I am not sure whether to ask this in Physics, Math, or Computer Science, but I will try Math.
I read a paper "Nonlinear nonequilibrium statistical mechanics approach to C3 systems." Lester Ingber. 9th MIT/ONR Workshop on C3 Systems: Naval Postgraduate School, Monterey, CA, 2-5 June 1986. pp. 237-244. In this paper Ingber claims a method based on neural physiology to solve constraint-based problems, such as those found in military operations research. Without detailing all the twists and turns of the paper, the basic theory is that:
...learning and retrieval mechanisms can be developed by first determining expansion coefficients of eigenfunction expansions of the differential Fokker--Planck distributions, e.g., considering stationary states as Hermite polynomials in neighborhoods of minima.
The most relevant extract from the blizzard of math in this paper is the following segment:
The idea being that one can solve a complex system of non-linear relationships which cannot be modelled as ODEs using the above Greek salad.
As a computer scientist I am unable to interpret the validity of these claims. I am hoping for an assessment of the validity of the paper before spending my time trying to understand it. Is this purported technique a potentially valid way to solve complex constraint systems, such as those found in operations research, or is it just a bunch of concocted BS?