak
about continuous domains. The point here, instead, was whether one
can have a continuous domain with exceptions, that is the claim I
originally criticized.
| First, as shown in the above, the axiom of completeness doesn't
simply apply to the case when time intervals are involved.
Therefore, your claim that "alternative notions of continuity are
clearly equivalent" is unjustified, at least it is not clear!
Second, the question of "whether one can have a continuous
domain with exceptions" depends on how do you understand the real
meaning. It is important to note that neither Pat nor myself claims
that one can have such a structure as you understood and hence
described by "a continuous domain with exceptions". Of course, if you
have already assumed that the domain as a whole is continuous, then it
must be continuous - no exception! This is just like if you impose
that "The traffic light was green throughout last week", then, of
course, it was green any time during last week, no exception.
Similarly, if you impose that "The traffic light was red throughout
last week", then it was red any time during last week. Again, no
exception. However, if you don't have either of them, why can't one
have the case that over the last week, the traffic light was
sometimes red, and sometimes green, and even sometimes yellow?
As I said earlier, when Pat talked about "continuous with
exceptions", he actually meant that "except at those time moments,
the time is continuous", or more specially, "except for time moments,
each time interval is decomposable". I don't think he would actually
assume, in the first place, the continuity of the whole domain, then
expect there are some exceptions. Do I understand your meaning
rightly, Pat?
Jixin
References:
j-cj-37-114 | Jixin Ma and Brian Knight.
A General Temporal Theory.
Computer Journal, vol. 37 (1994), pp. 114-123. |
Pat Hayes wrote (ENRAC 21.4.1998):
| Why cannot time be continuous in some places but discontinuous at others?
There is no mathematical objection to such a structure, and it has been
argued that a continuum punctuated by a sparse collection of points of
discontinuity might be a plausible mathematical picture of time which seems
to 'flow smoothly' except when things happen suddenly. (Similar arguments
can be made for describing spatial boundaries, by the way; and elementary
physics makes similar assumptions, where velocity is supposed to change
smoothly except when 'impact' occurs.)
| If a given temporal structure includes the solution to the problem of
representing ``perceived smooth'' flux and ``perceived fast'' flux of
time, then that temporal structure is necessarily agent-centric,
since different agents may have a different perception of the world.
In being agent-centric, this structure can not aim at generality. In
fact, if we design an agent-centric temporal structure and the world
is inhabited by more than one agent, then we must design a more
general structure that reconciles the different views from the
different agents. I say ``must'' because, otherwise, we pre-destine
agents to never interact with each other, which would be a major
restriction.
Sergio
John McCarthy wrote
| If axioms are guaranteed to be used only in a particular program or
set of programs, they need be no stronger than necessary.
As to the rhetorical "what use", suppose the theory is to tolerate the
elaboration that two successive events, shooting Pat and his falling
to the ground, occurred between successive ticks of the clock. If you
guarantee that no such elaborations will be required or that you are
willing to do major surgery on your theory should elaboration be
required, then you are ok with a weak theory even if it is
unextendable.
| in answer to my remarks:
| In my view, axioms are tools which we can manipulate at will; they are not
set in stone or somehow inevitable. Different formal accounts of time might
be appropriate for different purposes or to capture different intuitions.
...
| and
| A more mundane example is given by temporal databases, which usually assume
in their basic ontology that time is discrete:...
| If I understand what John is saying, then I completely agree with him. (If
he intended to disagree with me, then maybe I dont understand his point.)
However, note that both Jixin and I are trying to give a theory which is as
elaboration-tolerant as possible, without being completely vacuous.
Answers to Sergio Brandano
Sergio seems to be on a different planet, as his responses to both Jixin
and I seem to quite miss the point of our debate, and often to be
completely free of content.
| I can just quote myself ...
| Well, you can; but to do so is at best unhelpful, and at worst arrogant.
If someone fails to understand you and asks for clarification, to simply
repeat yourself is obviously unlikely to give them the clarification they
need.
| Here in Pisa, we write ``continuity'' and we read ``axiom of
completeness'', which is what everyone commonly means when speaking
about (the founding notion of) continuity.
| May I ask in return if "everyone" here is meant to refer to everyone in
Pisa, or to a broader community? If the former, my advice is to travel
more; if the latter, then you are simply wrong.
| The word "continuity", even at the ontological level, can not be read
as "continuous with some exception".
| There is an entire mathematical theory of punctuated continua, ie spaces
which are continuous everywhere except for a non-dense set of points. Such
structures even arise naturally from purely continuous phenomana in, for
example, catastrophe theory.
The formal trick, you see, is to alter the axiom so that instead of reading
'for all points...' it reads 'there exists a set S such that for all points
not in S ...'. The result is also an axiom, believe it or not.
| For a point-based model, the continuity is usually characterized as
"Between any two points, there is a third"; while for an
interval-based model (like that of Allen), it is characterized as
"Every interval can be decomposed into two adjacent sub-intervals".
|
|
The axiom of completeness states:
Let be A and B non empty subsets of S such that a < b
for all a in A and b in B . Then exists xi in S such that
a < xi < b for all a in A and b in B .
Now, the set S , that is your domain, may consists as well either of
time-points or time-intervals; S holds real numbers on the former
case, intervals from the real line on the latter case.
| You havent said what < means for intervals. If it means that the
endpoint of a is point- < the first point of b , then this axiom seems
false; for consider a point p and the set A = { <p1, p> } for any
p1 < p and B = { <p, p2> } for any p2 > p . This satisfies your
premise, but there is no
interval between any of the intervals in A and any interval in B (unless
you allow intervals consisting of a single point.) But in any case, you are
here assuming that the real line is your intended model. But this axiom
doesnt characterize the real line. Its true on the rationals, for one
thing, but thats not all. For example, here is a nonstandard model of your
axiom: interpret points as pairs <n,q> where n is an integer and q is a
rational number or the symbol " i ", and say that <a,b> < <c,d> just when
a < c v (a = c ^ d = "i") v (a = c ^ b < d) .
