Misc Notes‎ > ‎

Physics and Computation


Newsgroups: net.physics
From: bre...@h-sc1.UUCP
Date: Sat, 29-Mar-86 23:22:25 EST
Local: Sun, Mar 30 1986 6:22 am
Subject: Re: grammars in physics

||Since any calculation can be performed by a turing machine,
||and since the language accepted by a turing manchine can
||be generated by an unrestricted grammar, what is the grammar
||that describes nature? More specifically is there a grammar
||for quantum mechanics?
|There are several things wrong with this idea, not the least
|of which is:  Nature does not have to perform calculations.
Yes, there are several things wrong with this idea, but
it is not that 'nature does not have to perform calculations'
(see below).

It is indeed interesting to look at physical laws and principles
from a computational point of view. 'Grammars' and 'Turing machines'
are, however, relatively inconvenient devices of doing so.
More appropriate models of physical reality based on computational
devices are 'cellular automata'. Cellular automata are
large (infinite) arrays of simple computing elements capable
of local communication.

Cellular automata can be the discrete counterparts of differential
equations, and, in fact, are used to model and compute approximations
to field equations, boundary value problems, &c.

Cellular automata also display dynamical properties that are
not found in their smooth counterparts. One of the most fundamental
differences is, of course, by definition, that they have finite
resolution in space, time, and, often, paramter space.

From the computational point of view, cellular automata
can be Turing equivalent.

Common examples of cellular automata are Life and video feedback.
The connection machine developed at M.I.T., and several computer prototypes
used for vision research have architectures reminiscent of
cellular automata.

After my thesis is done, I might post references and a more comprehensive
and detailed survey of cellular automata and their relationship to physics,
if interest warrants it.

Finally, there is (Doug, we do seem to have quite differing views
of physical reality) absolutely no reason why we couldn't or
shouldn't use physical models based on computation rather than on
differential equations for modelling nature. Since all the verification
of our current theories involves computation, I would argue that,
in a weird sort of way, our models of nature are in fact already
purely computational.


PS: let me stress again: I object to the statement 'Nature does not
have to perform calculations'. The reason is that it contains a
semantic confusion: 'calculations' are a human model of nature,
just like 'differential equations', 'groups', &c. Obviously (or
perhaps not so obviously), nature does not operate through the models
that we make of it. I might therefore with equal right say:
'Nature does not obey physics', but clearly this statement is
either tautological (namely if you interpret it to mean that
the existence of our physical models of nature is not a prerequisite
to the operation of nature), or it is a malformed version of
the sentence 'physics is probably not an accurate model of nature'.
Anyhow, as I have argued above, *computation may be as accurate
a model of nature as modern physics is*.

Newsgroups: net.physics
From: t...@talcott.UUCP (Thomas M. Breuel)
Date: Tue, 13-Aug-85 13:04:33 EDT
Local: Tues, Aug 13 1985 7:04 pm
Subject: QM and Multiple Worlds

In article <1...@sdcsvax.UUCP> , david...@sdcsvax.UUCP (Greg Davidson) writes:
>> From: g...@brl-tgr.ARPA (Doug Gwyn <gwyn>)
>>>  The multiple worlds interpretation has some severe problems ...
>> Not all the alternate worlds are equiprobable!  There is no
>> observable difference between the alternate-worlds QM and the
>> Copenhagen QM.  It happens that there IS a small chance that
> The probability of the multiple worlds is irrelevant to the dwellers
> therein.  Each world is a complete spacetime continuum separate from
> the others.  In some moments, at some places, in some of them, there is
> Maybe the latter worlds are less common, or maybe your thoughts fo
> normality, including your notions of what is probable, are due to
> the peculiar accidents of your world.  Once you accept the multiple

The 'multiple worlds interpretation' of QM is not a physical theory:
you cannot design an experiment to disprove it, since it postulates that
there is not interaction between its different worlds.

It makes no sense to speak of 'observing differences' among worlds,
or to generalise notions of normality to a fictitious ensemble of
such separate worlds. WE are the dwelles of THIS world, and by the
very defition of the word 'world' (in this context), the best thing
that natural science can do is to describe statistically the physical
laws that govern this world.