libatoms / matscipy Goto Github PK

I just encountered behavior from neighbour_list, which I had not expected and which seems strange / undesirable to me when I call the function and give a single number for the cutoff

• When I use a float, e.g. 5.0, everything works just fine.
• When I give an integer, e.g. 5, I get an error ValueError: object of too small depth for desired array from _matscipy.neighbour_list.

I realize that the documentation states that you can give "A single float" specifically, but I don't see an obvious reason why "5" should not be interpreted as "5.0".

The following tests fail in CI:

• FAILED test_c2d.py::c2dCliTest::test_c2d_input_format_npz_output_format_xyz
• FAILED test_c2d.py::c2dCliTest::test_c2d_input_format_txt_output_format_xyz
• FAILED test_c2d.py::c2dCliTest::test_pnp_c2d_pipeline_mode

Dear authors,

Generally speaking, the software runs more or less as described in the documentation. However, there are some comments in this regard:

• It's not clear where the user should start using the package after installing it. The way the provided examples are placed is a bit difficult to follow. Except for the electrochemistry and the fracture_mechanics directories, the rest of the examples directories are not clear regarding which functionality they belong to. I suggest that you add a new example directory for getting started with the package, putting some basic examples grouped per feature.
• It's not clear how the distributed computation example is supposed to work.
• Generally speaking, the example directory contains a large number of non-trivial files, most of which are not documented unfortunately. I think this makes it difficult to get started with the package.
• Running the samples_pb_c2d.py and samples_pnp_c2d.py examples results in the following error: ModuleNotFoundError: No module named 'matscipy.electrochemistry.utility'. The error for samples_pnp_by_fenics_fem.py is related to lack of matscipy.electrochemistry.poisson_nernst_planck_solver_fenics module.
• Some examples rely on IPython, which doesn't get installed via the installation procedure of the package.

P.S. This issue is related to openjournals/joss-reviews#5668

There are symlinks in matscipy/cli/electrochemistry that make trouble when checking out on non-Linux system.

@jotelha - please remove those symlinks

The current release fails with latest numpy and ASE, so we're overdue for a new one. Does anyone have anything they'd like to merge before we go ahead and do this? I think all the new wheels are working, so will be nice to get them on PyPI too.

On Windows, when I execute "python setup.py build", it says

c = os.path.basename(self.compiler.compiler[0])
AttributeError: 'MSVCCompiler' object has no attribute 'compiler'. Did you mean: 'compile'?

What should I do? Thanks

Should we switch the C code to C++ and pybind11? My guess is this would lead to a C++ code that is half the size of what we have currently, but we would have a dependency on pybind11 (which is actually just a bunch of C++ headers).

The electrochemistry module samples discrete coordinate sets from classical continuum solutions to Poisson-Nernst-Planck systems. It consists of three components,

• a 1D solver for the static Poisson-Nernst-Planck (PNP) equation, either by
  • simple finite differences without any dependencies beyond numpy and scipy (poisson_nernst_planck_solver.py) , or
  • finite elements via fencis ( poisson_nernst_planck_solver_fenics.py) , more robust
• a sampler continuous2discrete.py that samples point coordinates from continuous distributions
• steric_correction.py that uses the algorithm of packmol, described in L. Martinez, R. Andrade, E. G. Birgin, and J. M. Martínez, “PACKMOL: A package for building initial configurations for molecular dynamics simulations,” J. Comput. Chem., vol. 30, no. 13, pp. 2157–2164, 2009, doi: 10/chm6f7, to assure a minimum distance between point coordinates.

A chain of these components may generate molecular coordinates that approximately resemble ion distributions in classical electrochemical double layer theory and may serve as initial configurations for atomistic simulations.

@jotelha elaborate and cast this into markdown snippet for JOSS submission.

Hi guys,
when converting a Voigt strain to a 3x3 strain and back, I don't get the same results.
I think the diagonal entries in the definition below should be 1.0*e1 and so on instead of using the "+".

def Voigt_6_to_full_3x3_strain(strain_vector):
    """
    Form a 3x3 strain matrix from a 6 component vector in Voigt notation
    """
    e1, e2, e3, e4, e5, e6 = np.transpose(strain_vector)
    return np.transpose([[1.0+e1, 0.5*e6, 0.5*e5],
                         [0.5*e6, 1.0+e2, 0.5*e4],
                         [0.5*e5, 0.5*e4, 1.0+e3]])

This is a brilliant project, many thanks for the help in researches.

How about adding a citation guide so that we can cite this repo?

I'm keen to try out the new EAM Hessian functionality - thanks @wgnoehring for contributing this!

I'd like to apply to a system which is large in 2 dimension but narrow in the 3rd dimension, and I see the comment here requires i-j pairs to be unique.

Is there a fundamental reason why this is restricted to systems larger than cutoff in all dimensions? Couldn't we label matrix element by (i, j, shift) rather than just (i, j)?

Is there a reason why the EAM calculator lies about its energy not being force_consistent (in the ASE sense) ?

import ase
import ase.build
from ase.constraints import FixAtoms,StrainFilter,UnitCellFilter
from ase.optimize.precon import PreconLBFGS
from matscipy.calculators.eam import EAM

calc = EAM("PotentialB.fs", kind="eam/fs") 
Fe = ase.build.bulk("Fe", "bcc", a=2.8, cubic=True)
Fe.set_calculator(calc)
opt = PreconLBFGS(UnitCellFilter(Fe, mask=[1,1,1,0,0,0]))
opt.run()

errors with

PropertyNotImplementedError: Force consistent/free energy ("free_energy") not provided by EAM calculator

PotentialB.fs.zip

I'm trying to set up a github CI that uses matscipy, and attempting to install using from the github repo fails, e.g. https://github.com/libAtoms/pymatnext/actions/runs/4779137831/jobs/8496016926

I'm going to try to reproduce outside the github CI.

There is some problem with the poisson ratio calculations.

Here is an example script (needs quippy for the potential, but other EAMs should work), which first computes the poisson ratio in the cubic bcc unit cell. Then I ask it to compute by pulling in the (0,1,1) direction, and it gives a silly number (0.6), then I create a rotated unit cell which has the y axis along the (0,1,1), and now the poisson ratio function (pulling along y) gives a sensible answer.

I know there are other ways of getting at the poisson ratio, using the elastic_moduli function for example, I tried that and both answers are returned in an unsymmetric nu matrix. someone who understands these things should take a good look at this, I'm pretty sure the poisson_ratio() function is broken.
poisson_ratio_bug.tar.gz

ASE's atoms object has a new celldisp parameters that specifies the origin of the cell. The neighbor list will break is it is present and not (0,0,0).

I get a compile error related to the -std=c++0x option (I think). Full output is below.