This amounts to N copies of
Q laid end-to-end with points at infinity placed between them. It satisfies
your axiom. Im sure that anyone with a little imagination can easily cook
up lots more such nonstandard worlds.
| Premise: It is evident that if you assume the axiom of completeness,
the domain S can just be continuous, while if you do not assume the
axiom of completeness then S is necessarily discrete.
| If you do not assume the axiom, then S may be discrete, continuous or any
mixture. Did you mean to say, if you assume the negation of the axiom?
But the negation of your axiom simply says that some point of discontinuity
exists; it does not impose a discrete structure on the whole of S . It is
much more difficult to axiomatise a discrete structure than a dense one; in
fact, it cannot be done in first-order logic.
| Question: Suppose that you define your neither continuous nor discrete
Temporal Structure. What is your domain S ? What is your
replacement for the axiom of completeness?
| See above, but modify the domain to exclude the "i" symbols. This structure
( N copies of Q ) is dense almost everywhere, but your axiom fails to hold
when the sets A and B are infinite in a particular way. There are 2|Q
subsets in the power set of this domain, and only N|2 of them fail your
axiom, so by almost any standard it is true 'almost' everywhere. (Another
interesting example is got by reversing N and Q , so that one has Q copies
of N laid end-to-end. This fails your axiom 'locally', ie when the subests
are only finitely separated, but satisfies it for sufficiently separated
sets. It is like a discrete space which changes to a dense one when the
scale is reduced sufficiently. See J.F.A.K.Van Benthem, The Logic of
Time, [mb-Benthem-83] for a lovely discussion of such examples.
| Why cannot time be continuous in some places but discontinuous at others?
|
|
Places? If we shall understand time like (physicists) understands the
space, ...
| Yes, that is more or less what I have in mind. Do you propose to formalise
a theory of time which is incompatible with physics? (Why??)
| There is no mathematical objection to such a structure, and it has been
|
|
If a Temporal Structure exists in this sense, may I have a look at
its domain (that is at the S domain, as stated above)?
| See above examples and use your imagination.
But this is a trivial challenge. It can be done for any set S with a
(strict) ordering < . Select a subset P of S , and define a new order
relation « on S+P as follows: x « y iff
(x in S ^ y in P ^ x = y) xor x < y .
( + here is disjoint union, xor is exclusive-or). This inserts a
'twin' of each point in P just after it, with no points between them. If S
is dense/continuous/whatever, then this new structure is that too
everywhere except at points in P . If P is a dense subset of S , this
construction effectively makes two copies of the original set in that
region with a 'sawtooth' ordering that jumps back and forth between them,
inserting a discrete section into the originally dense ordering:
P . . . . . . . .
S ...........|\|\|\|\|\|\.........|\......
looks like this when 'straightened out':
............. .. .. .. .. .. .. .. .. .............. .............
(BTW, another way to describe this is that each point in the P -subset of S
is replaced by a 'two-sided' point.) If S is dense, then your axiom applies
everywhere except at points in P .
(Aside to Jixin: this is the intuition behind the idea of replacing moments
by points. The endpoints of a moment can be thought of as the result of
this construction on a smaller set of points, and the construction can be
reversed by identifying the endpoints of the moment, ie treating the moment
as being pointlike. The result is a timeline with some points identified as
being 'interval-like', ie capable of having something true at them. If
moments never meet, then all the axioms of the Allen-Hayes theory apply to
the S-line iff they applied to the original. This is why your theory and
ours are essentially the same. )
Erik Sandewall wrote
| ..... I have no problems accepting that
a function of time may be piecewise continuous, or that it may be
undefined for some points along the time axis. However, it seems to
me that there are several problems with saying that time itself
is piecewise continuous (btw - do you mean piecewise dense?).
| (Yes, most of this discussion is really about density rather than
continuity. Ive just let this ride for now.)
| The first problem is with respect to motivation. For what reasons would
Time suddenly skip over potential timepoints? If the reason is, as you
wrote, that
|
| The 'dividing point' problem which gave rise to this discussion would do.
According to the modern account of the continuum, this point must exist,
and since all intervals consist of points, the light is therefore either on
or off at it. But it seems more natural, as well as formally simpler, to
just say that the question is meaningless; perhaps (though this is no
longer my own preference) because that point doesn't exist.
|
|
then exactly what events in the world would be allowed to contribute to
the continuity faults? Does the next time I hit a key on my keyboard
qualify?
| It all depends on whether you find it useful to describe things that way.
The idea of intervals simply meeting seems to be a very useful way to
think about time, and it immediately gives rise to all these problems.
| And what about the midpoint halfway between two continuity faults,
is it also a continuity fault, recursively?
| Not necessarily. (Im not sure quite what your point is here. Must there be
a waterfall exactly between two waterfalls?)
| The other problem is with respect to the axiomatizations. Since your
article "A catalog of temporal theories" characterizes the various
theories through axiomatizations, I thought I'd go back to that article
and check how you had done this formally. However I was not able to
find it; the closest I got was the denseness axiom on page 15.
| Throughout the catalog, I give density and discreteness axioms. As I say in
the text, you can take your pick; or, if you like, you can say that time is
dense sometimes and discrete others, making obvious slight changes to the
axioms to make these assertions. The axioms in the theories of the catalog
are offered to you like pieces of an erector set. I make no committment to
their truth, only that they fit together properly.
