Some remarks on the tensorial linearization of general and linearly separable boolean functions

The coefficients of the tensorial linearized expression of any boolean (in particular, linearly separable) function are exhibited as solutions of simple algebraic equations. A nontrivial connection with three-valued functions is shown to provide a tool which might serve to express a linearly separable function in terms of its Chow parameters. This formalism is meant for applications in neural net theory and in the study of statistical and critical phenomena of various kinds. © 1973 Springer-Verlag.