This is on OS X 10.10.5, Anaconda Python 2.3.0, andApple LLVM version 6.1.0 (clang-602.0.53) (based on LLVM 3.6.0svn)

Sunrise:matscipy ortner$ CC=gcc CXX=g++ python setup.py build
running build
running config_cc
unifing config_cc, config, build_clib, build_ext, build commands --compiler options
running config_fc
unifing config_fc, config, build_clib, build_ext, build commands --fcompiler options
running build_src
build_src
building extension "_matscipy" sources
build_src: building npy-pkg config files
running build_py
running build_ext
customize UnixCCompiler
customize UnixCCompiler using build_ext
customize UnixCCompiler
customize UnixCCompiler using build_ext
building '_matscipy' extension
compiling C sources
C compiler: gcc -fno-strict-aliasing -I/Users/ortner/anaconda/include -arch x86_64 -DNDEBUG -g -fwrapv -O3 -Wall -Wstrict-prototypes

compile options: '-I/Users/ortner/anaconda/lib/python2.7/site-packages/numpy/core/include -I/Users/ortner/anaconda/include/python2.7 -c'
extra options: '-std=c++0x'
gcc: c/tools.c
error: invalid argument '-std=c++0x' not allowed with 'C/ObjC'
error: invalid argument '-std=c++0x' not allowed with 'C/ObjC'
error: Command "gcc -fno-strict-aliasing -I/Users/ortner/anaconda/include -arch x86_64 -DNDEBUG -g -fwrapv -O3 -Wall -Wstrict-prototypes -I/Users/ortner/anaconda/lib/python2.7/site-packages/numpy/core/include -I/Users/ortner/anaconda/include/python2.7 -c c/tools.c -o build/temp.macosx-10.5-x86_64-2.7/c/tools.o -std=c++0x" failed with exit status 1

Test should run when pytest is executed in the main directory.

See closed PR #122 for previous attempt

Hello,
I was searching for rings in compound AB2, and the only type of rings i need is (A-B)-(A-B)-(...). I modified functions step_colser(), step_away() and find_shortest_distances() with type[j] != type[j], but it say "Distance map and graph do not match". How can i do with it ?
Best regard,
yunai

The current API to compute/get a hessian in matscipy is the following function:

def get_hessian(self,
                atoms: ase.Atoms,
                format: str = 'sparse',
                divide_by_masses: bool = False) -> Union[scipy.sparse.spmatrix, np.ndarray]:
    ...

This has the following disatvantages:

1. interoperability with ASE's property system is limited since there are function parameters. I've worked around that in MatscipyCalculator by defining different properties for different parameter combinations (e.g. "dynamical_matrix" calls with divide_by_masses=True.
2. Not every calculator returns a sparse matrix (e.g. Ewald), which means conversions must sometimes happen
3. The sparse format should be bsr_matrix for all calculators: this makes it harder to change for specific calculators in case it's sub-optimal, and makes us less interoperable with other codes or future calculators that might use a different format, e.g. hierarchical matrices, or matrix-free formulations (other kspace implementations, fast-multipole, etc.)
4. There are two accepted formats: "sparse" and "neighbor-list", which are not interoperable

To solve these issues, I propose a new API:

def get_hessian(self,
                atoms: ase.Atoms) -> scipy.sparse.linalg.LinearOperator:
    ...

The abstract class LinearOperator is designed to abstract away details of how matrix-vector (or matrix-matrix) products are done and are fully composable, i.e. P = A * B and S = A + B can be defined for A and B that have different representation without needing conversion. They are also accepted as parameters to the sparse solve/diagonalization routines in scipy. Most calculators in matscipy would just need to call scipy.sparse.linalg.aslinearoperator() on the sparse matrix that they assemble, which returns a concrete instance of MatrixLinearOperator, from which the matrix representation can be retrieved, if need be (e.g. for tests). This addresses points 2 & 3 above.

For point 1, the divide by masses operation should be done in a separate function. For point 4, after discussing with @griessej , we figured that the neighbor-list format is only really used in the pair and polydisperse calculators for the calculation of the Born constants. It then makes sense to move whatever code in MatscipyCalculator that uses that format to these calculators (and actually have the polydisperse calculator inherit from the pair calculator).

LMK what you think of this change. The main burden of making this change would probably be to change the tests of hessian values.

Travis-CI.org is shutting down imminently and the payment plans on the replacement Travis-CI.com are exorbitant, with no free open source plan any more. I think we need to follow the rest of the open source world to GitHub actions - would that be ok? Happy to try to migrate the travis.yml, as I'm doing the same for QUIP and friends.

It may not be worth fixing, so maybe it's just a "won't fix", but the ASE new FixSymmetry constraint confuses matscipy.elasticity. If the constraint is applied when you start computing elastic constants, the strains that it thinks it's applying don't actually take effect, and you get garbage elastic constants as a result. One consistent alternative is for matscipy to have a keep_symmetry argument, which applies a re-computed symmetry constraint after each strain but before the relaxation (so strains can break the symmetry but the relaxations cannot break it further).

Consider a dimer:

from ase import Atoms
from ase.units import eV, Ang, GPa
from matscipy.neighbours import neighbour_list

d = 2.5*Ang
a = Atoms('2Cu', positions=[(0., 0., 0.), (0., 0., d)])
ij = neighbour_list('ij', a, 3.0)

The ase.Atoms constructor correctly sets pbc=False for the zero-length vectors.
The current implementation fails with a LinAlgError when trying to invert the cell.

Traceback (most recent call last):
  File "neigh_pbc_test.py", line 7, in <module>
    ij = neighbour_list('ij', a, 2.0)
  File "/Users/jameskermode/.julia/v0.5/Conda/deps/usr/lib/python2.7/site-packages/matscipy/neighbours.py", line 127, in neighbour_list
    np.linalg.inv(a.cell.T), a.pbc,
  File "/Users/jameskermode/.julia/v0.5/Conda/deps/usr/lib/python2.7/site-packages/numpy/linalg/linalg.py", line 513, in inv
    ainv = _umath_linalg.inv(a, signature=signature, extobj=extobj)
  File "/Users/jameskermode/.julia/v0.5/Conda/deps/usr/lib/python2.7/site-packages/numpy/linalg/linalg.py", line 90, in _raise_linalgerror_singular
    raise LinAlgError("Singular matrix")
numpy.linalg.linalg.LinAlgError: Singular matrix

Since the inverse shouldn't be needed when pbc=False, I tried the following patch:

index 30096ba..c8567de 100644
--- a/matscipy/neighbours.py
+++ b/matscipy/neighbours.py
@@ -123,7 +123,12 @@ def neighbour_list(quantities, a, cutoff, *args):
                    np.bincount(i, weights=pair_forces[:, 2], minlength=len(a))
     """

+    if np.any(a.pbc):
+        invcell = np.linalg.inv(a.cell.T)
+    else:
+        invcell = np.eye(3)
+
     return _matscipy.neighbour_list(quantities, a.cell,
-                                    np.linalg.inv(a.cell.T), a.pbc,
+                                    invcell, a.pbc,
                                     a.positions, cutoff, *args)

However, we need to think about this a bit more, as this leads to zero cell volume, and would screw up the division of box into cells within C neighbour list code.

$ python neigh_pbc_test.py
Traceback (most recent call last):
  File "neigh_pbc_test.py", line 7, in <module>
    ij = neighbour_list('ij', a, 2.0)
  File "/Users/jameskermode/.julia/v0.5/Conda/deps/usr/lib/python2.7/site-packages/matscipy/neighbours.py", line 133, in neighbour_list
    a.positions, cutoff, *args)
RuntimeError: Zero cell volume.