As to whether time really is discrete or continuous, etc., the only
people who can answer questions like that are physicists, not we who merely
craft ontologies.
| If the
intuitive notion is that time itself is continuous in some places but
not in others, wouldn't it be natural to start with an axiomatization
of continuous time (such as the real numbers) and then to proceed from
there? For example, a domain of piecewise continuous time could be
represented as a twotuple <R,D> where R is the real numbers and
D is a "small" subset of it; the intention being that R-D is the
modified time domain in question. The notions of non-standard intervals
could then be constructed as the natural next step.
| Yes, that is a possible approach. However, (1) the real numbers are already
a very compicated domain to axiomatise, requiring such things as set theory
and notions of limits, etc..; I was looking for something much more
mundane; and (2) as Ive said repeatedly, the real line isnt a very good
model of our temporal intuitions, in my view, but comes along with a lot of
misleading assumptions which are not necessary for temporal reasoning.
| Maybe I'm missing something - are constructs of this kind subsumed by
the axioms in your report, or can they be inferred as theorems?
| Neither. The time axioms are far too weak to be able to infer anything
about real analysis. However, it should be possible to construct models of
the time axioms using ordinary mathematical notions like the integers and
the reals, and indeed I try to do that for every theory in the catalog. At
the very least, this helps guide ones intuitions about just what it is ones
axioms really say, instead of what one hopes they ought to say.
Pat
References:
In reply to Jixin Ma (ENRAC 23.4.1998)
| So, you didn't refer "the classical one" to "the Basic Time
Structure" you developed, did you? If No, why did you develop it?
What is your convincing argument(s) on the need of such a structure?
Is it also an alternative to the classical one? (Sorry, I am here
using the similar question raised by youself to ask you, though I
don't have to). If Yes, I shouldn't ask this question.
| By "the classical one" I mean the classical notion of continuity.
By "the basic time structure" I mean a basic (minimal) time structure.
By "the time structure X" I mean the temporal structure we like to deal
with. It is obtained from the basic time structure via additional axioms.
You also invited me to be more explicit with respect to the following
sentence.
| The axiom of completeness states:
Let A and B be non empty subsets of S such that a < b
for all a in A and b in B . Then there
exists xi in S such that
a < xi < b for all a in A and b in B .
Now, the set S , that is your domain, may consists as well either of
time-points or time-intervals; S holds real numbers on the former
case, intervals from the real line on the latter case.
| The (temporal) domain S , as I meant, may consist either of time-points
xor of time-intervals (exclusive "or").
An interval from the real-line is an ordered set of real numbers limited
by its end-points, which are not necessarily included in the set.
Suppose S consists of intervals from the real line. Assume
<s1,t1> in A and <s2,t2> in B , intervals in S . We say that
<s1,t1> < <s2,t2> iff t1 < s2 . The strict order relation
< is an abbreviation for < ^ =/ .
Suppose now that <s1,t1> < <s2,t2> . The axiom of completeness states
the existence of xi in S
such that <s1,t1> < xi < <s2,t2> .
I reply in advance to your next question: "Why did you write < instead
of < ?". The reply is that < means "less or equal", that is
xi may
not be equal to t1 or s2 , but it can do so. Note that since
xi
belongs to S , then xi is an interval. This is also meant as a reply
to your question about the dividing instant problem.
I could not penetrate the rest of your message.
Best Regards
Sergio
A comment on Sergio's reply to Jixin:
| An interval from the real-line is an ordered set of real numbers limited
by its end-points, which are not necessarily included in the set.
| It seems from this that the set of intervals is supposed to include open,
half-open and closed intervals; is that right? (Or do you mean to say that
there may be some doubt about whether a particular interval does or does
not include its endpoints? If the latter, this is not the usual notion of
'interval' as used in real analysis, and you need to explain further.)
| Suppose S consists of intervals from the real line. Assume
<s1,t1> in A and <s2,t2> in B , intervals in S . We say that
<s1,t1> < <s2,t2> iff t1 < s2 . The strict order relation
< is an abbreviation for < logical-and =/ .
| It follows then that for intervals, < implies < except for
pointlike intervals (single-point closed intervals) since if t1 < s2 ,
the intervals <s1,t1> and <s2,t2> cannot be equal unless
s1 = t1 = s2 = t2 .
| Suppose now that <s1,t1> < <s2,t2> . The axiom of completeness states
the existence of xi in S such that
<s1,t1> < xi < <s2,t2> .
| Consider the closed intervals [p, q] and [q, r] with p < q < r .
These satisfy < and hence satisfy < , but there is no interval
between them. Hence, your axiom is false for intervals on the real line.
Pat Hayes
To Sergio, who wrote:
| The (temporal) domain S , as I meant, may consist either of
time-points xor of time-intervals (exclusive "or").
An interval from the real-line is an ordered set of real numbers
limited by its end-points, which are not necessarily included in the
set.
Suppose S consists of intervals from the real line. Assume
<s1,t1> in A and <s2,t2> in B , intervals in S . We say that
<s1,t1> < <s2,t2> iff t1 < s2 . The strict order relation
< is an abbreviation for < logical-and =/ .
Suppose now that <s1,t1> < <s2,t2> . The axiom of completeness
states the existence of
xi in S such that <s1,t1> < xi < <s2,t2> .
I reply in advance to your next question: "Why did you
write < instead of < ?". The reply is that < means "less or
equal", that is xi may not be equal to t1 or s2 , but it can
do so. Note that since xi belongs to S , then xi is an
interval. This is also meant as a reply to your question about the
dividing instant problem.
I could not penetrate the rest of your message.
| First of all, what you wrote in the above didn't solve the Dividing
Instant Problem at all!
Anyway, you have claimed that "The (temporal) domain S may
consist either of time-points or (exclusive-or) of time-intervals",
and "an interval from the real-line is an ordered set of real numbers
limited by its end-points, which are not necessarily included in the
set."
In this case, can your intervals be "pointlike"? That is, for an interval
<s,t> , is s allowed to be equal to t ? In other words, can a set
representing an interval be a singleton? As I suggested in my former
response, the anwser has to be yes (see below). That is, if
your domain S contains non-pointlike intervals, then, to satisfy the
so-called completeness property, the domain S must contains
singletons (or namely points!) as well. Therefore, all my
former questions for you still apply.
I have shown in my former message that if your
domain S contains intervals, it must contain points as well.
However, you
claimed that S does not consist both of time-points and intervals
since you specially claimed that your "or" is exclusive-or). I would
like to use your notation to show this again.
