topic numerical integrator dodesol_mk52lfn and tolerance in Intel® oneAPI Math Kernel Library

numerical integrator dodesol_mk52lfn and tolerance

Brian_Murphy — Fri, 08 Mar 2019 19:22:46 GMT

I am using dodesol_mk52lfn as a replacement for the Gear numerical integrator in the IMSL. dodesol_mk52lfn has a tolerance input argument, ep, which I do not understand. dodesol_mk52lfn output matches IMSL quite well with ep on the order of 0.1 to 1.0, whereas it seems that this argument ought to be on the order of 1e-6. For instance, the example problem in the documentation uses ep=1e-6 and says this produces accuracy of "about 5%".

ep DOUBLE PRECISION

(input) Relative error tolerance, which must be small enough. The code cannot ensure the requested accuracy for ep<1.d-9. This parameter is used to control the step size.

Thanks,

Brian Murphy

Brian, may be would be better

Gennady_F_Intel — Sat, 09 Mar 2019 01:00:48 GMT

Brian, may be would be better to post such question to the imsl forum? https://forums.roguewave.com/forumdisplay.php?110-IMSL-Numerical-Libraries.

But my question is about the

Brian_Murphy — Sat, 09 Mar 2019 13:42:36 GMT

But my question is about the MKL routine, not about the IMSL routine. I need help with the MKL routine. What the documentation says about the "relative error tolerance" is not consistent with the results I get using the routine. Is it possible the "relative error" is defined in a non-standard way?