when I pip install matscipy with windows10 VS2019,errors occurred

ERROR: Command errored out with exit status 1:
command: 'D:\program\Anaconda3\envs\tmp\python.exe' -u -c 'import sys, setuptools, tokenize; sys.argv[0] = '"'"'C:\Users\ADMINI1\AppData\Local\Temp\pip-install-8ief1gm9\matscipy\setup.py'"'"'; file='"'"'C:\Users\ADMINI1\AppData\Local\Temp\pip-install-8ief1gm9\matscipy\setup.py'"'"';f=getattr(tokenize, '"'"'open'"'"', open)(file);code=f.read().replace('"'"'\r\n'"'"', '"'"'\n'"'"');f.close();exec(compile(code, file, '"'"'exec'"'"'))' bdist_wheel -d 'C:\Users\ADMINI1\AppData\Local\Temp\pip-wheel-hhahycjt'
cwd: C:\Users\ADMINI1\AppData\Local\Temp\pip-install-8ief1gm9\matscipy
Complete output (142 lines):
running bdist_wheel
running build
running config_cc
unifing config_cc, config, build_clib, build_ext, build commands --compiler options
running config_fc
unifing config_fc, config, build_clib, build_ext, build commands --fcompiler options
running build_src
build_src
building extension "matscipy" sources
build_src: building npy-pkg config files
running build_py
creating build
creating build\lib.win-amd64-3.7
creating build\lib.win-amd64-3.7\matscipy
copying matscipy\angle_distribution.py -> build\lib.win-amd64-3.7\matscipy
copying matscipy\atomic_strain.py -> build\lib.win-amd64-3.7\matscipy
copying matscipy\deformation.py -> build\lib.win-amd64-3.7\matscipy
copying matscipy\dislocation.py -> build\lib.win-amd64-3.7\matscipy
copying matscipy\distributed_computation.py -> build\lib.win-amd64-3.7\matscipy
copying matscipy\drift.py -> build\lib.win-amd64-3.7\matscipy
copying matscipy\elasticity.py -> build\lib.win-amd64-3.7\matscipy
copying matscipy\hessian_finite_differences.py -> build\lib.win-amd64-3.7\matscipy
copying matscipy\hydrogenate.py -> build\lib.win-amd64-3.7\matscipy
copying matscipy\logger.py -> build\lib.win-amd64-3.7\matscipy
copying matscipy\neighbours.py -> build\lib.win-amd64-3.7\matscipy
copying matscipy\opls.py -> build\lib.win-amd64-3.7\matscipy
copying matscipy\pressurecoupling.py -> build\lib.win-amd64-3.7\matscipy
copying matscipy\rings.py -> build\lib.win-amd64-3.7\matscipy
copying matscipy\socketcalc.py -> build\lib.win-amd64-3.7\matscipy
copying matscipy\spatial_correlation_function.py -> build\lib.win-amd64-3.7\matscipy
copying matscipy\surface.py -> build\lib.win-amd64-3.7\matscipy
copying matscipy\visualise.py -> build\lib.win-amd64-3.7\matscipy
copying matscipy_version.py -> build\lib.win-amd64-3.7\matscipy
copying matscipy_init.py -> build\lib.win-amd64-3.7\matscipy
creating build\lib.win-amd64-3.7\matscipy\fracture_mechanics
copying matscipy\fracture_mechanics\clusters.py -> build\lib.win-amd64-3.7\matscipy\fracture_mechanics
copying matscipy\fracture_mechanics\crack.py -> build\lib.win-amd64-3.7\matscipy\fracture_mechanics
copying matscipy\fracture_mechanics\crackpathsel.py -> build\lib.win-amd64-3.7\matscipy\fracture_mechanics
copying matscipy\fracture_mechanics\energy_release.py -> build\lib.win-amd64-3.7\matscipy\fracture_mechanics
copying matscipy\fracture_mechanics\idealbrittlesolid.py -> build\lib.win-amd64-3.7\matscipy\fracture_mechanics
copying matscipy\fracture_mechanics_init_.py -> build\lib.win-amd64-3.7\matscipy\fracture_mechanics
creating build\lib.win-amd64-3.7\matscipy\contact_mechanics
copying matscipy\contact_mechanics\analysis.py -> build\lib.win-amd64-3.7\matscipy\contact_mechanics
copying matscipy\contact_mechanics\DMT.py -> build\lib.win-amd64-3.7\matscipy\contact_mechanics
copying matscipy\contact_mechanics\greens_function.py -> build\lib.win-amd64-3.7\matscipy\contact_mechanics
copying matscipy\contact_mechanics\Hertz.py -> build\lib.win-amd64-3.7\matscipy\contact_mechanics
copying matscipy\contact_mechanics\JKR.py -> build\lib.win-amd64-3.7\matscipy\contact_mechanics
copying matscipy\contact_mechanics_init_.py -> build\lib.win-amd64-3.7\matscipy\contact_mechanics
creating build\lib.win-amd64-3.7\matscipy\electrochemistry
copying matscipy\electrochemistry\continuous2discrete.py -> build\lib.win-amd64-3.7\matscipy\electrochemistry
copying matscipy\electrochemistry\poisson_boltzmann_distribution.py -> build\lib.win-amd64-3.7\matscipy\electrochemistry
copying matscipy\electrochemistry\poisson_nernst_planck_solver.py -> build\lib.win-amd64-3.7\matscipy\electrochemistry
copying matscipy\electrochemistry_init_.py -> build\lib.win-amd64-3.7\matscipy\electrochemistry
creating build\lib.win-amd64-3.7\matscipy\calculators
copying matscipy\calculators\fitting.py -> build\lib.win-amd64-3.7\matscipy\calculators
copying matscipy\calculators\supercell_calculator.py -> build\lib.win-amd64-3.7\matscipy\calculators
copying matscipy\calculators_init_.py -> build\lib.win-amd64-3.7\matscipy\calculators
creating build\lib.win-amd64-3.7\matscipy\calculators\eam
copying matscipy\calculators\eam\average_atom.py -> build\lib.win-amd64-3.7\matscipy\calculators\eam
copying matscipy\calculators\eam\calculator.py -> build\lib.win-amd64-3.7\matscipy\calculators\eam
copying matscipy\calculators\eam\io.py -> build\lib.win-amd64-3.7\matscipy\calculators\eam
copying matscipy\calculators\eam_init_.py -> build\lib.win-amd64-3.7\matscipy\calculators\eam
creating build\lib.win-amd64-3.7\matscipy\io
copying matscipy\io\metis.py -> build\lib.win-amd64-3.7\matscipy\io
copying matscipy\io\opls.py -> build\lib.win-amd64-3.7\matscipy\io
copying matscipy\io\tbl.py -> build\lib.win-amd64-3.7\matscipy\io
copying matscipy\io_init_.py -> build\lib.win-amd64-3.7\matscipy\io
creating build\lib.win-amd64-3.7\matscipy\tool
copying matscipy\tool\plot.py -> build\lib.win-amd64-3.7\matscipy\tool
copying matscipy\tool_init_.py -> build\lib.win-amd64-3.7\matscipy\tool
creating build\lib.win-amd64-3.7\matscipy\calculators\pair_potential
copying matscipy\calculators\pair_potential\calculator.py -> build\lib.win-amd64-3.7\matscipy\calculators\pair_potential
copying matscipy\calculators\pair_potential_init_.py -> build\lib.win-amd64-3.7\matscipy\calculators\pair_potential
creating build\lib.win-amd64-3.7\matscipy\calculators\polydisperse
copying matscipy\calculators\polydisperse\calculator.py -> build\lib.win-amd64-3.7\matscipy\calculators\polydisperse
copying matscipy\calculators\polydisperse_init_.py -> build\lib.win-amd64-3.7\matscipy\calculators\polydisperse
creating build\lib.win-amd64-3.7\matscipy\calculators\mcfm
copying matscipy\calculators\mcfm\calculator.py -> build\lib.win-amd64-3.7\matscipy\calculators\mcfm
copying matscipy\calculators\mcfm\cluster_data.py -> build\lib.win-amd64-3.7\matscipy\calculators\mcfm
copying matscipy\calculators\mcfm\qm_cluster.py -> build\lib.win-amd64-3.7\matscipy\calculators\mcfm
copying matscipy\calculators\mcfm_init_.py -> build\lib.win-amd64-3.7\matscipy\calculators\mcfm
creating build\lib.win-amd64-3.7\matscipy\calculators\mcfm\mcfm_parallel
copying matscipy\calculators\mcfm\mcfm_parallel\mcfm_parallel_control.py -> build\lib.win-amd64-3.7\matscipy\calculators\mcfm\mcfm_parallel
copying matscipy\calculators\mcfm\mcfm_parallel\mcfm_parallel_worker.py -> build\lib.win-amd64-3.7\matscipy\calculators\mcfm\mcfm_parallel
copying matscipy\calculators\mcfm\mcfm_parallel_init_.py -> build\lib.win-amd64-3.7\matscipy\calculators\mcfm\mcfm_parallel
creating build\lib.win-amd64-3.7\matscipy\calculators\mcfm\neighbour_list_mcfm
copying matscipy\calculators\mcfm\neighbour_list_mcfm\neighbour_list_base.py -> build\lib.win-amd64-3.7\matscipy\calculators\mcfm\neighbour_list_mcfm
copying matscipy\calculators\mcfm\neighbour_list_mcfm\neighbour_list_mcfm.py -> build\lib.win-amd64-3.7\matscipy\calculators\mcfm\neighbour_list_mcfm
copying matscipy\calculators\mcfm\neighbour_list_mcfm_init_.py -> build\lib.win-amd64-3.7\matscipy\calculators\mcfm\neighbour_list_mcfm
creating build\lib.win-amd64-3.7\matscipy\calculators\mcfm\qm_cluster_tools
copying matscipy\calculators\mcfm\qm_cluster_tools\base_qm_cluster_tool.py -> build\lib.win-amd64-3.7\matscipy\calculators\mcfm\qm_cluster_tools
copying matscipy\calculators\mcfm\qm_cluster_tools\qm_clustering_tool.py -> build\lib.win-amd64-3.7\matscipy\calculators\mcfm\qm_cluster_tools
copying matscipy\calculators\mcfm\qm_cluster_tools\qm_flagging_tool.py -> build\lib.win-amd64-3.7\matscipy\calculators\mcfm\qm_cluster_tools
copying matscipy\calculators\mcfm\qm_cluster_tools_init_.py -> build\lib.win-amd64-3.7\matscipy\calculators\mcfm\qm_cluster_tools
UPDATING build\lib.win-amd64-3.7\matscipy/_version.py
set build\lib.win-amd64-3.7\matscipy/_version.py to '0.6.0'
running build_ext
No module named 'numpy.distutils._msvccompiler' in numpy.distutils; trying from distutils
customize MSVCCompiler
customize MSVCCompiler using build_ext_subclass
No module named 'numpy.distutils._msvccompiler' in numpy.distutils; trying from distutils
customize MSVCCompiler
Missing compiler_cxx fix for MSVCCompiler
customize MSVCCompiler using build_ext_subclass
building 'matscipy' extension
Traceback (most recent call last):
File "", line 1, in
File "C:\Users\ADMINI~1\AppData\Local\Temp\pip-install-8ief1gm9\matscipy\setup.py", line 281, in
url="https://github.com/libAtoms/matscipy"
File "D:\program\Anaconda3\envs\tmp\lib\site-packages\numpy\distutils\core.py", line 171, in setup
return old_setup(**new_attr)
File "D:\program\Anaconda3\envs\tmp\lib\site-packages\setuptools_init.py", line 153, in setup
return distutils.core.setup(**attrs)
File "D:\program\Anaconda3\envs\tmp\lib\distutils\core.py", line 148, in setup
dist.run_commands()
File "D:\program\Anaconda3\envs\tmp\lib\distutils\dist.py", line 966, in run_commands
self.run_command(cmd)
File "D:\program\Anaconda3\envs\tmp\lib\distutils\dist.py", line 985, in run_command
cmd_obj.run()
File "D:\program\Anaconda3\envs\tmp\lib\site-packages\wheel\bdist_wheel.py", line 290, in run
self.run_command('build')
File "D:\program\Anaconda3\envs\tmp\lib\distutils\cmd.py", line 313, in run_command
self.distribution.run_command(command)
File "D:\program\Anaconda3\envs\tmp\lib\distutils\dist.py", line 985, in run_command
cmd_obj.run()
File "D:\program\Anaconda3\envs\tmp\lib\site-packages\numpy\distutils\command\build.py", line 47, in run
old_build.run(self)
File "D:\program\Anaconda3\envs\tmp\lib\distutils\command\build.py", line 135, in run
self.run_command(cmd_name)
File "D:\program\Anaconda3\envs\tmp\lib\distutils\cmd.py", line 313, in run_command
self.distribution.run_command(command)
File "D:\program\Anaconda3\envs\tmp\lib\distutils\dist.py", line 985, in run_command
cmd_obj.run()
File "D:\program\Anaconda3\envs\tmp\lib\site-packages\numpy\distutils\command\build_ext.py", line 261, in run
self.build_extensions()
File "D:\program\Anaconda3\envs\tmp\lib\distutils\command\build_ext.py", line 449, in build_extensions
self._build_extensions_serial()
File "D:\program\Anaconda3\envs\tmp\lib\distutils\command\build_ext.py", line 474, in _build_extensions_serial
self.build_extension(ext)
File "C:\Users\ADMINI~1\AppData\Local\Temp\pip-install-8ief1gm9\matscipy\setup.py", line 81, in build_extension
c = os.path.basename(self.compiler.compiler[0])
AttributeError: 'MSVCCompiler' object has no attribute 'compiler'