In fact, you define the (partial) relation " < " as follows:
<s1,t1> < <s2,t2> iff t1 < s2 . Consider
the case that interval <s1,t1> in A and interval <s2,t2> in B ,
satisfying <s1,t1> < <s2,t2> , and t1 = s2 (this is a valid case
according to your definition). To fulfil the completeness property,
there exists a xi in domain S such that
<s1,t1> < xi < <s2,t2> .
Let xi = <s,t> . Again, by the definition of " < "
between intervals,
we have t1 < s and t < s2 . However, remember t1 = s2 , we infer that
it is impossible for s < t . Therefore, we reach that s = t . That is
xi must be a point (pointlike)!
By the way, it seems that your description of the axiom of
completeness is not a first-order one.
Jixin
In ENRAC 24.4.1998 I made a typing mistake. I wrote: " xi may not
be equal to t1 or s2 , but it can do so", while it should obviously
be: "assuming xi = <xi1, xi2> ,
then xi1 may not be equal to t1
and xi2 may not be equal to s2 , but they can do so".
In reply to Jixin Ma (ENRAC 23.4 and 24.4 1998) -- completion:
| So, you think intervals are not needed? Anyway, our arguments...
| I am actually skeptic about the need of a temporal domain which
includes time-intervals. There are many convincing arguments that a
temporal domain consisting of time-points is good enough in many
different situations (Newtonian mechanics and Thermodynamics, for
instance, as well as Sandewall's underlying semantics for K-IA ),
and I see no reason why I should pursue a different path.
| ... about the convenience of using intervals are based on the belief
of the need of them.
| This is why I originally asked for some convincing argument(s) for the
plausibility of this approach. According to the standard scientific
methodology, in fact, we shall build on top of already existent
solutions, and be consistent. Just to make an example, suppose one
refuses a classical notion (continuity?), and encounters the problems
that this notion was used to solve (the dividing instant?); it is
surely not consistent to justify the need for a novel approach via
the claim that the problems he encountered can not be solved by the
notion he just refused. The notion of semi-continuity, for instance,
has dignity, and its plausibility is far to be based on the belief
that continuity is not needed... The case of time-intervals is clearly
safer; one may simply give a preliminary example and show some objective
advantages when using time-intervals instead of time-points.
| Premise: It is evident that if you assume the axiom of completeness,
the domain S can just be continuous, while if you do not assume the
axiom of completeness then S is necessarily discrete.
|
|
Wrong! Even if you do not asssume the axiom of completeness, it is
still not nessarily discrete.
| Yes, I agree. I realize I wrote that sentence having in mind the
basic time structure on my paper. The question holds properly if you
do not assume any axiom of density other than the one I stated.
Concerning the dividing instant problem, which seems to summarize
what is left from your objections, please read below.
In reply to Pat Hayes (ENRAC 24.4.1998):
As posted in my original message, I have not yet seen any explanation
why an alternative notion of continuous structure is needed at all?
Probably, in order to prevent any misunderstanding, I should have
included an additional sentence like "... is needed at all, within the
search of those non-monotonic logics which purpose is to formalize
common sense reasoning when reasoning about actions and change", but
I thought it was evident, as the title of this Newsletter reminds.
In particular, in the same message, I asked to give at least one
convincing argument on the need of a notion which is an alternative
to the classical one, along the lines: "the problem P of temporal
reasoning about actions and change can not be solved adopting the
axiom of completeness", or "the axiom of completeness is too strong
an assumption for our purposes; axiom A is better suited,
because..." (
>>>>star
)
You and Jixin Ma proposed the "dividing instant problem", apropos of
the problem of switching on the light, and argued the axiom of
completeness inadequate for solving that problem. The formulation
I gave in ENRAC 24.4.1998, with today's minor adjustment, gives the
evidence on how the axiom of completeness is, instead, safe with
respect to the dividing instant problem. You and Jixin based your
argument on the fact that I do not allow the domain S to hold points
"and" intervals, so that if S admits just intervals then the
dividing point p can not exist. I refuted that argument by simply
observing an interval from the real line may have equal end-points.
You also gave other examples, but you did not explain how they
relate to the world of "Reasoning about Actions and Change". In
particular, and I somehow repeat myself, it is not evident that one
needs a temporal domain with non-homogeneous continuity (let me say it
is even less evident the need of the imaginary number i in our
temporal structure). Does there exist at least one representative
problem of reasoning about actions and change that can not be solved
adopting the axiom of completeness, so that to justify a temporal
domain with non-homogeneous continuity? (and I repeated (
>>>>star
))
You also gave an informal argument on the plausibility of a
temporal structure which formalizes the perceived smooth flux and
perceived fast flux of time (ENRAC 21.4.1998). I refuted that
plausibility with my contribution to ENRAC 23.4.1998.
(Is it really ``free of context'' to you ?)
Best Regards
Sergio
In reply to Pat Hayes (ENRAC 3.5.1998)
| An interval from the real-line is an ordered set of real numbers limited
by its end-points, which are not necessarily included in the set.
|
| It seems from this that the set of intervals is supposed to include open,
half-open and closed intervals; is that right? (Or do you mean to say that
there may be some doubt about whether a particular interval does or does
not include its endpoints? If the latter, this is not the usual notion of
'interval' as used in real analysis, and you need to explain further.)
| The former case is the one I meant.
You posed a good question, which may call into the present debate the
possible relations between epistemological and ontological
assumptions, at least within the "Features and Fluents" framework.
If we assume the epistemological assumption K (accurate and
complete information about actions), then occurrences of actions are
also supposed to give no doubtful information whether the scheduled
time interval where they are supposed to be performed does or does
not include its endpoints, so that the latter case from the quoted
text must not hold. Probably the case may hold within "Mo".
| Suppose S consists of intervals from the real line. Assume
<s1,t1> in A and <s2,t2> in B , intervals in S . We say that
<s1,t1> < <s2,t2> iff t1 < s2 . The strict order relation
|
| < is an abbreviation for < logical-and =/ .
