home *** CD-ROM | disk | FTP | other *** search
- Xref: sparky sci.math:15278 sci.physics:19283
- Path: sparky!uunet!sun-barr!cs.utexas.edu!asuvax!chnews!sedona!bhoughto
- From: bhoughto@sedona.intel.com (Blair P. Houghton)
- Newsgroups: sci.math,sci.physics
- Subject: Re: Three-sided coin
- Date: 20 Nov 1992 06:20:49 GMT
- Organization: Intel Corp., Chandler, Arizona
- Lines: 45
- Message-ID: <1ei041INNec2@chnews.intel.com>
- References: <1992Nov17.130147.26746@oracorp.com>
- NNTP-Posting-Host: alfalfa.intel.com
-
- In article <1992Nov17.130147.26746@oracorp.com> daryl@oracorp.com (Daryl McCullough) writes:
- >bhoughto@sedona.intel.com (Blair P. Houghton) writes:
- >>Summary: probability is distributed according to entropy, not energy
- >
- >If we are talking about a flipped coin, I don't think it is
- >distributed according to either. The probability distribution on coins
- >depends on the way the coin is flipped and the initial orientation.
- >For example, if my coin is a cylinder (three sides), and flipping
- >consists of rolling it, then it has essentially zero probability of
- >landing on either of its two ends.
-
- Don't do that.
-
- You are further constraining the system, and thus
- altering its entropy at the outset.
-
- You are correct that, given the new initial conditions, the
- probability is almost unity that the coin will remain
- "on-edge." This is because you have made the change of
- state from on-edge to on-face an entropy *decreasing*
- event.
-
- You can't constrain the system with anything other than
- "what is the probability that the coin will end up on its
- edge?" Even if you say the coin is flipped, you have
- constrained the form of motion and therefore altered the
- entropy.
-
- In order for the entropic analysis to work, the method of
- reorienting the coin, and indeed whether the coin is
- reoriented or left alone, must remain unknown, and be
- allowed to assume all possible values; otherwise, you end
- up with a morass of conditions and trajectories and phase
- spaces and the need to define *everything* [*] rather than
- *nothing*.
-
- --Blair
- "It's sort of how it
- all works, really..."
-
- [*] This is okay if you *can* define everything. You
- simply partition the entropy further according to the
- number of similar states and the gross probability of the
- dichotomies, but in this example you can't even define the
- probabilities of the type of floor covering...
-