ERROR: Failed building wheel for matscipy
Running setup.py clean for matscipy
Failed to build matscipy
Installing collected packages: matscipy
Running setup.py install for matscipy ... error
ERROR: Command errored out with exit status 1:
command: 'D:\program\Anaconda3\envs\tmp\python.exe' -u -c 'import sys, setuptools, tokenize; sys.argv[0] = '"'"'C:\Users\ADMINI1\AppData\Local\Temp\pip-install-8ief1gm9\matscipy\setup.py'"'"'; file='"'"'C:\Users\ADMINI1\AppData\Local\Temp\pip-install-8ief1gm9\matscipy\setup.py'"'"';f=getattr(tokenize, '"'"'open'"'"', open)(file);code=f.read().replace('"'"'\r\n'"'"', '"'"'\n'"'"');f.close();exec(compile(code, file, '"'"'exec'"'"'))' install --record 'C:\Users\ADMINI1\AppData\Local\Temp\pip-record-l0hegtj0\install-record.txt' --single-version-externally-managed --compile --install-headers 'D:\program\Anaconda3\envs\tmp\Include\matscipy'
cwd: C:\Users\ADMINI1\AppData\Local\Temp\pip-install-8ief1gm9\matscipy
Complete output (146 lines):
running install
running build
running config_cc
unifing config_cc, config, build_clib, build_ext, build commands --compiler options
running config_fc
unifing config_fc, config, build_clib, build_ext, build commands --fcompiler options
running build_src
build_src
building extension "matscipy" sources
build_src: building npy-pkg config files
running build_py
creating build
creating build\lib.win-amd64-3.7
creating build\lib.win-amd64-3.7\matscipy
copying matscipy\angle_distribution.py -> build\lib.win-amd64-3.7\matscipy
copying matscipy\atomic_strain.py -> build\lib.win-amd64-3.7\matscipy
copying matscipy\deformation.py -> build\lib.win-amd64-3.7\matscipy
copying matscipy\dislocation.py -> build\lib.win-amd64-3.7\matscipy
copying matscipy\distributed_computation.py -> build\lib.win-amd64-3.7\matscipy
copying matscipy\drift.py -> build\lib.win-amd64-3.7\matscipy
copying matscipy\elasticity.py -> build\lib.win-amd64-3.7\matscipy
copying matscipy\hessian_finite_differences.py -> build\lib.win-amd64-3.7\matscipy
copying matscipy\hydrogenate.py -> build\lib.win-amd64-3.7\matscipy
copying matscipy\logger.py -> build\lib.win-amd64-3.7\matscipy
copying matscipy\neighbours.py -> build\lib.win-amd64-3.7\matscipy
copying matscipy\opls.py -> build\lib.win-amd64-3.7\matscipy
copying matscipy\pressurecoupling.py -> build\lib.win-amd64-3.7\matscipy
copying matscipy\rings.py -> build\lib.win-amd64-3.7\matscipy
copying matscipy\socketcalc.py -> build\lib.win-amd64-3.7\matscipy
copying matscipy\spatial_correlation_function.py -> build\lib.win-amd64-3.7\matscipy
copying matscipy\surface.py -> build\lib.win-amd64-3.7\matscipy
copying matscipy\visualise.py -> build\lib.win-amd64-3.7\matscipy
copying matscipy_version.py -> build\lib.win-amd64-3.7\matscipy
copying matscipy_init.py -> build\lib.win-amd64-3.7\matscipy
creating build\lib.win-amd64-3.7\matscipy\fracture_mechanics
copying matscipy\fracture_mechanics\clusters.py -> build\lib.win-amd64-3.7\matscipy\fracture_mechanics
copying matscipy\fracture_mechanics\crack.py -> build\lib.win-amd64-3.7\matscipy\fracture_mechanics
copying matscipy\fracture_mechanics\crackpathsel.py -> build\lib.win-amd64-3.7\matscipy\fracture_mechanics
copying matscipy\fracture_mechanics\energy_release.py -> build\lib.win-amd64-3.7\matscipy\fracture_mechanics
copying matscipy\fracture_mechanics\idealbrittlesolid.py -> build\lib.win-amd64-3.7\matscipy\fracture_mechanics
copying matscipy\fracture_mechanics_init_.py -> build\lib.win-amd64-3.7\matscipy\fracture_mechanics
creating build\lib.win-amd64-3.7\matscipy\contact_mechanics
copying matscipy\contact_mechanics\analysis.py -> build\lib.win-amd64-3.7\matscipy\contact_mechanics
copying matscipy\contact_mechanics\DMT.py -> build\lib.win-amd64-3.7\matscipy\contact_mechanics
copying matscipy\contact_mechanics\greens_function.py -> build\lib.win-amd64-3.7\matscipy\contact_mechanics
copying matscipy\contact_mechanics\Hertz.py -> build\lib.win-amd64-3.7\matscipy\contact_mechanics
copying matscipy\contact_mechanics\JKR.py -> build\lib.win-amd64-3.7\matscipy\contact_mechanics
copying matscipy\contact_mechanics_init_.py -> build\lib.win-amd64-3.7\matscipy\contact_mechanics
creating build\lib.win-amd64-3.7\matscipy\electrochemistry
copying matscipy\electrochemistry\continuous2discrete.py -> build\lib.win-amd64-3.7\matscipy\electrochemistry
copying matscipy\electrochemistry\poisson_boltzmann_distribution.py -> build\lib.win-amd64-3.7\matscipy\electrochemistry
copying matscipy\electrochemistry\poisson_nernst_planck_solver.py -> build\lib.win-amd64-3.7\matscipy\electrochemistry
copying matscipy\electrochemistry_init_.py -> build\lib.win-amd64-3.7\matscipy\electrochemistry
creating build\lib.win-amd64-3.7\matscipy\calculators
copying matscipy\calculators\fitting.py -> build\lib.win-amd64-3.7\matscipy\calculators
copying matscipy\calculators\supercell_calculator.py -> build\lib.win-amd64-3.7\matscipy\calculators
copying matscipy\calculators_init_.py -> build\lib.win-amd64-3.7\matscipy\calculators
creating build\lib.win-amd64-3.7\matscipy\calculators\eam
copying matscipy\calculators\eam\average_atom.py -> build\lib.win-amd64-3.7\matscipy\calculators\eam
copying matscipy\calculators\eam\calculator.py -> build\lib.win-amd64-3.7\matscipy\calculators\eam
copying matscipy\calculators\eam\io.py -> build\lib.win-amd64-3.7\matscipy\calculators\eam
copying matscipy\calculators\eam_init_.py -> build\lib.win-amd64-3.7\matscipy\calculators\eam
creating build\lib.win-amd64-3.7\matscipy\io
copying matscipy\io\metis.py -> build\lib.win-amd64-3.7\matscipy\io
copying matscipy\io\opls.py -> build\lib.win-amd64-3.7\matscipy\io
copying matscipy\io\tbl.py -> build\lib.win-amd64-3.7\matscipy\io
copying matscipy\io_init_.py -> build\lib.win-amd64-3.7\matscipy\io
creating build\lib.win-amd64-3.7\matscipy\tool
copying matscipy\tool\plot.py -> build\lib.win-amd64-3.7\matscipy\tool
copying matscipy\tool_init_.py -> build\lib.win-amd64-3.7\matscipy\tool
creating build\lib.win-amd64-3.7\matscipy\calculators\pair_potential
copying matscipy\calculators\pair_potential\calculator.py -> build\lib.win-amd64-3.7\matscipy\calculators\pair_potential
copying matscipy\calculators\pair_potential_init_.py -> build\lib.win-amd64-3.7\matscipy\calculators\pair_potential
