home *** CD-ROM | disk | FTP | other *** search
- Xref: sparky sci.math:17387 sci.physics:21727
- Newsgroups: sci.math,sci.physics
- Path: sparky!uunet!psinntp!scylla!daryl
- From: daryl@oracorp.com (Daryl McCullough)
- Subject: Re: Bayes' theorem and QM
- Message-ID: <1992Dec24.101452.16194@oracorp.com>
- Organization: ORA Corporation
- Date: Thu, 24 Dec 1992 10:14:52 GMT
- Lines: 90
-
- bhoughto@sedona.intel.com (Blair P. Houghton) writes:
-
- >How would a "non-classical probability theory" look? What is
- >the limitation of "classical probability theory," and just
- >what is a "classical probability theory?"
-
- Classical and quantum probability are described in the book "Quantum
- Probability" by Stanley Gudder. There is nothing especially "quantum"
- about quantum probability, except for the fact that it can be applied
- to quantum mechanics.
-
- Briefly, the differences between classical and nonclassical
- probability theory is mostly about the issue of nonmeasurable sets.
- (Probability theory being a special case of measure theory).
-
- Nonmeasurable sets:
-
- Take the simple case of the good old Lebesgue measure for
- three-dimensional space. In this case measure is essentially
- the same as volume.
-
- Suppose that we know that a point particle is somewhere inside the
- unit cube centered at the origin (that is, its x coordinate is between
- + 1/2 and - 1/2, and so is its y coordinate, and so is its z
- coordinate). Suppose further that the particle is equally likely to be
- anywhere in that region. Then we can answer a question like: What is
- the probability that the point is in region A (where A is a subset of
- the unit cube)? The probability is simply the three-dimensional volume
- of region A.
-
- It would be nice if the volume of a region of space were always
- well-defined, but it is not. A demonstration due to Banach and Tarski
- showed that it is possible (mathematically, rather than physically
- possible) to decompose a sphere into a finite number of pieces and
- then recombine them by rotations and translations to get two complete
- spheres. It is easy to see that these pieces cannot possibly have
- volumes, since that would lead to the conclusion that the volume of
- two spheres equals the volume of one sphere.
-
- So one cannot always assume that *every* set has a measure (or volume,
- or probability).
-
- Sigma algebras:
-
- Of course, these pieces are very weird, and are not regions one is
- typically interested in. Therefore, a measure theory does not have to
- assign measures to every set, but only to the sets of interest. The
- assumption behind classical measure theory is that the measurable sets
- form a sigma algebra. All this means is that we have some collection
- of basis sets for which measures are given, and all other measurable
- sets are formed from these by combinations of the operations of:
-
- set difference,
- set intersection (countable intersections),
- set union (countable unions)
-
- These are the operations of a sigma algebra. These operations, though
- they won't necessarily give you all sets, gives a nice collection of
- sets, which have a nice, intuitive interpretation in the case of
- probability measures: Every set can be thought of as a proposition;
- for instance we can associate a set A with the proposition "The
- particle is somewhere in set A". The fact that the measurable sets
- form a sigma algebra means that any logical combination of such
- statements is given a probability of being true.
-
- Nonclassical measure theory:
-
- Nonclassical measure theory simply drops the assumption that the
- measurable sets form a sigma algebra. It allows for the possibility
- that proposition A may have a well-defined probability, and proposition
- B may also have a well-defined probability, while the proposition A and
- B may be given no probability whatsoever.
-
- The applicability to quantum mechanics should be obvious: wave
- functions allow us to calculate the probability that the particle can
- be found in region A, and they allow us to calculate the probability
- that its momentum is in region B, but they don't allow us to calculate
- the probability that the particle is in region A *and* its momentum is
- in region B. Quantum mechanics does not give a unique answer to such
- questions.
-
- So nonclassical probability theory differs from classical probability
- theory in that it may fail to give probabilities for logical
- combinations of meaningful propositions.
-
- Daryl McCullough
- ORA Corp.
- Ithaca, NY
-
-
-