October 25th, 2012, 3:27 pm
In the book (page 290) the authors mention the Predictor Corrector method for 1 factor SDE (btw I suppose alpha and beta are in range [0,1]?)In general, we can choose:1. Standard drift2. Corrected drift3. Averaging using trapezoidal [n, n+1] (2 function calls per step)4. Averaging using midpoint [n,n+1] (1 function call per step)JK/DW use options 1 + 3. In MC1 each of the 4 options is supported.From experiments options 2 + 4 give best results and better than 1 + 3. A deeper analysis of 4 options using Matlab and C++ is a good test. E.g. 1+4 in Matlab. BTW no one seems to use 4 altho' you see it all over the place in numerical analysis.
Last edited by
Cuchulainn on October 24th, 2012, 10:00 pm, edited 1 time in total.