creating build\lib.win-amd64-3.7\matscipy\calculators\polydisperse
copying matscipy\calculators\polydisperse\calculator.py -> build\lib.win-amd64-3.7\matscipy\calculators\polydisperse
copying matscipy\calculators\polydisperse_init_.py -> build\lib.win-amd64-3.7\matscipy\calculators\polydisperse
creating build\lib.win-amd64-3.7\matscipy\calculators\mcfm
copying matscipy\calculators\mcfm\calculator.py -> build\lib.win-amd64-3.7\matscipy\calculators\mcfm
copying matscipy\calculators\mcfm\cluster_data.py -> build\lib.win-amd64-3.7\matscipy\calculators\mcfm
copying matscipy\calculators\mcfm\qm_cluster.py -> build\lib.win-amd64-3.7\matscipy\calculators\mcfm
copying matscipy\calculators\mcfm_init_.py -> build\lib.win-amd64-3.7\matscipy\calculators\mcfm
creating build\lib.win-amd64-3.7\matscipy\calculators\mcfm\mcfm_parallel
copying matscipy\calculators\mcfm\mcfm_parallel\mcfm_parallel_control.py -> build\lib.win-amd64-3.7\matscipy\calculators\mcfm\mcfm_parallel
copying matscipy\calculators\mcfm\mcfm_parallel\mcfm_parallel_worker.py -> build\lib.win-amd64-3.7\matscipy\calculators\mcfm\mcfm_parallel
copying matscipy\calculators\mcfm\mcfm_parallel_init_.py -> build\lib.win-amd64-3.7\matscipy\calculators\mcfm\mcfm_parallel
creating build\lib.win-amd64-3.7\matscipy\calculators\mcfm\neighbour_list_mcfm
copying matscipy\calculators\mcfm\neighbour_list_mcfm\neighbour_list_base.py -> build\lib.win-amd64-3.7\matscipy\calculators\mcfm\neighbour_list_mcfm
copying matscipy\calculators\mcfm\neighbour_list_mcfm\neighbour_list_mcfm.py -> build\lib.win-amd64-3.7\matscipy\calculators\mcfm\neighbour_list_mcfm
copying matscipy\calculators\mcfm\neighbour_list_mcfm_init_.py -> build\lib.win-amd64-3.7\matscipy\calculators\mcfm\neighbour_list_mcfm
creating build\lib.win-amd64-3.7\matscipy\calculators\mcfm\qm_cluster_tools
copying matscipy\calculators\mcfm\qm_cluster_tools\base_qm_cluster_tool.py -> build\lib.win-amd64-3.7\matscipy\calculators\mcfm\qm_cluster_tools
copying matscipy\calculators\mcfm\qm_cluster_tools\qm_clustering_tool.py -> build\lib.win-amd64-3.7\matscipy\calculators\mcfm\qm_cluster_tools
copying matscipy\calculators\mcfm\qm_cluster_tools\qm_flagging_tool.py -> build\lib.win-amd64-3.7\matscipy\calculators\mcfm\qm_cluster_tools
copying matscipy\calculators\mcfm\qm_cluster_tools_init_.py -> build\lib.win-amd64-3.7\matscipy\calculators\mcfm\qm_cluster_tools
UPDATING build\lib.win-amd64-3.7\matscipy/_version.py
set build\lib.win-amd64-3.7\matscipy/_version.py to '0.6.0'
running build_ext
No module named 'numpy.distutils._msvccompiler' in numpy.distutils; trying from distutils
customize MSVCCompiler
customize MSVCCompiler using build_ext_subclass
No module named 'numpy.distutils._msvccompiler' in numpy.distutils; trying from distutils
customize MSVCCompiler
Missing compiler_cxx fix for MSVCCompiler
customize MSVCCompiler using build_ext_subclass
building 'matscipy' extension
Traceback (most recent call last):
File "", line 1, in
File "C:\Users\ADMINI~1\AppData\Local\Temp\pip-install-8ief1gm9\matscipy\setup.py", line 281, in
url="https://github.com/libAtoms/matscipy"
File "D:\program\Anaconda3\envs\tmp\lib\site-packages\numpy\distutils\core.py", line 171, in setup
return old_setup(**new_attr)
File "D:\program\Anaconda3\envs\tmp\lib\site-packages\setuptools_init.py", line 153, in setup
return distutils.core.setup(**attrs)
File "D:\program\Anaconda3\envs\tmp\lib\distutils\core.py", line 148, in setup
dist.run_commands()
File "D:\program\Anaconda3\envs\tmp\lib\distutils\dist.py", line 966, in run_commands
self.run_command(cmd)
File "D:\program\Anaconda3\envs\tmp\lib\distutils\dist.py", line 985, in run_command
cmd_obj.run()
File "D:\program\Anaconda3\envs\tmp\lib\site-packages\numpy\distutils\command\install.py", line 62, in run
r = self.setuptools_run()
File "D:\program\Anaconda3\envs\tmp\lib\site-packages\numpy\distutils\command\install.py", line 36, in setuptools_run
return distutils_install.run(self)
File "D:\program\Anaconda3\envs\tmp\lib\distutils\command\install.py", line 545, in run
self.run_command('build')
File "D:\program\Anaconda3\envs\tmp\lib\distutils\cmd.py", line 313, in run_command
self.distribution.run_command(command)
File "D:\program\Anaconda3\envs\tmp\lib\distutils\dist.py", line 985, in run_command
cmd_obj.run()
File "D:\program\Anaconda3\envs\tmp\lib\site-packages\numpy\distutils\command\build.py", line 47, in run
old_build.run(self)
File "D:\program\Anaconda3\envs\tmp\lib\distutils\command\build.py", line 135, in run
self.run_command(cmd_name)
File "D:\program\Anaconda3\envs\tmp\lib\distutils\cmd.py", line 313, in run_command
self.distribution.run_command(command)
File "D:\program\Anaconda3\envs\tmp\lib\distutils\dist.py", line 985, in run_command
cmd_obj.run()
File "D:\program\Anaconda3\envs\tmp\lib\site-packages\numpy\distutils\command\build_ext.py", line 261, in run
self.build_extensions()
File "D:\program\Anaconda3\envs\tmp\lib\distutils\command\build_ext.py", line 449, in build_extensions
self._build_extensions_serial()
File "D:\program\Anaconda3\envs\tmp\lib\distutils\command\build_ext.py", line 474, in _build_extensions_serial
self.build_extension(ext)
File "C:\Users\ADMINI1\AppData\Local\Temp\pip-install-8ief1gm9\matscipy\setup.py", line 81, in build_extension
c = os.path.basename(self.compiler.compiler[0])
AttributeError: 'MSVCCompiler' object has no attribute 'compiler'
----------------------------------------
ERROR: Command errored out with exit status 1: 'D:\program\Anaconda3\envs\tmp\python.exe' -u -c 'import sys, setuptools, tokenize; sys.argv[0] = '"'"'C:\Users\ADMINI1\AppData\Local\Temp\pip-install-8ief1gm9\matscipy\setup.py'"'"'; file='"'"'C:\Users\ADMINI1\AppData\Local\Temp\pip-install-8ief1gm9\matscipy\setup.py'"'"';f=getattr(tokenize, '"'"'open'"'"', open)(file);code=f.read().replace('"'"'\r\n'"'"', '"'"'\n'"'"');f.close();exec(compile(code, file, '"'"'exec'"'"'))' install --record 'C:\Users\ADMINI1\AppData\Local\Temp\pip-record-l0hegtj0\install-record.txt' --single-version-externally-managed --compile --install-headers 'D:\program\Anaconda3\envs\tmp\Include\matscipy' Check the logs for full command output.

