Hi all,yet another thread on Malliavin calculus...Basically, I'd like to apply the Malliavin technique tocompute greeks, say delta. The paper by Fournie et algives a quite generic formula:where .It seems quite typical to choose a_t = constant, however, did anyof you play with other shapes for a_t ?It seems to me that a good choice of a_t could lead to anestimator of delta with minimum variance, though the derivationis of course payoff dependent.So what do you use for a_t ? Also, any research paper tacklingwith this problem ?Any help would be greatly appreciated, thanks in advance.

Perhaps you find this paper by Glasserman interesting.Malliavin Greeks Without Malliavin Calculus

Thanks, I'm using this paper.The idea is that when you have densities, and usediscretization, the Likelyhood ratio estimators, whenaveraged along the path, converge to the Malliavin estimators.And Glasserman gives some condition on the a_t(or really a_i since we have discretized) for convergence,however it still leaves a lot of freedom in choosing a_t.Any other ideas ?