Abstract
As we saw in Sect. 1.3, the complete solution of a statistical mechanical model is known once an explicit expression has been obtained for the partition function. In this chapter we shall discuss methods which involve the derivation of expansions of partition functions and free energies, or their derivatives, in series of powers of temperature-dependent variables. These series are called low temperature, or high temperature, respectively according to whether the expansion variable tends to zero as T → 0 or T → ∞. If an expression for the coefficient of the general term were known then a series would constitute an exact solution. However, the phrase ‘series methods’ refers primarily to the term-by-term calculation of as many coefficients as possible and the deduction of results for the model from the behaviour of a limited number of coefficient values. Low- and high-temperature series expansions provide approximations to the thermodynamic functions of a model in their respective ranges of temperature. They are, however, more often used to study critical properties, specifically to obtain critical points and exponents. Suppose that
is some thermodynamic function of the model of interest, with ζ a function of temperature appropriate to high-temperature or low-temperature series.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Preview
Unable to display preview. Download preview PDF.
References
This terminology is not related to the exponent renormalization described in Volume 1, Sect. 9.3.
See also the article by Joyce and Guttmann in Graves-Morris (1973) and a discussion of this method in Guttmann (1989).
See Griffiths (1972) for this and more general cases and Sect. 4.2 for a discussion of the thermodynamic limit for a one-component lattice gas.
This identity was conjectured by Feynmann in some unpublished lecture notes where he used it to develop a variation of the method of Kac and Ward (1952) for solving the Ising model.
For the form of 6(q) see Sykes (1961) or Domb (1974a).
Hence the alternative name ‘connected graph expansion’. Note that a multi-bonded graph is connected if and only if its silhouette is connected.
Other authors (for example Rushbrooke et al. 1974) use the standard square bracket pair [· · ·] for this quantity. We have used the special symbol [· · ·] to avoid confusion with the more general use of square brackets.
The name ‘D-vector model’ is due to Stanley (1968a, 1974). The names ‘n-vector model’ and ‘O(n) model’ are also used, the latter because, in the absence of symmetry breaking fields, there is invariance under rotations in the n-dimensional space of s(r). We use D rather than n to avoid confusion with other usages.
It is now generally believed (Stanley 1987) that the exponents of the quantum Heisenberg model are independent of the spin s and, since the classical Heisenberg model is given by the limit s → ∞, this too will have the same exponents.
Personal communication to Professor Guttmann.
Square, honeycomb, simple cubic, body-centre cubic and diamond (see Volume 1, Appendix A.1).
Author information
Authors and Affiliations
Rights and permissions
Copyright information
© 1999 Springer-Verlag Berlin Heidelberg
About this chapter
Cite this chapter
Lavis, D.A., Bell, G.M. (1999). Series Methods. In: Statistical Mechanics of Lattice Systems. Texts and Monographs in Physics. Springer, Berlin, Heidelberg. https://doi.org/10.1007/978-3-662-10020-2_7
Download citation
DOI: https://doi.org/10.1007/978-3-662-10020-2_7
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-642-08410-2
Online ISBN: 978-3-662-10020-2
eBook Packages: Springer Book Archive