I am a bit confused about the license, the LICENSE file states GPL 2.0 while the setup.py file states LGPL 2.1+.

push!(LOAD_PATH, "..")
using PyCall
using ASE
@pyimport matscipy.neighbours as matscipy_nlist
at = bulk("Al"; cubic=true)
at = repeat(at, (3, 3, 3))
@time NL = matscipy_nlist.neighbour_list("ijd", at.po, 3.1)

this returns neighbours as expected.

Changing to

at = bulk("Al"; cubic=false)

returns an empty neighbourlist.

(Int32[],Int32[],Float64[])

(also, it takes 6.8 second instead of 0.02 seconds)

Very rarely, I get a crash inside matscipy's neighbor list code. The crash happens in neighbours.c, specifically in

PyObject *py_ret = PyTuple_New(strlen(quantities));

I added earlier references to strlen(quantities), and they work until you get to

#if PY_MAJOR_VERSION >= 3
    Py_DECREF(py_bquantities);
#endif

Indeed, quantities is set by what looks to me like it could just be a reference to the incoming python object, not a copy

#if PY_MAJOR_VERSION >= 3
    PyObject *py_bquantities = PyUnicode_AsASCIIString(py_quantities);
    char *quantities = PyBytes_AS_STRING(py_bquantities);
#else
    char *quantities = PyString_AS_STRING(py_quantities);
#endif

If the Py_DECREF makes the pointer invalid (python documentation suggests that it can, from what I can tell), that would explain why accessing later sometimes leads to a seg fault. Should it be copied, or move the PyDECREF to later in the routine, once quantities is no longer needed?

