A proposed *MathFit* initiative offers funding from the EPSRC to encourage exploitation of mathematics in computer science departments.

This is one of a number of similar programs which have encouraged collaboration between academic disciplines, and hint at a kind of customer/supplier relationship between some academic departments.

On this my first observation is, that I think its a good idea. I think that supplementing the primary "peer review" process controlling allocation of research funding by customer/supplier-like relationships within academia is a promising idea.

Anyway, apart from that I don't have much to say about MathFit. I have stronger feelings about a proposal which has not been put. So I thought I'd sketch it out.

If we consider the natural customer/supplier relationships between the disciplines funded by the EPSRC (Engineering and Physical Sciences Research Council) then it seems to me that:

- Logic
- provides reasoning tools for mathematics which enable greater certainty, rigour, and precision.
- Mathematics
- supplies to the physical sciences the wherewithal to build and test mathematical models of physical systems
- Physical Sciences
- supply experimentally tested mathematical models of physical systems to the engineering sciences
- Engineering Sciences
- fill out the details necessary for real world applications of scientific theories

- Information Technology
- provides the means throughout the chain to automate as much as possible of the above processes.

In the light of this (admittedly simplistic, possibly controversial) perspective (seen through the eyes of an information technologist obsessed with logic), the following programme is suggested:

A kind of stacked "T" shape is suggested, giving a broad general coverage with two nested foci of interest and support.

The first proposed focus is on *symbolic* techniques, the second on methods grounded in symbolic *logic*.
It is presumed that the scope of symbolic techniques subsumes that of numerical techniques, and that both are properly contained in what can be realised within a (sufficiently automated) formal logical framework.

A particular interest is suggested in methods which in a practically beneficial way promote methods from numeric to symbolic or from symbolic to logical frameworks.

An even balance is suggested in the influence of strategic considerations such as these and more concrete short term requirements of *the customers*, scientists and engineers.
The funding will also support full participation by scientists and engineers in the programme.

NB: this is just a fantasy. There is no funding (for MathKit, there is for *MathFit*), and the EPSRC doesn't take much notice of me.

© created 1996/2/23 modified 1996/8/6