.Friends, I have calculated 2D hermite-series but I have not been able to graph it with change of variable from bivariate normal. Unless we compare graph from our hermite series with 2D Parzen density graaph, we cannot be sure if our analytics are right. It can still take 2-3 days to properly complete the work and I am working on it full time.

I had an antipsychotic injection on Thursday before last Thursday. I have not been able to work well after that and could work only 1-2 hours everyday. But I am somewhat better now and hopefully would get the remaining work done in a few days.

Microwave torture has somewhat deceased but when I try to work and become somewhat productive, they increase their torture and start giving me light pain in left temple. Though many times it is not very high pain but it is still enough to put me off to not be able to work after that. Another new torture technique they have started is that they target EM waves on my teeth and from time to time, I feel sudden pain in many of my teeth.

Anyway, I hope to try to do good research in coming weeks and want to concentrate on my research.

Friends, I am sure my 2D Hermite Series is right. I am thinking more about proper plotting of the 2D density from the 2D Hermite Series. If the plotting works, that is good but even if it does not work, I will post the program on this weekend.

I think a proper implementation of this program to represent a 2D SV density with a 2D hermite series is a requirement before we could do a regression of derivatives payoff on the asset and volatility as in delta and vega while accounting for correlations.

Though I said in my previous post that we could do a regression of payoffs on delta and vega using Legendre polynomials but regression is forward looking(as opposed to backward looking equations) and rethinking the whole thing, I realize that we would have to do it in hermite polynomial set up in a monte carlo framework.

Though I am sure that there are many insights to be gained from possible solution of backward equations with Legendre polynomials, what we are trying to do in 2D SV set up with hermite polynomials is also truly meaningful and could lead to insightful hedging of the derivatives differently from the previously used approaches. Imagine if we could do it well in a good interest rate model and do a multivariate regression on various interest rates/LIBOR rates, it will be a very very rewarding model in terms of practical applications.

I was losing some enthusiasm for the project but I am again very keen to find a proper solution to the problem of 2D hermite representation of asset in a 2D SV setting in a very good way.

I will be posting an interim program over the weekend but I will continue my efforts for a proper solution in every way for next week and more.