YY_MEL: OOMMF Magnetoelastic Extension Module
Release ver. 1.0.1
This file was last modified on 2015-03-03.
What's this:
YY_MEL module is an OOMMF extension to include the magnetoelastic copuling in
the micromagnetic simulation. The code is based on the work developed for
Yahagi et al, Phys. Rev. B, 90, 140405(R) (2014).
Basic principle:
The magnetoelastic (MEL) energy can be expressed in terms of the
magnetoelastic coefficients B1 and B2, as a function of the magnetization and
elastic strain. By taking the variational derivative of the MEL energy
density, the MEL contribution to the effective field is obtained. This module
takes either displacement or elastic strain, calculates MEL field and include
it as an additional energy term in the micromagnetic simulation.
Installation and usage:
Copy .cc and .h files into the directory $(OOMMF_DIR)/app/oxs/local/ and build
OOMMF by typing
tclsh oommf.tcl pimake
Include the following Oxs_Ext classes in your mif file and run it.
Oxs_Ext classes:
This extension contains three Oxs_Ext classes to include magnetoelastic (MEL)
energy terms in OOMMF simulations. The classes YY_FixedMEL and YY_StageMEL are
designed in the semblance of Oxs_FixedZeeman and Oxs_StageZeeman to account
for spatially nonuniform, non-time-varying strain. YY_TransformStageMEL
produces time-varying MEL field by applying a spatially uniform linear
transform, much in the same way as Oxs_TransformZeeman does. Unlike
Oxs_TransformZeeman which reads the input vector field only once at the
beginning of the simulation, YY_TransformStageMEL updates the untransformed
vector field (either displacement or strain) at each stage and provides more
flexibility.
Common features:
In all three YY_*MEL classes, some initialization strings are always required.
B1 and B2 take Oxs_ScalarField objects to specify the MEL coefficients in the
unit of J/m^3. In most cases, Oxs_AtlasScalarField will suffice. In
elastically isotropic media, B1 = B2 but you need to specify both of them
anyway.
The elastic strain needs to be specified, either directly with a strain matrix
or indirectly with a displacement vector field. If the displacement is
specified, the strain is automatically calculated and is used for the
subsequent computation. Only one of them can be specified in each Specify
block. Note that specifying the displacement may deteriorate the accuracy when
you model a thin structure with only a few cells across its thickness (e.g.,
multilayers), since YY_MEL extension considers the region with Ms = 0 as
vacuum and calculates one-side difference for the edge cells, as opposed to
the central difference for the inner cells. In these cases, the strain should
be directly specified instead.
The displacement u and strain e are handled as Oxs_VectorField objects. The
strain, however, is a symmetric real matrix and requires six elements to be
stored. In YY_*MEL modules, the diagonal and off-diagonal elements of the
strain are expressed with two Oxs_VectorField objects, together representing
a strain matrix. The indices of these vectors correspond to the matrix index
as depicted below.
Strain matrix Diag vec Off-diag vec
/ 11 12 13 \ / 0 \ / 2 1 \
| .. 22 23 | = | 1 | + | 0 |
\ .. .. 33 / \ 2 / \ /
YY_FixedMEL:
Specify YY_FixedMEL {
B1 scalar_field_spec
B2 scalar_field_spec
u_field vector_field_spec
e_diag_field vector_field_spec
e_offdiag_field vector_field_spec
multiplier number
}
YY_FixedMEL takes a fixed displacement or strain and calculates the MEL field
and energy. Note that even though the strain remains constant throughout the
simulation, the MEL field and energy are repeatedly updated reflecting the
magnetization evolution. The calculated MEL field and energy are multiplied by
the value of the multiplier parameter.
Either one of the displacement (u_field) or the strain (BOTH e_diag_field AND
e_offdiag_field) should be specified.
Example 1.
Specify Oxs_UniformScalarField:B {
value 7.85e6
}
Specify YY_FixedMEL:MELField {
B1 :B
B2 :B
u_field :fileufield
}
YY_StageMEL:
Specify YY_FixedMEL {
B1 scalar_field_spec
B2 scalar_field_spec
u_script Tcl_script
u_files { list_of_files }
e_diag_script Tcl_script
e_diag_files { list_of_files }
e_offdiag_script Tcl_script
e_offdiag_files { list_of_files }
multiplier number
stage_count number_of_stages
}
For B1, B2, multiplier, see YY_FixedMEL. The stage_count works as it does in
Oxs_StageZeeman and notifies the Oxs_Driver how many stages the YY_StageMEL
object wants. The default is 0, indicating no preference.
The displacement or the strain is specified as script, which are evaluated at
each stage and returns an Oxs_VectorField object. See the section for
Oxs_StageZeeman of the OOMMF documentation for an example. Alternatively, a
list of ovf files may be specified as well.
Either one of the displacement (u_script or u_files) or the strain
(e_*diag_script or e_*diag_files) should be specified. For strain input, both
the diagonal and off-diagonal entries need to be specified.
YY_TransformStageMEL:
Specify YY_TransformStageMEL {
B1 scalar_field_spec
B2 scalar_field_spec
u_script Tcl_script
u_files { list_of_files }
e_diag_script Tcl_script
e_diag_files { list_of_files }
e_offdiag_script Tcl_script
e_offdiag_files { list_of_files }
multiplier number
stage_count number_of_stages
type transform_type
script_args { args_request }
script Tcl_script
}
YY_TransformStageMEL is an extension of YY_StageMEL and has all the parameters
the latter has. See YY_StageMEL for descriptions for these parameters.
YY_TransformStageMEL applies a linear transformation to the displacement or
the strain, whichever is specified by the user, and the MEL field and energy
is calculated at each simulation step.
The rest of the parameters, type, script_args, and script behave as they are
defined in Oxs_TransformZeeman. See the OOMMF documentation for the detailed
description. Note, however, that the transform matrix is applied differently
to the displacement and the strain. When M denotes the transformation matrix,
the displacement is transformed in the same manner as the applied field is
transformed in Oxs_TransformZeeman,
u_new = M*u_old
while the strain is transformed as
e_new = M*e_old*transpose(M)
The type of the elastic input determines the type of linaer transformation.
License:
This software is released under the MIT license. See LICENSE.txt.
Contact:
<yuyahagi2@gmail.com>