Dear Developers, I've been using fipy-2.1.3 for quite some time and am very satisfied, but I decided to switch to newer version, 3.0.1-dev144-g950431e.
I made all required changes to my script and unfortunately I got different solutions with both versions. I rather strongly believe that solutions obtained with older fipy are correct. The original script is rather complicated. Here I attach a simpler version having though absolutely the same problem. The two equations describe the evolution of Br and Bt. The velocity field, v, is read from the attached file v.pkl. I tried following runs: RUN1: CFL~=10 (taking Vsh/R as the largest speed) timeStep=0.5 nrsteps=250 and RUN2: CFL~=1 (taking Vsh/R as the largest speed) timeStep=0.0125 nrsteps=1000 Fipy213 showed expected results in both cases with RUN2 being slightly better, and RUN1 pretty satisfactory. Fipy301 seems to show some fake diffusivity in both runs - solutions are not as sharp as in fipy213 case and are very smeared. Note, there are no diffusion terms in the equations. Besides, in RUN1 fipy301 solves bad for Br inner part: solution is higher than Bt whereas they should coincide. Moreover, fipy213 turned out to be ~15% faster. Could you please help me understand the problem? Thanks a lot! Regards, Igor. PS: I tried several versions of fipy-2.1.3: release from download page, trunk 569, 5214. 5214 is actually already solving exactly the same as fipy301. All prerequisite software is the same for all fipy versions. No problems with installation found.
test2vs3v2.py
Description: Binary data
v.pkl
Description: Binary data
_______________________________________________ fipy mailing list [email protected] http://www.ctcms.nist.gov/fipy [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]