It follows then that for intervals, < implies < except for
pointlike intervals (single-point closed intervals) since if t1 < s2 ,
the intervals <s1,t1> and <s2,t2> cannot be equal unless
s1 = t1 = s2 = t2 .
| < does not necessarily imply < , as in the case of
<2,5> < <5,9> , which is a valid case with respect to < .
You are right concerning the case whether <s1,t1> may be equal to
<s2,t2> , but this does not really affects the axiom of completeness
and, into the slightest question, it may be easily fixed.
| Suppose now that <s1,t1> < <s2,t2> . The axiom of completeness states
the existence of xi in S such that
<s1,t1> < xi < <s2,t2> .
|
| Consider the closed intervals [p, q] and [q, r] with
p < q < r .
These satisfy < and hence satisfy < , but there is no interval
between them. Hence, your axiom is false for intervals on the real line.
| The closed intervals [p, q] and [q, r] , with p < q < r , do not
fulfill the relation <p,q> < <q,r> , hence they do not make a
valid counterexample.
Best regards
Sergio
In ENRAC 3.5 (980521), Sergio wrote:
| I am actually skeptic about the need of a temporal domain which
includes time-intervals. There are many convincing arguments that a
temporal domain consisting of time-points is good enough in many
different situations (Newtonian mechanics and Thermodynamics, for
instance, as well as Sandewall's underlying semantics
for K-IA ), and
I see no reason why I should pursue a different path.
| You said here "a temporal domain consisting of time-points is good
enough in many different situations". Is this ("manyu") a convincing
argument for general treatments? Anyway, the fact that "you see no
reason why you should pursue a different path" does not mean others
don't see/have the reason (see below).
| ... about the convenience of using intervals are based on the
belief of the need of them.
|
| ... According to the standard scientific
methodology, in fact, we shall build on top of already existent
solutions, and be consistent. Just to make an example, suppose one
refuses a classical notion (continuity?), and encounters the
problems that this notion was used to solve (the dividing instant?);
it is surely not consistent to justify the need for a novel approach
via the claim that the problems he encountered can not be solved by
the notion he just refused.
| Have you applied the above arguments to that one proposed by youself?
Sorry, I am here again using your question to ask you.
Anyway, while I (and many others) have seen the convenience of using
intervals, I can also see the need of them. In fact, there have been
quite a lot of examples (many) in the literature that demonstrated the
need of time-intervals (or time-periods). Haven't you ever encountered
any one of them? Or you simply cannot see anyone of them is
convincing?
All right, let's just have a look at the example of throwing a ball up
into the air. As I showed in ENRAC 1.4 (98033) (one may disagree with
this), the motion of the ball can be modelled by a quantity space of
three elements: going-up, stationary, and going-down. Firstly, or at
least, we can see here the convenience of using intervals. In fact, we
can conveniently associate the property that "the ball changes its
position" with some time-intervals. Secondly, let's see if we indeed
need time-intervals. Without the notion of time-intervals (neither
primitive nor derived from time-points), can you just associate such a
property with time-points? Yes, we may associate it with a pair of
points. However, this doesn't mean that the property holds at these
points. What it really means is that the property holds for the time
periods denoted by the pair of points. Aren't these time periods in fact
time intervals?
It is important to note, up to now in the above, I just talked about
the need of the notion of intervals. As for how to characterise
intervals (e.g., are intervals taken as primitvie or derived
structures from time-points?) is another important issue, and this
issue, again, has been addressed in the literature for a long time.
The Point Is: while we were/are discussing/arguing about some broader
issues on temporal ontology, you just jumped in and asked "why an
alternative notion of continuous structure is needed at all?" First of
all, the "continuity" (or more truly, density) is not the main issue
we are talking about. The fundamental question is if we need to
address and how to addess time intervals. Based on such a discussion,
in the case that intervals are taken as temporal primitive, then, we
are talking about how to characterise some corresponding issues
including dense/discrete structures. But your questions and
arguments/replies do not seem to follow this. As stated in the former
replies from both Pat and myself, first of all, the dense structure
does not have to be characterised in terms of the only form of the
so-called "axiom of completeness". Also, in the case where
time-intervals are involved (even they are still point-based, let
alone in the case they are taken as primitive), such an axiom doesn't
simply apply. In fact, I have shown this twice with different notations
in this discussion. I will point out more problems in detail below in
my response to your reply to Pat.
| Concerning the dividing instant problem, which seems to summarize
what is left from your objections, please read below.
| As I already stated, your approach does not solve the
DIP at all. In fact, it seems that you don't realise the DIP in the
way as we are talking about (see below).
| In reply to Pat Hayes (ENRAC 24.4.1998):
|
| As posted in my original message, I have not yet seen any
explanation why an alternative notion of continuous structure is
needed at all?
| Still not yet?
| You and Jixin Ma proposed the "dividing instant problem", apropos of
the problem of switching on the light, and argued the axiom of
completeness inadequate for solving that problem. The formulation I
gave in ENRAC 24.4.1998, with today's minor adjustment, gives the
evidence on how the axiom of completeness is, instead, safe with
respect to the dividing instant problem. You and Jixin based your
argument on the fact that I do not allow the domain S to hold
points "and" intervals, so that if S admits just intervals then
the dividing point p can not exist. I refuted that argument by
simply observing an interval from the real line may have equal
end-points.
| You claimed already that your domain S contains points or
(exclusive-or) intervals? To fulfill the axiom of
completeness, you must allow your intervals to be possibly some
singletons (i.e., a set of single point). In other words, if your S
contains intervals, it should also contain singletons (points). The
real problem is that even if you allow your intervals to be singletons,
the Dividing Instant Problem is still there, and in fact more
obviously. Do you agree with this?
| The closed intervals [p, q] and [q, r] , with p < q < r , do not
fulfill the relation <p,q> < <q,r> , hence they do not make a
valid counterexample.
| Pat's example becomes invalid only after you made the "minor
adjustment" that replaces the relation < in your hypothesis
<s1,t1> < <s2,t2> by < , that is <s1,t1> < <s2,t2> .
