svn revision 27 should have functional parallel code! Certainly I’ve gotten much better at openmp in the past few days. Difficult to maintain a general API that is agnostic to the model details and still knows enough about the data formats to declare them appropriately for each thread. My lesson for my future coding: by writing with parallelization concepts in mind from the start I should be able to significantly reduce the overhead time when I go back and add openmp directives. Gillespie API still needs perfecting but should be able to push ahead with the science now.
still need to add const qualifiers and expand documentation.
- Next steps are the analytic approximations for the expected variance about the warning signals (variance of the variance estimate essentially). This also should be implemented in the code, which is more straight forward.
* * * * *
HTML Comment Box is loading comments…