home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!biosci!uwm.edu!rpi!utcsri!helios.physics.utoronto.ca!alchemy.chem.utoronto.ca!mroussel
- From: mroussel@alchemy.chem.utoronto.ca (Marc Roussel)
- Newsgroups: bionet.info-theory
- Subject: Re: A Mathematical Fireside Chat at 300K about 0K
- Message-ID: <1992Nov15.163436.29176@alchemy.chem.utoronto.ca>
- Date: 15 Nov 92 16:34:36 GMT
- References: <1992Nov14.160532.13636@alchemy.chem.utoronto.ca> <1992Nov14.234006.12491@news2.cis.umn.edu>
- Distribution: bionet
- Organization: Department of Chemistry, University of Toronto
- Lines: 72
-
- In article <1992Nov14.234006.12491@news2.cis.umn.edu> burchard@geom.umn.edu
- (Paul Burchard) writes:
- >In article <1992Nov14.160532.13636@alchemy.chem.utoronto.ca>
- >mroussel@alchemy.chem.utoronto.ca (Marc Roussel) writes:
- >> In article <1992Nov14.080455.6956@news2.cis.umn.edu>
- >> burchard@horizon.math.utah.edu (Paul Burchard) writes:
- >> >Thermodynamics may be understood as the study of energy and the way it
- >> >is distributed among the parts of a system. The fundamental assumption
- >> >that the subject makes about energy is that it is a *positive* quantity.
- >>
- >> That's wrong. We have no way of assigning a zero point to energy
- >> and we only measure differences. All zero points for energy are purely
- >> conventional and recognized as so by all physicists I know.
- >
- >I am not wrong. First of all, at this point I am only claiming that
- >thermodynamics *wants* positive energy. This is necessary because otherwise
- >there is no concept of a "heat bath" which is characterized purely by its
- >temperature---unless you impose the positivity constraint, there will be
- >no well-defined maximum entropy distribution.
-
- Thermodynamics gets on quite well without a positive energy scale.
- Temperature in thermodynamics is defined as that variable of state which
- characterizes thermal equilibrium. A heat bath is large body of
- constant temperature, i.e. one which is in internal thermal equilibrium.
- Thermal equilibrium means that no heat flows in the absence of an
- adiabatic wall and heat is a form of energy which flows without
- macroscopic work being done either on or by the system. I never had to
- involve ANY energy scale in this definition. If you want to stand
- physics on its head, you're going to have to tell us what's wrong with
- this, our normal way of thinking of temperature.
- Rather than thermodynamics, I think you want to discuss statistical
- mechanics. (They are not entirely the same thing, although modern usage
- sadly confuses the two more and more.) If you look at how we write down
- partition functions and derive other quantities from them, you find that
- differences in state functions are independent of the choice of zero
- point energy. I don't give a hoot about anything else because I can't
- measure absolute values of energy or entropy.
-
- >Second, many quantum field theories do provide a natural zero energy
- >reference point, such that a difference between this reference level and
- >the ground state has physical consequences.
-
- I will grant you that. However, I don't know of a q.f.t. that
- predicts thermodynamic consequences of this zero reference point. If
- you do, let me know.
-
- >> >In this way we find that our best guess is the exponential distribution
- >> >p(E) = (1/T) e^{-E/T}.
- [...]
- >That's just the *most generic* distribution having
- >a specific temperature; therefore it should be the distribution in the case
- >of a heat bath which we assume to have no further properties.
-
- This is backwards from the way we do physics. It doesn't tell me
- how to measure temperature, therefore it is physically meaningless. It
- is senseless to pretend that we don't know what temperature is.
- Temperature is a primitive concept which we get from thermodynamics.
- The job of statistical mechanics is to connect this primitive to
- microscopic mechanics, not to invent temperature de novo.
-
- >The entropy of any continuous probability distribution is infinitely larger
- >than that of a discrete one. I challenge you to make sense of the limiting
- >entropy of a sequence of continuous distributions, when the entropy measured
- >from the continuous point of view is approaching minus infinity.
-
- I am going to be honest and admit that I have no idea what you're
- talking about here. I honestly don't see the pathology you describe in
- the theory. Maybe if you rephrased this in terms of partition functions
- I would get it.
-
- Marc R. Roussel
- mroussel@alchemy.chem.utoronto.ca
-