(Is this an alternative?)
It follows that you do need alternation, doesn't it?
(Note that this is just for the case
when you construct intervals out of points. In the case where
intervals are taken as primitive, the need of such alternative is
indeed more conceptually necessary). However, your adjustment is not
enough, or you haven't reached the proper form for general treatments.
In fact, you need to address the issue regarding different cases. To see
this, you may just consider the difference between the case where at
least one of <s1,t1> and <s2,t2> is "closed" at t1 ( = s2 ), and the
case where both <s1,t1> and <s2,t2> are "open" at t1 ( = s2 ). In the
former case, you need use < in the hypothesis; otherwise, Pat's
example will be a valid counterexample. In the latter case, you need
use < in the hypothesis; otherwise, your axiom cannot prevent a
"gap" between <s1, t1) and (s2, t2> ,
that is, there is no guarantee
that the singleton [t1, t1] is contained in S (Do you think this is
consistent with the "classical" concept of contiunity?).
Jixin
In reply to Jixin Ma (ENRAC 7.5.1998)
| Pat's example becomes invalid only after you made the "minor
adjustment" that replaces the relation < in your hypothesis
<s1,t1> < <s2,t2> by < , that is <s1,t1> < <s2,t2> . (Is this an
alternative?)
| Ex falso sequitur quodlibet!
The only one "minor adjustment" I made consists in the first four
lines of my contribution to ENRAC 3.5.1998, where no inequality
appears at all. Concerning the hypothesis, I remind you what I wrote
in ENRAC 24.4.1998:
| Suppose now that <s1,t1> < <s2,t2> . The axiom of completeness states
the existence of xi in S such that
<s1,t1> < xi < <s2,t2> .
| I observe Pat quoted me correctly in ENRAC 3.5.1998.
| So, you do need alternation, don't? (And this is just for the case ...
| The axiom of completeness imposes < , so no "alternation" is
needed at all. The reason why I wrote < instead of < is
simply due to my need to stress the example, since the case
<s1,t1> = <s2,t2> is trivial. If you like to check, the reference
is ENRAC 24.4.1998.
| ...when you construct intervals out of points. In the case where
intervals are taken as primitive, the need of such alternative is
indeed more conceptually necessary). However, your adjustment is not
enough, or you haven't reached the proper form for general treatments.
In fact, you need address the issue regarding different cases. To see
this, you may just consider the difference between the case where at
least one of <s1,t1> and <s2,t2> is "closed" at t1 ( = s2 ), and the
case where both <s1,t1> and <s2,t2> are "open" at t1 ( = s2 ). In the
former case, you need use < in the hypothesis; otherwise, Pat's
example will be a valid counterexample. In the latter case, you need
| ... at least one is closed. So we have, since s2 = t1 :
1. [s1, t1] < [t1, t2]
2. [s1, t1] < (t1, t2]
3. [s1, t1) < [t1, t2]
where xi = [t1, t1] in S in all cases.
Note I used < , as required by the axiom of completeness.
If I use < , as you recommend, then all
cases trivially fail.
Pat's example:
- trivially fails when using < ,
- trivially succeeds ( xi = [q, q] ) when using < .
| use < in the hypothesis; otherwise, your axiom cannot not prevent a
"gap" between <s1, t1) and (s2, t2> , that is, there is no guarantee
that the singleton [t1, t1] is contained in S (Do you think this is
consistent with the "classical" concept of contiunity?).
| ... the latter case. So we have, since s2 = t1 :
4. <s1, t1) < (t1, t2>
where xi = [t1, t1] in S . I used < here too.
So, the axiom of completeness has no problems with your examples.
Concerning the first part of your message, as you wrote in it, it was
entirely based on the DIP problem and the above argument-examples.
| while I (and many others) have seen the convenience of using
intervals, I can also see the need of them. In fact, there have been
quite a lot of examples (MANY) in the literature that demonstrated the
need of time-intervals (or time-periods). Haven't you ever encountered
any one of them? Or you simply cannot see anyone of them is
convincing?
| The problem about intervals is whether one needs to introduce them
into the temporal domain, and the few argument-examples I encountered
are far from being convincing. Furthermore, in this debate, you and
Hayes proposed the DIP, and I refuted it.
There exists at least one problem (within R.A.C) that needs to
introduce intervals into the temporal domain?
The other problem is:
There exists at least one problem (within R.A.C.) that can not be
solved with a continuous temporal domain, so that to justify a
temporal domain with non-uniform continuity?
This debate aims at generality, surely does not aim at completeness
of case examples. If many examples do exist, then this is the proper
debate where at least the most representative of them should appear
"naked" under the spotlight, for general benefit. On the other
hand, I note that more than two weeks are now passed from my
criticism, and no such representative example appeared.
Sergio
Reply to Sergio Brandano (ENRAC 8.5.1998)
| The only one "minor adjustment" I made consists in the first four
lines of my contribution to ENRAC 3.5.1998, where no inequality
appears at all. Concerning the hypothesis, I remind you what I wrote
in ENRAC 24.4.1998:
| When you presented the (classical) axiom of completeness (ENRAC
23.4.1998), you used < in the hypothesis (You are now still using
it, see below). But for the case where elements in domain S are just
intervals, you used < instead (otherwise, Pat's example is valid,
see below).
| The axiom of completeness imposes < , so no "alternation" is
needed at all. The reason why I wrote < instead of < is
simply due to my need to stress the example, since the case
<s1,t1> = <s2,t2> is trivial. If you like to check, the reference
is ENRAC 24.4.1998.
| Again, if you re-claim that the axiom of completeness imposes < ,
then Pat's example is valid (see below).
| ...when you construct intervals out of points. In the case where
intervals are taken as primitive, the need of such alternative is
indeed more conceptually necessary). However, your adjustment is not
enough, or you haven't reached the proper form for general treatments.
