The meeting with Askari went very well. I will send you and Peter a summary 
of points we agreed on for initial phase of the project, for your approval. 
Let me know if you want any clarification or changes. Basically it consists 
of just getting what we have now to run under NT, and testing it using a 
standard data set from J.P. Morgan (Hon tells me Miloje had already done this,
although I haven't been able to find the data set, and more importantly the 
data set parsing program, among his files - I will look some more and ask 
Miloje about it). Only things we agreed to add to the existing code are 
fairly simple - a couple of new distribution functions (should be able to 
get algorithms and/or code for this from textbooks or papers), and variable 
timesteps (this should be straightforward and not take long to implement).

Also agreed to look into effort required to do two other things (but not to
actually do them in the first phase, unless they turn out to be trivial).
One is probably trivial (checkpointing - code may already exist to do this, 
in which case it can be implemented very easily) and one probably non-trivial 
(computing interest rates using a stochastic equation rather than Monte Carlo). 

I think all this should be feasible in 7-8 weeks (i.e. beginning of July). 
However we will need the following ASAP:

1) PC with Windows NT for Chao-Wei to use in his office
2) Access to multi-processor NT machine
3) Visual C++ version 5.0 installed on both the above machines (this is 
   what Askari use for their program development)
4) Chao-Wei needs to be able to talk to someone at NPAC with expertise in 
   C++ development and multi-threading under NT
5) Also, eventually will probably need to upgrade the multi-processor NT 
   machine from 128 Mbytes of RAM to 256 Mbytes, which is a more likely 
   configuration for Wall St company running Askari software.