Skip to content

Inhomogeneous Msat#115

Merged
godsic merged 8 commits intomasterfrom
interfaces
Jul 19, 2017
Merged

Inhomogeneous Msat#115
godsic merged 8 commits intomasterfrom
interfaces

Conversation

@JeroenMulkers
Copy link
Copy Markdown
Collaborator

These changes resolve the occurring "increasing-energy" problems for samples with an inhomogeneous Msat.

  • The harmonic mean A_i/M_i and A_j/M_j is no longer used as prefactor for the exchange interaction between cell i and j. Instead, the prefactor A_ij/M_i is used in cell i and A_ij/M_j in cell j.
  • The default value for A_ij is the harmonic mean of A_i and A_j (arbitrary choice).
  • It is possible to rescale A_ij with ext_ScaleExchange (old feature).
  • It is also possible to set A_ij to a specified value with ext_InterExchange (new feature).
  • The same approach is used for the DMI.
  • Simulations with homogeneous Msat before and after changes should give the same result.
  • All tests run fine after minor trivial changes in test/exchcoupling.mx3 and test/dindcoupling.mx3

There is no longer a difference between aexchParam and dexchParam structs when eliminating the 1e18 and 1e9 factors in their update methods. The deduplication of these structs to a single exchParam makes to code cleaner.

@JeroenMulkers JeroenMulkers requested review from barnex and godsic July 13, 2017 15:06
@godsic
Copy link
Copy Markdown
Contributor

godsic commented Jul 14, 2017

@JeroenMulkers Have you tested with non-homogeneous Aex? OOMMF also does harmonic mean, so it makes sense to check against OOMMF to ensure mumax3 implementation is consistent.

@JeroenMulkers
Copy link
Copy Markdown
Collaborator Author

I did not test a non-homogeneous Aex explicitly (only through the already existing unit tests) because the changes are irrelevant for a non-homogeneous Aex.

Problems did arise for a non-homogeneous Msat. I used this script to check if the made changes solve the increasing-energy problem. I think this script is not really suitable as a unit test due to its complexity and the long run time.

Shall I write a more simple unit test with a non-homogeneous Msat, possibly combined with a non-homogeneous Aex?

@godsic
Copy link
Copy Markdown
Contributor

godsic commented Jul 14, 2017

@JeroenMulkers Are you OK to test against OOMMF?

@JeroenMulkers
Copy link
Copy Markdown
Collaborator Author

I will give it a try.

@JeroenMulkers
Copy link
Copy Markdown
Collaborator Author

In the unit test I calculate the exchange energy for a system with an inhomogeneous Aex and Msat, and a random magnetization. The obtained numerical value corresponds exactly with the exchange energy obtained using oommf.

@godsic
Copy link
Copy Markdown
Contributor

godsic commented Jul 19, 2017 via email

@JeroenMulkers
Copy link
Copy Markdown
Collaborator Author

Thank you. The OOMMF script is also added in commit 328a1df.

@godsic godsic merged commit e2e8506 into master Jul 19, 2017
@godsic
Copy link
Copy Markdown
Contributor

godsic commented Jul 19, 2017

@JeroenMulkers Just checked the values of the exchange energy and they are consistent with yours.

godsic added a commit that referenced this pull request Jul 19, 2017
Fix exchange boundary conditions, many thanks to Attila Kakay for repoting this issue.
@JeroenMulkers JeroenMulkers deleted the interfaces branch August 5, 2020 12:51
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants