Name: Hrayr Matevosyan Date: 03/05/07-04:11:58 AM Z
Thank you for the suggestions. I will look more closely at FORM.
Actually I had two things in mind when I considered MathCode to generate stand-alone C++ code from Mathematica. I ran FeynCalc on a farm of 300 machines to successfully calculate Dirac amplitudes for some 40,000 diagrams and have publishable results, but the number of processes I could have going was limited by the number of licenses for Mathematica (could run only about 20 kernels at a time). So with stand-alone code the number of the processes can go up an order of magnitude, allowing to go to the next order of the calculation. Secondly, it might still be faster to run a binary file rather than a Mathematica notebook (no I/O delay, etc).
FORM seems to be useful for the grid-computing as well, but I simply
prefer to work with Mathematica and C++ (easier to develop complex code)
if it will be reasonable to do.
P.S. I am thinking to try and get a trial version of MathCode to test it
when I will have some free time (a good while later).