Seite - 27 - in Differential Geometrical Theory of Statistics
Bild der Seite - 27 -
Text der Seite - 27 -
Entropy2016,18, 370
6.2. ThermodynamicEquilibria andThermodynamicFunctions
6.2.1.AssumptionsMadein thisSection.
Any Hamiltonian H defined on a symplectic manifold (M,ω) considered in this section
will be assumed to be smooth, bounded by below and such that for any real b > 0, each one
of the three functions, defined on M, z → exp(−bH(z)), z → ∣∣H(z)∣∣exp(−bH(z)) and z
→(
H(z) )2exp(−bH(z)) iseverywheresmaller thansomefunctiondefinedonM integrablewithrespect
to theLiouvillemeasureλω. The integralswhichdefine
P(b)= ∫
M exp(−bH)dλω and Eρb(H)= ∫
M Hexp(−bH)dλω
thereforeconverge.
Proposition11. LetHbeaHamiltoniandefinedonasymplecticmanifold (M,ω) satisfying theassumptions
indicated inSection6.2.1. Foranyreal b>0 let
P(b)= ∫
M exp(−bH)dλω and ρb= 1P(b) exp(−bH)
be thevalueat bof thepartition functionPandtheprobabilitydensityof theGibbs statistical stateassociated
tob, and
E(b)=Eρb(H)= 1
P(b) ∫
M Hexp(−bH)dλω
be themeanvalueofHwithrespect to theprobabilitydensityρb. Thefirst andsecondderivativeswith respect to
b of thepartition functionPexist, are continuous functionsof bgivenby
dP(b)
db =−P(b)E(b) , d
2P(b)
db2 = ∫
M H2exp(−bH)dλω=P(b)Eρb(H2) .
Thederivativewith respect to b of the functionEexists and is a continuous functionof bgivenby
dE(b)
db =− 1
P(b) ∫
M ( H−Eρb(H) )2dλω=−Eρb((H−Eρb(H))2) .
LetS(b)be the entropy s(ρb)of theGibbs statistical state associated to b. The functionScanbe expressed
in termsofPandEas
S(b)= log (
P(b) )
+bE(b) .
Itsderivativewith respect to b exists and is a continuous functionof bgivenby
dS(b)
db = b dE(b)
db .
Proof. UsingtheassumptionsSection6.2.1,wesee that the functionsb →P(b)andb →Eρb(H)=E(b),
definedbyintegralsonM,haveaderivativewithrespect tobwhich iscontinuousandwhichcanbe
calculatedbyderivationunder the sign ∫
M . The indicated results easily follow, ifweobserve that
for any function f on M such that Eρb(f) and Eρb(f2) exist, we have the formula,well known in
Probability theory,
Eρb(f2)− (Eρb(f))2=Eρb((f−Eρb(f))2) .
6.2.2. PhysicalMeaningof the IntroducedFunctions
Let us consider aphysical system, for example agas contained in avessel boundedby rigid,
thermally insulatedwalls, at rest in aGalilean reference frame. Weassume that its evolution can
27
Differential Geometrical Theory of Statistics
- Titel
- Differential Geometrical Theory of Statistics
- Autoren
- Frédéric Barbaresco
- Frank Nielsen
- Herausgeber
- MDPI
- Ort
- Basel
- Datum
- 2017
- Sprache
- englisch
- Lizenz
- CC BY-NC-ND 4.0
- ISBN
- 978-3-03842-425-3
- Abmessungen
- 17.0 x 24.4 cm
- Seiten
- 476
- Schlagwörter
- Entropy, Coding Theory, Maximum entropy, Information geometry, Computational Information Geometry, Hessian Geometry, Divergence Geometry, Information topology, Cohomology, Shape Space, Statistical physics, Thermodynamics
- Kategorien
- Naturwissenschaften Physik