In fact, you need address the issue regarding different cases. To see
this, you may just consider the difference between the case where at
least one of <s1,t1> and <s2,t2> is "closed" at t1 ( = s2 ),
and the
case where both <s1,t1> and <s2,t2> are "open" at t1 ( = s2 ). In the
former case, you need use < in the hypothesis; otherwise, Pat's
example will be a valid counterexample. In the latter case, you need
|
|
... at least one is closed. So we have, since s2 = t1 :
-
[s1, t1] < [t1, t2]
[s1, t1] < (t1, t2]
[s1, t1) < [t1, t2]
where xi = [t1, t1] in S in all cases.
Note I used < , as required by the axiom of completeness.
If I use $lt$, as you recommend, then all cases trivially fail.
Pat's example:
- trivially fails when using
< ,
trivially succeeds ( xi = [q, q] ) when using < .
|
This is exactly what I wanted to show and have shown a few times now.
That is, in the case there your domain S contains intervals,
to fulfill the axiom of completeness, S has to contains singletons
(single points) as well, not as you specially claimed that S contains
points or (exclusive-or) intervals. My observation that Pat's
example would be valid is under your assumption that the domain S
refuses to take both intervals and singletons (points). I think when
Pat gave the example, he also followed this assumption of yours.
(Actually, Pat did specially claim that "unless you allow intervals
consisting of a single point" when he gave the example in ENRAC
24.4.1998).
| use < in the hypothesis; otherwise, your axiom cannot not prevent a
"gap" between <s1, t1) and (s2, t2> , that is, there is no guarantee
that the singleton [t1, t1] is contained in S (Do you think this is
consistent with the "classical" concept of contiunity?).
|
|
... the latter case. So we have, since s2 = t1 :
4. [s1, t1) < (t1, t2]
where xi = [t1, t1] in S . I used < here too.
| This is exactly what I have suggested in my message to you (see
above).
| The problem about intervals is whether one needs to introduce them
into the temporal domain, and the few argument-examples I encountered
are far from being convincing. Furthermore, in this debate, you and
Hayes proposed the DIP, and I refuted it.
| The DIP was proposed much earlier in the literature. What you have
done, is just a re-Writing of a model where the DIP arises. Could
you please check carefully what exactly is the problem and if it can
be solved by your formulation?
Jixin
[S.B.]
| Suppose S consists of intervals from the real line. Assume
<s1,t1> in A and <s2,t2> in B , intervals in S . We say that
<s1,t1> < <s2,t2> iff t1 < s2 . The strict order relation
< is an abbreviation for < ^ =/ .
| [P.H.]
| It follows then that for intervals, < implies < except for
pointlike intervals (single-point closed intervals) since if t1 < s2 ,
the intervals <s1,t1> and <s2,t2> cannot be equal unless
s1 = t1 = s2 = t2 .
| [S.B.]
| < does not necessarily imply < , as in the case of
<2,5> < <5,9> , which is a valid case with respect to < .
| Clearly, <2,5> is not equal to <5,9> , ie <2,5> =/ <5,9> .
That is, both < and =/ hold between those intervals. According
to Sergio's definition (in italics above) it follows that the relation
< must hold between them. In general, if p is not equal to q , then
the intervals <p,q> and <q,r> cannot be equal, so must be =/ , but are
also < , and therefore must be < . The rest of his message in ENRAC
4.5 (98042) makes the same error, and the subsequent confusion has been
noted by Jixin in the later discussion.
Sergios point seems to be that one can describe the line in terms of
conventional open and closed intervals in such a way that no 'gaps' appear,
so that the 'interval' between the open intervals
(a, b) and (b, c) is the
closed interval [b] containing a single point. Yes, of course: that
is not
at issue. We are not claiming to have found some basic flaw in conventional
real analysis. The question is whether this standard mathematical view of
the line is the most suitable for capturing linguistic intuition or for
action reasoning. For example, we want to be able to assert that a light is
off before time t and on after time t without having to commit ourselves
to its being either on or off at that time, but also without sacrificing
the assumption that it is always either on or off. Of course we could just
decide that periods of off-ness, say, shall be left-open-right-closed, or
some other convention: but this is arbitrary, ad-hoc and theoretically
unsatisfactory, since the intuition we would like to capture is that the
question (of the light being on or off at b) simply doesnt arise: it just
goes on then, that's all. That is the intuition which Allens interval
calculus, where intervals can simply meet, is intended to capture. The
fact that this calculus violates Sergios pet axiom doesnt seem to be a very
important matter for further discussion.
In any case, many users of temporal ontologies do not wish to assume
continuity or even density, for reasons of their own, and so a
general-purpose temporal ontology should therefore not make such
unnecessarily strong assumptions as Sergio's 'completeness' axiom. Temporal
database technology usually assumes times are discrete, for example.
Pat Hayes
In reply to Pat Hayes [ENRAC 17.5]
|
Sergios point seems to be that one can describe the line in terms of
conventional open and closed intervals in such a way that no 'gaps' appear,
so that the 'interval' between the open intervals
(a, b) and (b, c) is the
closed interval [b] containing a single point.
|
In my example, the temporal domain $S$ consists exclusively of
intervals from the real-line and no total order relation was imposed,
so that the interpretation for which one can describe the line with
that domain is surely not correct. If one restricts S to the case
of intervals from the positive real-line, for example, then S is a
tree with a continuum of branches.
| The question is whether this standard mathematical view of the line
is the most suitable for capturing linguistic intuition or for
action reasoning. For example, we want to be able to assert that a
light is off before time t and on after time t
| In your approach the switching action takes place at time t , while
the process of turning on the light does not necessarily require a
single time-point. Anyway, according to the classical approach
(linear point-based temporal-domain) one may easily capture the
proper intuition with a simple function of time:
\[ to\_switch(s,\tau,t) =
\cases{ {\rm off} & if $\tau = s$ \cr
{\rm on} \vee {\rm off} & if $\tau \in (s,t)$ \cr
{\rm on} & if $\tau = t$}
\]
where the interval [s, t] is the switching interval, which may also
be point-like, of course. The switching interval is properly that
time-interval where the action of switching on the light is
performed. Please note that s and t may also be variables.
