µMAG Evening Session
9th Joint MMM/Intermag Conference
January 7, 2004
Anaheim, CA
Bob McMichael, Session Chair
Agenda:
Bob McMichael, NIST
Werner Scholz, Seagate
Don Porter and Mike Donahue, NIST
Bob McMichael, moderator
Session report:
The MAG evening session ran from 7:30 to approximately 9:30 and drew
approximately 90 participants.
Welcome and Introduction
Bob McMichael opened the session with a very brief introduction to
muMAG and its objective of fostering improved communication within the
micromagnetic community and encouraging improvements in the reliability
of micromagnetic computations. The two technical efforts within muMAG
are
- Public micromagnetic code
- Standard Problems
magpar, finite element micromagnetics
Werner Scholz then gave a description of the finite element
micromagnetic software he developed as a graduate student at the
Technical University of Vienna under the guidance of Thomas Schrefl and
Josef Fidler. A few weeks before this meeting, Werner had announced
the public release of this software via the muMAG mailing list.
The features of magpar include
- entirely based on portable, free, open source software packages,
- highly portable to different hardware platforms, which range from
- simple PCs to massively parallel supercomputers,
- highly optimized and scalable,
- well integrated, combining static energy minimization and dynamic
- time integration.
- magpar is distributed under the GNU General Public License.
Werner described the open source software packages he incorporated to
handle the large matrices and vectors, to perform energy minimization or
LLG time integration and ultimately, to make the program run
efficiently on parallel processors. The efforts to parallelize the code
appear to be quite successful since the computation time scales
inversely with the number of processors used.
Official website and distribution site: http://magnet.atp.tuwien.ac.at/scholz/magpar/
The complete source code including documentation and a package
of examples are available for download.
Werner requests that all correspondence concerning magpar be directed
to the following address only: magpar@magnet.atp.tuwien.ac.at
Werner's presentation ended with a double round of applause and thanks
for this very important contribution to the micromagnetic community.
OMMF: Where it is, where is it going?
Don Porter and Mike Donahue reported on the status and future
directions for the NIST Object Oriented MicroMagnetic Framework,
(OOMMF).
Don described a maintenance release in the 1.1 branch of the OOMMF
software, which is centered on the 2D solver mmSolve2D. The new
release has basically the same capabilities as the previous release,
but includes updates to maintain compatibility with various operating
systems and the Tcl/Tk programming language. Support for the Mac
operating system (OS X) is included for the first time.
Mike Donahue described an upcoming release in the 1.2 branch of
OOMMF. This distribution will include all updates in the new 1.1
release, but it is centered on the 3D solver, Oxsii. The new release
incorporates a number of bug fixes and several important new
features, including a Runge-Kutta solver for integration of the LLG
equations of motion and a problem restart capability.
For users of the 2D mmSolve2D program, Mike described an input file
translation utility that translates input files to the format
required for 3D Oxsii. Mike presented some examples of the 3D input
syntax and showed how to use the Tcl scripting features for flexible
input. He also described what is involved in creating an extension
module for Oxsii through an example hexagonal anisotropy class.
Mike discussed planned features for the future 1.3 branch of OOMMF,
and announced plans to open a contributor website to facilitate user
contributions to the code.
Open Discussion: Toward a thermal
standard problem
The final item was a discussion geared toward designing a new
micromagnetic standard problem to test thermal effects in
micromagnetics. As with previous successful standard problems, the
new standard problem should have the following properties.
- solvable with a reasonable amount of effort
- ability to distinguish good technique from bad
- solvable by a wide number of techniques.
- availability of analytical results are especially desirable
The lively discussion centered primarily around two main issues:
whether the Langevin approach, where a fluctuating, random applied field
is used to simulate the effects of temperature, is thermodynamicaly
rigorous and what kinds of standard problems are best suited as tests
for thermodynamic rigor.
A list of suggested problems was generated, and the ideas fell
basically into two categories: steady state problems and thermal
switching.
Steady state problems
- Calculation of M vs. T for a block of material
- Temperature dependence of noise, perhaps including power spectra
- Thermally induced vortex motion
- Superparamagnetism of a particle (M vs. T)
- Checking energy statistics against Boltzmann distribution
Switching problems
- Switching times of a particle large enough to have many internal
degrees of freedom.
- Sweep rate dependence of coercivity.
Other comments included a suggestion to create a situation where the
magnetostatic part of the calculation could be simplified, either by
eliminating magnetostatics or by using high anisotropy thin films where
the magnetostatics is relatively unimportant and magnetostatic effects
can be restricted to short range. The audience was reminded that
results should be demonstrated to be independent of discretization.
To wrap up the discussion, it was decided that this report would be
written up and disseminated via the µMAG mailing list where the
discussion could continue online.
Report submitted by Bob
McMichael, Jan 22, 2004.