Grid functions of nonstandard analysis in the theory of distributions and in partial differential equations (Q1721948): Difference between revisions

This article treats a theory of generalized functions called grid functions which is based on nonstandard analysis and generalizes both distribution theory and the concept of Young measures. Grid functions are defined as functions on a uniform hyperfinite grid. In place of derivatives and integrals of these functions, one employs finite difference operators with infinitesimal step and hyperfinite sums, respectively. Based on a class \(S^\alpha\) of grid functions, which is the counterpart of classical \(C^\alpha\) functions, an algebra of test functions is defined which serves as a nonstandard counterpart of the usual space of test functions. This way, the author obtains that the space of grid functions is a vector space over \({}^* \mathbb{R}\), extending the space of distributions, compatible with the pointwise multiplication of \(S^0\)-functions and with a derivative which generalizes the distributional derivative, but for which a certain discrete version of the product rule holds. The embedding of distributions is non-canonical and is not fully coherent with derivatives. Moreover, a submodule of the space of grid functions is identified whose elements correspond to Young measures. As an application, the formulation of PDEs in the setting of grid functions is given and the relation between solutions by means of grid functions and solutions in the classical formulation is studied. Finally, the author obtains the notorious formula ``\(H H' = \frac{1}{2}H'\)'' for the Heaviside function \(H\), using the theory of grid functions; moreover, he gives an example of a variational problem without a classical minimum, but whose minimum in the sense of Young measures can also be obtained via the developed theory.