The rescaling formulas , bring to mind formulas related to the Maslov dequantization of real numbers, see e.g. [4], [5]. The core of the Maslov dequantization is a family of semirings (recall that a semiring is a sort of ring, but without subtraction). As a set, each of is . The semiring operations and in are defined as follows: for by
Applying the terminology of quantization to this deformation, we must call a classical object, and with , quantum ones. The analogy with Quantum Mechanics motivated the following:
Correspondence Principle formulated by Litvinov and Maslov [4]. ``There exists a (heuristic) correspondence, in the spirit of the correspondence principle in Quantum Mechanics, between important, useful and interesting constructions and results over the field of real (or complex) numbers (or the semiring of all nonnegative numbers) and similar constructions and results over idempotent semirings.''
This principle proved to be very fruitful in a number of situations, see [4], [5]. According to the correspondence principle, the idempotent counterpart of a polynomial is a convex PL-function . As we have seen above, and are related not only on an heuristic level. In Section 1.6 we connected the graph of on logarithmic paper and the graph of by a continuous family of graphs .
Furthermore, is the graph of the function defined by
At last, the graph of on is the the graph of .
We see that the whole job of deforming to the graph of a piecewise linear convex function can be done by the Maslov dequantization: the deformation consists of the graphs of the same polynomial on for . The coefficients of this polynomial are logarithms of the coefficients of the original polynomial: . Since the map was denoted above by , we denote by the polynomial obtained from a polynomial with positive coefficients by replacing its coefficients with their logarithms. Thus . Since is a semiring homomorphism, the graph of on log paper is the graph of on . The other graphs involved into the deformation are the graphs of the same polynomial on . They coincide with the graphs on log paper of the preimages of under . Indeed, and .
For a real polynomial with positive coefficients, we shall call with the dequantizing family of polynomials.
The notion of polynomial is central in algebraic geometry. (I believe the subject of algebraic geometry would be better described by the name of polynomial geometry.) Since a polynomial over is presented so explicitly as a quantization of a piecewise linear convex function, one may expect to find along this line explicit relations between other objects and phenomena of algebraic geometry over and piecewise linear geometry. Indeed, in piecewise linear geometry the notion of piecewise linear convex function plays almost the same rôle as the notion of polynomial in algebraic geometry.
A representation of real algebraic geometry as a quantized PL-geometry may be rewarding in many ways. For example, in any quantization there are classical objects, i.e., objects which do not change much under the quantization. Objects of PL-geometry are easier to construct. If we knew conditions under which a PL object gives rise to a real algebraic object, which is classical with respect to the Maslov quantization, then we would have a simple way to construct real algebraic objects with controlled properties.