Economic Downturn

Topological Paradoxes of Time Measurement

Lecture Notes for a talk given
October 30,2001

U Illinois Circle Campus at Chicago

The ideas and constructions presented here are taken from the treatise Time, Euclidean Geometry and Relativity written in 1967 and revised several times, the latest being in the year 2000. It has formed the basis for presentations at the Institut Poincaré (1968), Trinity College in Dublin(1970), and Wesleyan University (2000) In the original paper we describe the axiomatic structure of time, and the associated issues of clock construction, in a linear, homogeneous, relativistic, cyclic , and linear-cyclic, ( which bears some resemblance to a quantum) , universe.

Definition: A linear universe U is one in which there is some way translating the temporal dimension into a spatial dimension. That is to say, the structure of U allows the measurement of duration by clocks to be replaced by a measurement of distance by rulers. An obvious example of a linear universe is the universe of Special Relativity, or any universe in which the Postulate of Relativity ( constancy of the speed of light) applies.

Let W be a non-linear universe, one in which no isomorphism can be established between time measurement and length measurement. We make the following assumptions:

  1. Given a length L, it is possible to fashion a ruler of exactly length L.

  2. Given a ruler of length R, it is always possible to fashion a ruler of some length R'< R (This does not assume that an R' can be with its length having some given proportion to R, only that one can always make a ruler shorter than any given ruler.

  3. Given a clock measuring duration T, it is always possible to fashion a clock measuring some shorter duration T'< T. Once again there is no assumption about the ratio of T to T'

  4. No time reversal. Time is always measured in the forward direction, which is asummed to be known.

  5. It is assumed that SPATIAL sub-spaces of W, those subspaces defined at any given instant of time, are Euclid-Hilbert. (Finite or Infinite)

  6. No such restriction applies to space. Rulers can be freely transported in all directions, freely rotated, etc.

Theorem

Given the above set of assumptions for a a non-linear universe W it is not possible, from the existence of a clock Co measuring a duration of length T, to construct, save by trial and error a clock measuring a duration of length d = (1/2) T . More generally, it is not possible to construct a clock measuring an interval of time aT, where a is any constant 0< a < 1. By the measurement of a duration by a clock, one means that the clock ticks only at the beginning and the end of the duration T.

However, in either a 1-dimensional, 2-dimensional or 3-dimensional Euclidean space, it is possible , using a ruler, to determine,from a given length L, the midpoint l= (1/2) L.

Corollary: if I have a finite collection of clocks ticking off durations t1 < t2 < t3 ....< tn , then I cannot, save by accident or trial and error, construct a clock which ticks in any predetermined interval less than the minimum t1.

Clocks versus Rulers

(A) RULERS

Definition : A RULER is a mechanical system that functions as instrument for measuring lengths. It's basic property is that the distance between its endpoints remains invariant under rotations, translations and reflections.

Fix a moment in time. Let L be a pre-assigned distance in W between end-points p1 and p2. The axioms of a Euclid-Hilbert universe allow one to construct the entire line segment S connecting p1 and p2 . We look at methods for constructing the midpoint of S in

    One dimension

  1. Two dimensions

  2. Three dimensions

THE ONE DIMENSIONAL CONSTRUCTION:

To determine the midpoint of a segment S = [p1,p2] of length L, by the motions of rulers in one spatial dimension.

Our assumptions allow us to make a ruler of length R < L. Lay off integral lengths of R along S, starting from p1. If R goes into L an integral number of times n , then L = nR . If n is even our work is finished.

If n > 1 is odd, or if R does not divide L exactly, then make a ruler of length R*< R , and compare the following lengths: (1) Ra = R* ; (2)Rb = R-R*. It is important to note that the property of free translation in space has made it possible to construct the length Rb .

Place the leftmost endpoints of rulers R and R* next to each other. If y is the terminal point of R*, z the terminal point of R, then Rb = Length[y,z ].

Either Ra or Rb must be less than (1/2)R in length.

Choose the shorter of these two lengths, and label it R1. Next lay off R1 against L . If R1 goes into L exactly, then we can compute a new number n1 such that L = n1R1. If n1 is even we are finished. If r1 does not exactly divide L , or if n1 is odd, then make a shorter ruler R1* The minimum of the two lengths R1** and R1 - R1* will be our next ruler R2.

