Skip to content

Latest commit

 

History

History
158 lines (117 loc) · 8.71 KB

CHANGELOG.md

File metadata and controls

158 lines (117 loc) · 8.71 KB
Raw

v4.0.0

New Features

  • Add a function to output Mantid MDHistogramWorkspaces (sw_spec2MDHisto)
  • Add Python plotting of magnetic structure using vispy
  • Add mex files to compute main loop in spinwave() enabling a 2x - 4x speed up depending on system size
  • Add python wrapper for all Matlab functions so e.g. sw_plotspec etc can be called without the m. prefix in pyspinw:q

Improvements

  • Replace spinwavefast() method with a new fastmode option in the main spinwave() method to reduce confusion
  • Adds a neutron_output option to spinwave() to compute only the neutron (Sperp) cross-section and not the full spin-spin correlation tensor, saving memory

Bug Fixes

  • Corrects equation for Q-range from thetaMin in sw_instrument and add thetaMax option
  • Fixes sw_issymspec to recognise powder spectra
  • Fixes a parsing error in the spinw.fourier method if no sublat option given.
  • Fixes several bugs in sw_plotspec where it ignores user options in 'auto' mode, and where it inverts user supplied colormaps.
  • Fixes several bugs in .fitspec() for handling twins and where it only outputs the final chi^2 values.

v3.2.0

Initial public beta of PySpinW

This is an initial public beta version of PySpinW released on PyPI.

Please install using:

pip install spinw

This will install a module called pyspinw (note the py at the start).

You can then run SpinW with:

import numpy as np
import matplotlib.pyplot as plt
from pyspinw import Matlab
m = Matlab()
swobj = m.spinw()
swobj.genlattice('lat_const', [3, 3, 6], 'angled', [90, 90, 120], 'sym', 'P 1');
swobj.addatom('r', [0, 0, 0], 'S', 1/2, 'label', 'MCu2')
swobj.gencoupling('maxDistance', 5)
swobj.addmatrix('label', 'J1', 'value', 1.00, 'color', 'g')
swobj.addcoupling('mat', 'J1', 'bond', 1)
swobj.genmagstr('mode', 'helical', 'k', [-1/3, -1/3, 0], 'n',[0, 0, 1], 'unit', 'lu', 'S', [[1], [0], [0]])
spec = swobj.spinwave([[-1/2, 0, 0], [0, 0, 0], [1/2, 1/2, 0], 100], 'hermit', False)
spec = m.sw_egrid(spec, 'component', 'Sxx+Syy', 'imagChk', False, 'Evect', np.linspace(0, 3, 100))
ax = plt.imshow(np.real(np.flipud(spec['swConv'])), aspect='auto', vmax=1)
plt.show()

On Windows and Linux systems, as long as you're running PySpinW locally, Matlab plotting commands like m.plot(swobj) will work. This is not the case on MacOS (a known bug) and on remote systems (e.g. via JupyterHub).

v0.0.1

pySpinW

This is an initial release of pySpinW as a pip installable wheel for python >= 3.8 and MATLAB >= R2021a

Installation

Please install with

pip install pyspinw*.whl

This package can now be used in python if you have a version of MATLAB or MCR available on the machine. The package will try to automatically detect your installation, however if it is in a non-standard location, the path and version will have to be specified.

from pyspinw import Matlab
m = Matlab(matlab_version='R2023a', matlab_path='/usr/local/MATLAB/R2023a/')

Example

An example would be:

import numpy as np
from pyspinw import Matlab

m = Matlab()

# Create a spinw model, in this case a triangular antiferromagnet
s = m.sw_model('triAF', 1)

# Specify the start and end points of the q grid and the number of points
q_start = [0, 0, 0]
q_end = [1, 1, 0]
pts = 501

# Calculate the spin wave spectrum
spec = m.spinwave(s, [q_start, q_end, pts])

Known limitations

At the moment graphics will not work on macOS systems and is disabled.

v3.1.2

Improvements

  • Change to preallocation of output energies in spinwave to reduce memory usage and improve calculation speed
  • Use a mex function (if mex is enabled) for matrix multiplication in spinwave with hermit=false that reduces memory usage and improves calculation speed for large magnetic cells (in an example with 216 magnetic atoms the execution time was reduced by ~65%)