Now, assume the light is initially "off". Then it will remain "off"
until s , because of the assumption of inertia. From s to t the
value will change according to the definition of the switching
action, while at t the light will be surely "on". The light will be
"on" at t and at every timepoint after t until another action
will occur to change it's state. As simple as this...
| In any case, many users of temporal ontologies do not wish to assume
continuity or even density, for reasons of their own, and so a
general-purpose temporal ontology should therefore not make such
unnecessarily strong assumptions as Sergio's 'completeness'
axiom. Temporal database technology usually assumes times are
discrete, for example.
| (a) When implementing whatever formalism on a resource-bounded
machine, compromises are never enough, otherwise we shall bound
ourselves when designing them. On the other hand, I like to observe that
humans formulated the notion of continuity in their mind, although
humans too are resource-bounded. It is also true that not every human
knows about continuity, but some of them does it. In any case the
notion of continuity does exists, at least on paper.
(b) The problem that comes when aiming at generality (as in this
debate?), is one shall include all possible cases, instead of
excluding those which are of no interest for someone. Personally, I
like to consider problems within Newtonian physics as being an
important part of Reasoning about Actions and Changes, where
continuity plays an fundamental role when idealizing the physical
world. Furthermore a continuous temporal-domain subsumes the case of
integer time, as well as the case of rational time, so that it is
general enough to embrace all possible cases.
Therefore, in my opinion, the axiom of
completeness is a good assumption, at least
within the classical point-based approach. The purpose of my example
with S , was an ad-hoc example to show the axiom of completeness
adequate as well for an interval-based temporal domain. As an
immediate consequence of that result, the dividing instant problem is
then solved for those interval-based formalisms where this axiom will
be properly adopted. Still remains the question whether at least one
problem exists that needs to introduce intervals into the temporal
domain. Personally, I think this problem does not really exists.
Finally, I would like to resume Jixin's original statements in ENRAC 13.3:
| (1) For general treatment, both intervals and points are needed.
(2) To overcome the so-called Dividing Instant Problem, that is the
problem in specifying whether intervals is "open" or "closed" at their
ending-points, both intervals and points should be treated as
primitive on the same footing.
| My reply to (1) was: [ENRAC 8.5]
| Does there exist at least one problem (within R.A.C) that needs to
introduce intervals into the temporal domain?
| My reply to (2) was the example with S , where just intervals are
needed.
Best Regards
Sergio
Pat,
You wrote (ENRAC 17.5]:
|
The question is whether this standard mathematical view of
the line is the most suitable for capturing linguistic intuition or for
action reasoning. For example, we want to be able to assert that a light is
off before time t and on after time t without having to commit ourselves
to its being either on or off at that time, but also without sacrificing
the assumption that it is always either on or off.
|
The most natural way of dealing with such a situation, it seems to me,
is to admit that one's axioms allow two kinds of models: those where
the light is on at time t , and those where it is off. The assumption
that light is either on or off holds in each of the models, but the
axioms don't imply one or the other.
The alternative that has occurred in the present discussion, and proposed
as a way of dealing with the "dividing instant problem"
is the use of a punctuated timeline where the domain
of timepoints is chosen e.g. as Re - B where Re is the real numbers
and B is a finite set of "breakpoints", typically chosen as the
times where actions start or end. If we ignore questions of how
this domain is specified in terms of axioms or how intervals are
formally defined, this seems to be the essence of how one would deal
with the DIP without going to very unconventional concepts of time.
Given that the standard view already exists, it seems worthwhile to
understand what the concrete reasons would be for replacing it with
the punctuated timeline ontology. Apart from the purely subjective reasons
(that is, some people prefer to do it that way) I wonder what
results have been obtained using punctuated or other nonstandard
ontologies, and which are not trivially translatable back to the standard
view. The following would seem to be interesting results for this
purpose:
- Expressiveness in a concrete scenario. This ought to be a scenario
that can't be properly expressed in the standard ontology, but which
can be expressed in the nonstandard one.
- Algorithms or other implementation techniques. Is there some
algorithm that works with a punctuated timeline and that outperforms
those that don't?
- Validation and range of applicability results. These are results
stating that or when a certain entailment method is correct, that
is, it obtains exactly the intended models viz conclusions. Have some
such results been obtained using nonstandard ontologies?
Please feel free to add more categories to the list. (In the methodology
discussion at the NRAC workshop at IJCAI last year we got to more than
ten such categories).
I realize of course the purely philosophical and/or logical interest
in analyzing different possible concepts of time, but from an AI point
of view there is no point in pursuing an approach if it doesn't
deliver any results. I haven't been able to see any indications of such
concrete results from the present discussion or from the articles that
have been referenced in it. (That most work in temporal databases uses
discrete time doesn't say anything about the choice between a standard
or a punctuated real timeline, does it?)
Although I'd prefer to see algorithmic results or range of applicability
results, a few words about expressiveness since after all that's also
a relevant type of achievement. One nice scenario for hybrid change (that
is, continuous and discrete) is the impact problem that Persson and
Staflin used in their ECAI 1990 paper [c-ecai-90-497]. It goes like
this: two solid spheres B and C are lying side by side on a horizontal
surface, B to the left of C. A third sphere, A, comes rolling from the
left and hits B. As we know from physics, the result of the impact is
that A stops, B stays, and C starts moving towards the right. This
particular exercise has been solved in 1990, but is there now some
harder one that makes essential use of punctuated time ontology?
Erik
References:
c-ecai-90-497 | Tommy Persson and Lennart Staflin.
A Causation Theory for a Logic of Continuous Change.
Proc. European Conference on Artificial Intelligence, 1990, pp. 497-502.
Also available as
Linköping technical report Nr. 90-18 [postscript]. |
|