This process, known as the Euclidean algorithm, can be continued indefinitely One thereby builds up a sequence of remainder lengths, R1, R2, R3..... If R and L are incommensurable, this sequence is infinite. It must converge to zero however since each remainder is less than or equal to 1/2 of the previous one.

Each Rk goes into L a certain number of times, say nk =[L/Rk] . Let hk = [(1/2)nk] and locate the point on the segment S at the distance dk= hkRk. Then the sequence of lengths {dk} must converge to the point (1/2)L .

Since there are no temporal restrictions on the measuring process, one can get around the Zeno Paradox by positing that each operation takes half the length of time of the previous. It is sufficient for our purposes to observe that the succession of rulers converges to zero.

THE TWO DIMENSIONAL CONSTRUCTION:Here One can find the midpoint of any segment by using the familiar construction from Euclidean Geometry involving parallel lines. All that is needed is a way of constructing parallel lines. This can be done with marked rulers, which are certainly permitted from our initial assumptions. One can restate this as follows: since rulers are postulated to be able to move about freely they can be employed effectively on the plane as compasses . The issues surrounding the use of compasses or marked rulers have nothing to do with the mechanical laws governing the space of the plane in which the construction takes place.

THE THREE DIMENSIONAL CONSTRUCTION. The compass as a mechanical system is allowed by the assumptions governing the universe W.

(B)CLOCKS

It is clear that the mechanical process of finding the mid-point of a temporal duration T, in the absence of the postulate of relativity,( or some other unambiguous way of mechanically setting up an isomorphism between temporal duration and spatial length. ) involves a host of new difficulties

Any isolated and closed system M with an identical complete set of state variables V at two distinct moments in time t1 and t2 , will pulse forever from and to this state in durations of equal length T = [t1, t2]

One may in fact take this as the definition of what is meant by an equal interval of time. Note that it does not depend on the measurement of time, nor on the assigning of a numerical metric to the instants t1 and t2.

This axiom is itself dependant on numerous conditions discussed in the papers cited above. It is also assumed of course that W is deterministic, not quantum.

  • Therefore: in a deterministic non-quantum, non-relativistic, non-linear universe W , all clocks are periodic non-reversible dynamical systems .

  • In a relativistic or linear universe, the postulate of relativity allows one to escape this periodicity requirement precisely because it asserts that every light quantum is a system with a constant dynamical state variable,c. We will now show that under our set of assumptions for a non-linear W that, given a clock C which pulses in periods of duration T, there can be no procedure (other than lucky accident) for constructing a clock C* of period d= (1/2)T .

    By assumption, given C with period T a clock C1 with period T1 < T can always be constructed. We wind up both clocks and set them going simultaneously at time t = 0. Obviously T1 and T tick together in the first period of C , we can compute n such that T = nT1. If n is even we are finished. Note that C1 was a lucky accident.

    If n is odd, we select a new clock, call it C1 and start again. If T1 does not exactly divide T there is an integer m >1 such that 0 < (m-1)T1 The interval J = duration[y,z] , between the terminal pulse of mC and the terminal pulse of C1 could, in the spatial situation, be used in the production of a Euclidean algorithm process leading to convergence to a midpoint, perhaps in the infinite limit. However because of the irreversibility of time there is no way, given only the existence of C and C1, to construct a clock pulsing in the period J.

    Any such construction must involve some way of "pushing" the initial point of the second cycle of C back to the terminal point of C1 , a mechanical action that is easily achieved with a ruler.

    This statement, which is clear yet informal here, is given a more rigorous treatment in the original paper.

    Note that, even if it were possible to construct a clock with period J, and J were incommensurable with T, the convergence of remainders cannot be guaranteed . This is because the procedure of building clocks to select between Ta = J , and Tb = C1- J , cannot be made without bringing in time reversal.

    It is perhaps a supercilious play on words to say that the carrying out of the full Euclidean algorithm process could "never" be accomplished because it would require an infinite amount ot time! In fact it isn't even possible to set it up.

    In the world of daily life in which both quantum and relativistic effects are disregarded there is no way to subdivide the periods of a clock without treating time as a spatial dimension by making the assumption of constant velocity in some mechanical system that effectively replaces time duration with spatial length. Since length is reversible while duration is not, one is in some sense 'cheating' by doing so. But this is what is in fact done.

    Bringing back relativity there are no absolute velocities, and one is obliged to rely on the speed of the light quantum as the only reliable way to get around the limitations of clocks as periodic systems, and subdivide arbitrary intervals of time.


  • Return to

    Home Page