Behavioral modeling

The maximum-term method is a consequence of the large numbers encountered in statistical mechanics. It states that under appropriate conditions the logarithm of a summation is essentially equal to the logarithm of the maximum term in the summation.

These conditions are (see also proof below) that (1) the number of terms in the sum is large and (2) the terms themselves scale exponentially with this number. A typical application is the calculation of a thermodynamic potential from a partition function. These functions often contain terms with factorials $n!$ which scale as $n^{1/2}n^{n}/e^{n}$ (Stirling's approximation).

Example

$\lim _{M\rightarrow \infty }(\ln {\sum _{N=1}^{M}N!})=\ln {M!}\$

Proof

Consider the sum

$S=\sum _{N=1}^{M}T_{N}\$

where $T_{N}$>0 for all N. Since all the terms are positive, the value of S must be greater than the value of the largest term, $T_{max}$, and less than the product of the number of terms and the value of the largest term. So we have

$T_{max}\leq S\leq MT_{max}.\$

Taking logarithm gives

$\ln T_{max}\leq \ln S\leq \ln T_{max}+\ln M.\$

In statistical mechanics often $T_{max}$ will be $O(e^{M})$: see Big O notation.

Here we have

$O(M)\leq \ln S\leq O(M)+\ln M.\$

For large M, ln M is negligible with respect to M itself, and so we can see that ln S is bounded from above and below by $\ln T_{max}$, and so

$\ln S=\ln T_{max}.\$

References

• D.A. McQuarrie, Statistical Mechanics. New York: Harper & Row, 1976.
• T.L. Hill, An Introduction to Statistical Thermodynamics. New York: Dover Publications, 1987