10.07.2015 Views

Information Theory, Inference, and Learning ... - Inference Group

Information Theory, Inference, and Learning ... - Inference Group

Information Theory, Inference, and Learning ... - Inference Group

SHOW MORE
SHOW LESS
  • No tags were found...

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

Copyright Cambridge University Press 2003. On-screen viewing permitted. Printing not permitted. http://www.cambridge.org/0521642981You can buy this book for 30 pounds or $50. See http://www.inference.phy.cam.ac.uk/mackay/itila/ for links.B.1: About phase transitions 603log ZN beta epsilonN log (2)(a)beta(c)var(E) N=24var(E) N=8betaFigure B.1. (a) Partition functionof toy system which shows a phasetransition for large N. The arrowmarks the point β c = log 2/ɛ. (b)The same, for larger N.(c) The variance of the energy ofthe system as a function of β fortwo system sizes. As N increasesthe variance has an increasinglysharp peak at the critical point β c .Contrast with figure B.2.log ZN beta epsilonN log (2)(b)betalog ZN beta epsilonN log (2)(a)beta(b)var(E) N=24var(E) N=8betaFigure B.2. The partition function(a) <strong>and</strong> energy-variance (b) of asystem consisting of Nindependent spins. The partitionfunction changes gradually fromone asymptote to the other,regardless of how large N is; thevariance of the energy does nothave a peak. The fluctuations arelargest at high temperature (smallβ) <strong>and</strong> scale linearly with systemsize N.The arrow marks the pointβ = ln 2(B.11)ɛat which these two asymptotes intersect. In the limit N → ∞, the graph ofln Z(β) becomes more <strong>and</strong> more sharply bent at this point (figure B.1b).The second derivative of ln Z, which describes the variance of the energyof the system, has a peak value, at β = ln 2/ɛ, roughly equal toN 2 ɛ 24 , (B.12)which corresponds to the system spending half of its time in the ground state<strong>and</strong> half its time in the other states.At this critical point, the heat capacity of this system is thus proportionalto N 2 ; the heat capacity per spin is proportional to N, which, for infinite N, isinfinite, in contrast to the behaviour of systems away from phase transitions,whose capacity per atom is a finite number.For comparison, figure B.2 shows the partition function <strong>and</strong> energy-varianceof the ordinary independent-spin system.More generallyPhase transitions can be categorized into ‘first-order’ <strong>and</strong> ‘continuous’ transitions.In a first-order phase transition, there is a discontinuous change of one

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!