The neighbor list is currently sorted by first index only, it may be beneficial for some applications to have a fully sorted neighbor list. This would make neighbor list construction slower, but only slightly so. Any opinions on this?

pot.fs.zip

I am trying to use an EAM potential for FeH, and have a .fs file. When I pass it to the matscipy calculator, I get the same energy that the standard ASE EAM calculator gives, but the forces are different. Geometry optimisation fails with the matscipy calculator, so I that is the wrong one. Am I doing something wrong?

import matscipy
import ase
from ase.lattice.cubic import BodyCenteredCubic
import numpy as np

calc1 = matscipy.calculators.eam.EAM("pot.fs", kind="eam/fs")   
calc2 = ase.calculators.eam.EAM(potential='pot.fs')  
a = BodyCenteredCubic(directions=[(1,0,0), (0,1,0), (0,0,1)], size=(3,3,3), symbol='Fe', pbc=True, latticeconstant=2.855)
a.append(ase.Atom('H', (0.0,a0*0.25,a0*0.5)))
a.set_calculator(calc1)
e1 = a.get_potential_energy()
f1 = a.get_forces()
a.set_calculator(calc2)
e2 = a.get_potential_energy()
f2 = a.get_forces()
print(e1-e2)
print(np.max(f1-f2))

comes back with

2.842170943040401e-14
0.028209307473157738

if I don't add the hydrogen, the forces are the same (even after a rattle).

Hi, matscipy is a fantastic work, really love it.

But I got a question when generating the strained configuration:

In elasticity.py/generate_strained_configs function:

T = np.eye(3) + Voigt_6_to_full_3x3_strain(strain)
at.set_cell(np.dot(T, at.cell.T).T, scale_atoms=False)
at.positions[:] = np.dot(T, at.positions.T).T
at.info['strain'] = T - np.eye(3)

I wonder why the new cell is generated in that way, is there any underlying theory? np.dot(T, at.cell.T).T seems to be equivalent to a simpler form of np.dot(T, at.cell)

Thank you very much.

Following the recent changes to neighbour list, the following script gives a seg fault in ring_statistics.cpp

from io import StringIO
from ase.io import read
from matscipy.rings import ring_statistics

