PROC MODEL is capable of solving systems of differential equations by a variety of numerical integration techniques. Example 17.10 in the ERS documentation optimizes 5 base ODEs and 36 PDEs in the parameters.
How well it optimizes the parameters will boil down to how much data, how much noise, and how well the model is specified.
Thank for your advice. The problem at hand is actually on parameter estimation on Heston and Nandi model and Christoffersen model on garch option pricing. I can't find the ERS documentation... is it a typo and should be ETS? and I can't find example 17.10 ...
The documentation I have at hand is SAS/ETS 9.1 User Guide.