Bug Fixes

  • Fix generation of lattice from basis vectors in genlattice, see issue #28
  • sortMode in spinwave now correctly sorts the spin wave modes within each twin
  • A spinw object can now be correctly created from a structure figure
  • .cif files with a mixture of tabs and spaces or containing a ? in the comments can now be read correctly
  • Rotation matrix rotC in addtwin is now required to be a valid rotation or reflection matrix.
  • Spin of atom in addatom must have S>=0.
  • Anisotropic g-tensor in addg must be physically valid - i.e. :math:g^\dagger.g must be a symmetric positive definite matrix.
  • Fix bug in addcoupling that did not allow user to supply 'atom' with numeric array of atom indices (previously only worked for string or cell of strings corresponding to atom labels).
  • Renamed undocumented gencoupling parameter tol to tolMaxDist (see doc string of gencoupling for more details).
  • Added validation to gencoupling to ensure maxDistance > dMin.
  • Fixed uncaught error in gencoupling by checking if any bonds have length < maxSym
  • A warning will now be emitted if saveSabp is requested in spinwave for a commensurate structure
  • Fix bug in definition of rotation matrix transforming to spinw coordinate system when left-handed set of basis vectors supplied to genlattice, see issue #57
  • Validation added for perm and origin arguments supplied to genlattice (and warn users that these will be ignored if no symmetry/spacegroup is supplied in the same function call).
  • Deprecated spgr argument to genlattice (users should use sym instead).
  • Fix MATLAB:nonLogicalConditional error raised when using multiple k in genmagstr with helical mode
  • Raise error if invalid shape S or k is provided to genmagstr, previously they would be silently set to zero
  • Raise error if wrong number of spins S is provided to genmagstr in helical mode. Previously the structure would be silently initialised to a random structure.
  • Raise error if a complex spin S is provided to genmagstr in helical mode. Previously this meant it would silently ignore the n option, and behave exactly like fourier mode.
  • Raise error if rotate mode is used without first initialising a magnetic structure
  • Emit deprecation warning if the undocumented extend mode is used in genmagstr
  • Raise error if the first spin is parallel to n and no rotation angle is provided in rotate mode in genmagstr. Previously this would silently result in NaN
  • Raise error if phi or phid is not real in rotate mode in genmagstr. This was an undocumented feature which has been removed.
  • Emit warning that the spin amplitude will be moderated if components of S are parallel to n in helical mode in genmagstr
  • Emit warning if nExt is unnecessarily large compared to k in helical and fourier modes in genmagstr
  • Emit warning if arguments that will be ignored are passed to a particular mode in genmagstr (e.g. S is passed to random)
  • Raise error if complex values is provided for n in genmagstr. Previously this would've caused a crash.
  • Fix error when plotting progress of optmagsteep without existing figure
  • Correctly report magnetic moments in each iteration of optmagsteep.
  • Fix errors when calling intmatrix with dipolar bonds and symbolic spinw object with fitmode true and false
  • Ensure biquadratic exchange interactions are isotropic in addcoupling (previously checked in intmatrix)
  • Raise error if invalid shape kbase is provided to optmagk, previously it would be silently set to empty
  • Ensure varargin is correctly passed through to ndbase.pso from optmagk. Previously user provided TolFun, TolX and MaxIter would be overwritten by the defaults.
  • Warn users that that the results of spinwave have not been scientifically validated for supercell structures with an incommensurate modulation.
  • Emit warning if wrong length xmin, xmax or x0 is passed to optmagstr. Previously they would be silently ignored.
  • No longer require a magnetic structure be initialised with genmagstr before using optmagstr. If not intialised, a default nExt of [1 1 1] is used. This has also been clarified in the docstring.
  • Fix bug where powder spectra was not recognised in sw_plotspec, introduced by a previous update to provide more helpful error messages.
  • sw_instrument now calculates the limits for thetaMax, before it was using the continuation of the thetaMin line to high Q which is incorrect.
  • Fixes a parsing error in the spinw.fourier method if no sublat option given.