v3.0

v3.0 of Fidimag.

v3.0a2

• Includes a version of the Fast Multipole and Barnes Hut methods for calculating the dipolar field distribution in atomistic simulations.

• Includes 4th Order Uniaxial Anisotropy for the first time

Alpha release of the upcoming 3.0 version

Major update since the 2.9 version. Highlights:

• Updated NEBM class

  • Option to set both climbing and falling images
  • Added new integrators (Verlet)
  • Updated calculation of geodesic angle
  • Removed old Cartesian NEBM class
  • Restructured C libs
  • Added more tests

• Added String method as an alternative to NEBM. So far only working with Verlet integrator

• Moved String/NEBM to a new chain methods class

• Updated Fidimag build file to avoid repeated Cython compilations

• Updated micromagnetic DMI class with more DMI terms and more simple way to implement new DMI types

• Updates to demag class

• Updates to the TimeZeeman class

• Started tracking Fidimag citations

• Trying to fix Travis build

  • Notebooks are now in a separate build but still need to be updated to pass the nbval tests

• Added Steepest Descent method as a minimiser class

• In progress:

  • Fast Multipole method
  • VTK writer in C to speed up output files and remove pyVTK dependency

Stable release

A new stable release for Fidimag, where:

• The micromagnetic DMI class and the C library for the calculation of the DMI field, have been restructured and simplified. New DMI terms and nomenclature has been added

• The atomistic hexagonal mesh has been modified to compute up to 9 shells of neighbours

• The atomistic exchange class has been simplified and a new function has been added to calculate exchange beyond nearest neighbours

• Class managing the integrators has been updated to allow calculations with scipy

• Updated CI with Travis

• Finished moving drivers, like STT, to the common folder, to make the library structure more consistent

Stable Fidimag Release

This release comes with many bug fixes which highly improve the stability of the code.
In addition, documentation has been added since the latest release.

To keep improving the code, it might be necessary to move the C llg classes from the atomistic folder to the common folder, and simplify the VTK saving routines to separate them from the driver. This can be done in a next major release.

Classes refactoring

This new 2.0 release comes with a major refactoring of the old Sim and llg classes. Now Sim is properly used as a class to keep most of the simulation and material parameters. The integrators no longer derive from the llg class but from a new driver class. In the common folder we moved to different _base classes, common methods for both atomistic and micromagnetic simulations, avoiding code repetition. Most of the methods have also been documented. Regarding the driver classes, it now stores all of the LLG related parameters (alpha, gama, etc).

This version also integrates the new NEBM code that uses Geodesic distances.

There are still things that need to be improved:

• Functions like saving VTKs should be in a separated library rather than in the driver class
• Better and updated documentation for the new classes and Fidimag structure
• A plan to easily allow new integrators to be added to the code
• Revision/test of the stochastic methods
• Merge integrators into a common class since the main difference between the atomistic and micromagnetic codes comes from the field classes

Fidimag Stable Version

This release is before we make a major refactoring to the main code. Tests are passing and some of the standard problems are reproduced successfully. The code works perfectly with Python 2.7 but it seems we need Anaconda to make it work with Python 3.x.

v0.0.1

Merge branch 'master' of github.com:computationalmodelling/fidimag