search

mumax / 3

GPU-accelerated micromagnetic simulator
Other
447 stars 150 forks source link

Any differences in version 3.9.3 and version 3.10beta will lead to different result with the exactly same input txt #239

Closed yjzhang97 closed 4 years ago

yjzhang97 commented 4 years ago

The input txt: setgridsize(10, 10, 1) setcellsize(4e-9, 4e-9, 2e-9) setGeom(cylinder(40e-9, 2e-9)) Msat = 1e6 Aex = 2e-11 alpha = 0.0003 m = uniform(0, 0.8, 1) lambda = 1.5 epsilonprime = 0.083 Pol = 0.6 fixedlayer = vector(0, 0, 1) B_ext = vector(0, 0, 5e-1) J = vector(0, 0, -4e9) tableautosave(1e-11) run(4e-9)

The result in 3.9.3 version: 3.9.3.txt 3 9 3

The result in 3.10 beta version: 3.10beta.txt 3 10

Considering the change sign of the epsilonprime in version 3.10beta, we applied minus epsilonprime, but the result is still different as following:

3.10beta-differ.txt 3 10_differ

Any changes in version 3.10 we ignored cause these result? Or what's wrong with our code or settings?

JeroenMulkers commented 4 years ago

This is due to a bugfix in 3.10. See discussion in https://github.com/mumax/3/issues/99 for more information.