FP in Mathematical/Numerical Programming

Has anyone applied functional programming to general numerical applications?

I am not really talking about high performance mathematical applications

where Sisal might be appropriate, or other applications where more esoteric

languages such as J/APL could be employed, but something more akin to

exploratory statistical analysis or matrix programming. The languages

currently used might be Matlab, S or Gauss, which are just convenient

scripting interfaces to established numerical analysis libraries.

My question is whether FP is an appropriate "scripting" language paradigm

for such interactive numerical applications? One could argue that a matrix

equation expressed in Matlab is functional, but I am really considering the

aspects of FP such as higher order functions, laziness, formal reasoning

etc, resulting in possible elegance and/or efficiency. For instance, some

kind of laziness is implemented in S (a statistical analysis language

developed by bell labs), in order to efficiently handle large datasets.

Could modern lazy FPL theory be used instead?

I would be interested to find any information on this subject. I apologize

in advance if this is an irrelevant topic as I am no expert in FP.

Thanks,

Peter.