Statistical mechanics of combinatorial partitions, and their limit shapes (Q677678): Difference between revisions

By a partition we mean a division of a positive integer or a vector with positive integral coordinates into unordered summands of the same kind. Since Euler's time, this object has been of interest to specialists in different areas of mathematics, including number theory, combinatorics, probability theory, and others. A new problem in asymptotical partition theory, which arose some time ago, is the equation about the asymptotic (limit) shape, or configuration, of partitions with respect to some statistics. Here is the simplest setting of this problem. Let \(\mu^n\) be the uniform measure on the set of all partitions of the number \(n:\mu^n(\lambda)= p(n)^{-1}\), \(\lambda\in{\mathcal P}_n\), where \(p\) is Euler's function (partitio numerorum); the question is: whether one can normalize the partitions in such a way that, in some properly chosen space, the measures \(\mu_n\) have a weak limit on generalized partitions, and whether this limit is singular (a \(\delta\)-measure). In the last case, the limit measure is concentrated on a limit shape. An affirmative answer to these questions, as well as explicit formulas for limit shapes, for a number of statistics, are presented. We continue the study of similar problems and stress their close relationship with the statistical physics of ideal gas. The problem of limit configuration in this context is especially appropriate to traditional statistics (Bose, Fermi, and others), although it was not studied before, as far as I know. It can be solved by means of some variation of classical tools and is interpreted as a problem of limit distribution of the total energy of the system over the energy spectrum. Nevertheless, our passage to the limit is not literally equivalent to the case of thermodynamic limit (N-V-limit): in fact, we reduce the problem of limit distribution of energy to the case of fixed volume, as energy tends to infinity so that its growth does not necessarily match the growth of the number of particles. If the number of summands (particles) is not fixed (the case similar to the statistics of photons), then the natural growth of the number of particles is automatically determined by the statistics. The classical N-V-limit is also covered by our scheme by selecting the growth of the number of summands in a special way, but the corresponding combinatorial problems are still poorly studied; they are also related to the behavior of the convolution of distributions on the semigroup of partitions. The problem of limit shape for partitions was first posed by the author in connection with the asymptotic theory of classical groups and their representations, in particular, the symmetric group, but it should be considered in a more general framework of known problems of the random growth of shapes. Note that combinatorial schemes in combinatorics, group theory, representation theory, and additive number theory are much more various than ordinary models in the statistical physics of the ideal gas. In particular, the ergodicity of macro- and microcanonical ensembles can fail in natural problems. The stimulating nature of these problems becomes apparent, in particular, in the interaction of different methods of their solution. For example, the variational principle that suggests finding a limit shape as the minimum of some functional led to a new type of variational problems. In this paper, we present the main statements and a series of results concerning the limit shape. Their detailed exposition will appear elsewhere.