The Asymptotic Behavior of the Mayer Series Coefficients for a Dimer Gas on a Rectangular Lattice

Paul Federbush
2024-06-27
Abstract:This paper continues and complements the research in the earlier version of this paper (essentially Part 3 and Part 4 herein.) We now assume as suggested at the end of Part 4 that $b(n)$ is asymptotically of the form in eq.(A1). \begin{equation}\label{A1} \tag{A1} (-1)^{n+1} * b(n) \sim \exp{( k_{-1} n + k_{0} ln(n) + \frac{k_{1}}{n} + \frac{k_{2}}{n^{2}}...)} \end{equation} Using the details of the six approximations used in Part 3, corresponding to the six values of $r$, $ 1 \leq r \leq 6$, in dimensions $d$ equal $2$. $3$, $5$, $11$ and $20$ we find in Part 1 an approximate value for the right side of eq.(A1) keeping the three terms in the exponent in $k_{-1}$, $k_{0}$, and $k_{1}$. In the range $5 \leq n \leq 20$ the two sides of eq.(A1) may be said to agree roughly to 5 parts per 100. ( With an appropriate choice of a constant of proportionality. ) In Part 2 an approximation is found keeping the term in $k_{2}$ also that may be said to agree to 5 parts in 1000 in the range $8 \leq n \leq 20$. Not only is it amazing that the relation in eq.(A1) seems to hold, it is equally amazing that it is so accurate for such small values of $n$.
Mathematical Physics,Statistical Mechanics
What problem does this paper attempt to address?