buf = StringIO("""216
Lattice="13.844138387203303 0.0 0.0 0.0 13.844138387203303 0.0 0.0 0.0 13.844138387203303" Properties=species:S:1:pos:R:3 i=20652 time=10326.0 E=-1034.0492541143 pbc="T T T"
Si       8.76146557      10.35390567       0.11347189
Si      14.21807936      -3.56114904      -0.88593487
Si      13.78922021       5.94338441      11.82531443
Si      12.69713705      13.90634767       2.46041116
Si       1.08192336       8.49992707       4.21266214
Si       0.85104850      15.67425259      -2.57860263
Si       0.90029713      11.07042506       2.95418491
Si       3.85024689       4.76283378      17.86188400
Si      18.40046490       5.15887197       8.66436327
Si      11.26875133       2.12060084      13.49887492
Si       1.58350468       7.26296547      -4.71806042
Si      11.72714598      13.27014891       7.13562216
Si       8.65246048       3.44641145       0.25553552
Si      14.75414206      12.56433673      10.19498937
Si       1.11314818       5.67063297      14.71348905
Si       3.50879963       6.72344126       5.79210860
Si       7.15740861      27.76073598       7.72284615
Si      17.32886564      13.15921380      16.04127017
Si       3.50713845       2.37602214      14.82677479
Si       4.30343667      20.72354649      13.13594112
Si       6.69704328      11.52083452      -1.91956095
Si      -2.40183153       9.94217749      11.16083709
Si      10.02171540      11.12007019       4.66360210
Si      -7.49232500       5.13573774       5.99220070
Si       6.05611820       2.21979659      -8.58519128
Si      18.43981661       9.93240975       3.62400319
Si       9.06641982      10.03139824      10.79427260
Si       0.95461271      -1.80327653       7.05737680
Si       3.41338430      11.66944108       4.95967107
Si      11.82535593       7.67923982      10.56286792
Si       5.14616660       9.71482023      -2.34012219
Si       1.33172638      13.51288996      18.07489084
Si       8.49579219      14.36337409       4.47234068
Si       3.24023788       8.91541882       8.53056834
Si       4.42194552       0.54113385       5.20728715
Si      15.21659049       5.66741936       7.46127358
Si      -0.35217392       4.49012920       3.79097796
Si       1.85797522       3.00166706      -0.82181693
Si      -0.21150422       0.86491245       0.33452660
Si      14.82585907      10.26862238      10.64942136
Si       2.64381855       9.49171038       5.66138329
Si       8.45336107       8.57503025       5.06476787
Si      11.64444759       5.11333178      12.38091823
Si       9.60337187      12.58303496      11.88682843
Si      10.31617897      -1.00846924      14.62796849
Si       1.56662794      19.51190920      17.19874186
Si      -0.75752203      11.17122226       4.90210710
Si       5.07227242      -1.30346450       7.76358958
Si       3.70137830       2.14366207      -3.01185260
Si       9.81201525      14.59234686       9.71660334
Si       7.56831202       6.10514965      -0.96626846
Si       3.49447971       3.68380139       6.12331300
Si      13.70003808      17.44575169      14.40900401
Si      13.69119144       8.89082786       9.21911926
Si      -3.96455740       8.12620160      13.27386301
Si      16.28305016       1.62515452       5.69090233
Si       9.90362611       1.81161098       1.81279522
Si       8.50402936      11.63259578      -4.82089379
Si       7.75468312      -1.24494168       5.67513360
Si       5.43998939       1.79327633       9.06885178
Si      13.39056313      11.89994920      15.18072883
Si      11.36662443       3.65567325       1.72066924
Si       6.08662600      -1.07598778      14.00262661
Si      19.16189952       3.20222647       3.34446769
Si       7.03257924       4.71795393      11.04650111
Si       9.50416713       4.48007738      17.69895700
Si       2.71834343       5.36789748      10.06432496
Si      -5.29980247       3.14671596       6.37038110
Si       3.31451730      11.00785092      -2.79987168
Si      13.15128604      -3.29437878       7.34619020
Si       2.12814786       5.15268096      12.21819136
Si      -3.31539847      -0.30916988       3.88095304
Si      -3.09286206       2.07489432       6.79437799
Si      -6.32218457       3.32560867      -5.31338461
Si      11.70747608      12.13939376      -0.96972644
Si      -3.97829566      -2.88611492       7.22240020
Si       5.42858625       3.09318784      -0.13880866
Si      -4.17191967       7.39415311       9.56554548
Si      12.22910580      15.33071186      11.52646443
Si      13.87161726       8.40075975      15.46264589
Si      11.31849242       2.97980766       4.84215808
Si       6.88733656      14.62711195      -3.05272237
Si       8.31248146      13.73763357       0.07153271
Si       4.73036163       6.04915661     -12.37994938
Si       2.79867330      -0.25444259      -0.87207953
Si      22.01886363       6.12424105      18.62582225
Si       2.25602784       7.55499189      14.22415021
Si       2.59484030      11.32966587      13.23637996
Si       9.52199645       0.32360410       7.41756349
Si      10.36879046       4.33604006      -3.28374888
Si       6.68668791      -5.79652636      13.64121673
Si       8.57236117      11.51077818       2.79621436
Si      20.07798189      13.16478867     -10.70674256
Si       6.80871478       6.14840219       8.16413641
Si       4.44134763      -0.72767454      10.50020625
Si      -5.25228093      16.96115877      11.71737967
Si       6.11605549      -5.11008184       4.99605264
Si       6.94805641       5.41431254       2.08639903
Si      16.80987806       4.54258700      15.56436229
Si      11.58876696      10.43988916       2.15478721
Si      19.55826404       0.85594734      12.75366622
Si      25.39431017      11.76258123       9.64269490
Si       4.76953949      18.08335210      11.76039345
Si       8.27323368       8.33471120      16.49302591
Si       3.19605553      -2.51672444      -5.59885816
Si       3.95315019       9.90043974      14.46555609
Si       6.93544112       9.96748074      15.49468680
Si      12.73333674       0.62276106       8.98884265
C        7.89677204       2.30461601       4.65578985
C        8.49750911      11.09584776      12.27592240
C        9.56419282       5.94913299       1.28374584
C        0.65127716      14.88359615       6.44071844
C        1.91535742      -1.02746658       8.59185242
C        8.03644803      -0.44837897      16.70060629
C       -2.24594733      11.49240238       7.84365386
C        8.22779961       0.74092159      15.62561222
C       11.15698934       7.57331021       6.74070261
C       14.68521762       4.07221753       8.31959331
C        5.63262540       3.46348173       8.01902147
C        6.45227980      -1.06491397      10.39268230
C        0.60022667      15.41860739       7.89377403
C        9.76897454       9.58147855      19.78932515
C       19.68718231       1.00048600      16.21748136
C       14.86929197       1.27655436      16.65256522
C       -1.58899503      10.27582057      -0.99634815
C       12.81052562      13.09176273       5.52030120
C        4.97621036       8.76982236      -0.70673976
C       13.63286039      15.80260516       5.66809215
C       13.13815707       8.22645267       4.77414074
C       10.44142149       5.33975245       2.32219861
C       13.46401486       3.14419640       6.33364408
C       -0.98802031       6.01962310      16.70658086
C        8.15548753      17.05072008       3.43213396
C        4.12266647       0.10142012       8.82340247
C       11.97292037       8.76901887      15.25345477
C        3.22077796       5.35354431       7.07381472
C        6.94793150       2.81611743       1.15139671
C        6.70555171      12.35945462      -4.86918518
C        1.25617581       0.79649181      15.23519088
C       10.26338679       7.10228303       7.80618439
C       11.17278478       7.52193269      15.64324782
C       -2.69629703       3.07634269       9.37077460
C       -5.62850075       7.75326751      12.18088955
C       -4.49095083       5.49727500       9.46329030
C        4.07828004      14.37407040      11.83127458
C        3.77690879       8.42964144       6.61116531
C       11.22057784      20.82358522      12.51667185
C        7.11008912       7.94823634       8.39162535
C        5.18890260      10.67806397      15.82596289
C        1.78383411       3.87045670       9.38408800
C       -1.54963492       7.79729362      -0.80175207
C       11.42353797       1.72111112       9.88729524
C        7.85884423       5.14050726       9.34074462
C        5.76938240       7.19717854       3.89611143
C        2.06774423       2.39395520       9.76427671
C       14.52819130       6.36903427       5.83486257
C        2.67995596      14.06469426       8.70151775
C        5.41901654       3.48030314       6.57674620
C       -1.24043825      13.40954258      11.71632839
C       17.59969228       8.40746937       2.86658652
C        0.14853823       6.67876307       2.30130824
C       -0.93713014      13.03664622      10.25784672
C        4.87044535      -1.24789769       5.65940948
C       12.81685494      15.49258925       4.45705455
C        4.56872298       1.72940103       2.55158805
C        9.91176538      11.08010207       1.44110396
C        5.80033612       8.32026095       7.90913635
C       11.17825324       3.40214580       8.04780354
C       10.26247718       5.75778273       8.18454953
C       11.76017011       9.03674858      -0.16164587
C        9.05832906       0.50190528      12.37133354
C        9.83923017       7.40885707      15.02691528
C        4.65993616       7.39682769       2.87190889
C        6.13111512      12.54446059       4.84069958
C        5.75936548       8.65345768      10.13901130
C       11.54061697       6.35781720       2.50385705
C       11.93500341      17.79577530      13.76108995
C       11.14269160       4.76487802       7.61085349
C        2.19626854       1.36831415       3.83735017
C        5.25755628       8.47931015       6.60332433
C       -2.89020960       8.70051845       5.86084910
C       14.56056675       8.41518707      13.62204586
C        0.03209667      -0.03368247      12.54326608
C        0.33251605       3.00480579       7.58672012
C       12.15359295       6.59938305       6.33399396
C        1.76349729      20.61737189       4.90574275
C        9.93556847      12.67039588      10.02995869
C        8.25248265       8.27651942      10.74684014
C        7.02481273      17.43303227       2.48575753
C       -3.00338828       0.14002654      15.91629341
C        2.75584796       8.76317703       1.79835916
C        0.41968560      13.43083022       5.92727688
C        7.10708884       6.77548467       3.32765936
C       11.88469855       8.92679585       4.77273132
C        5.18953132      12.20577839       1.82487614
C        3.63921442      14.67689728       3.38249656
C       13.49964800       1.77638987      16.86428289
C       -0.23500989       7.45316231      12.92889259
C       -0.55982406       7.04356206       5.56874006
C       -1.56982850      14.05786461       4.46437973
C       11.97255038       5.18943331       6.50464686
C       16.29490583      11.16839718       6.41250105
C       12.64771783       4.16063749       5.64961194
C        1.23411956      -0.12958704      11.84226386
C        1.86933205       1.35745958       8.78276721
C        4.97947989       8.70167789       8.99323121
C        2.56009519       0.41806975      14.75208527
C        1.42640792       9.28969228       2.48276192
C        2.42912895      12.17932198       3.38118907
C       11.68675880      10.27150163       4.10547232
C        8.36779028       5.30351229       0.61626877
C        5.06785069       6.10480551      -9.05013437
C        7.07712740       8.28688650       9.84004648
C       13.09084320       3.19314444      16.32286183
C        6.93984751       1.44157691       1.65429719
C        8.64589104       1.29061544      -2.68293240
""")

a = read(buf, format='extxyz')
s = ring_statistics(a, cutoff=1.8)

Here's a backtrace:

Program received signal SIGSEGV, Segmentation fault.
0x00002aaabd90dec3 in find_shortest_distances (seed=seed@entry=0x7ffffffef9a0, neighbours=neighbours@entry=0xacc480, root=root@entry=0, dist=dist@entry=0xad0870)
    at c/ring_statistics.cpp:65
65	                int j = neighbours[ni];
Missing separate debuginfos, use: debuginfo-install glibc-2.17-307.el7.1.x86_64 keyutils-libs-1.5.8-3.el7.x86_64 krb5-libs-1.15.1-46.el7.x86_64 libcom_err-1.42.9-17.el7.x86_64 libselinux-2.5-15.el7.x86_64 nss-softokn-freebl-3.44.0-8.el7_7.x86_64 openssl-libs-1.0.2k-19.el7.x86_64 pcre-8.32-17.el7.x86_64
(gdb) bt
#0  0x00002aaabd90dec3 in find_shortest_distances (seed=seed@entry=0x7ffffffef9a0, neighbours=neighbours@entry=0xacc480, root=root@entry=0, dist=dist@entry=0xad0870)
    at c/ring_statistics.cpp:65
#1  0x00002aaabd90e078 in distances_on_graph (nat=nat@entry=216, seed=seed@entry=0x7ffffffef9a0, neighbours=0xacc480, dist=0xad0870, diameter=diameter@entry=0x0)
    at c/ring_statistics.cpp:102
#2  0x00002aaabd90e3ec in py_distances_on_graph (self=self@entry=0x2aaabd87b6b0, args=args@entry=0x2aaabd880a00) at c/ring_statistics.cpp:152
#3  0x00002aaaaae46db4 in _PyMethodDef_RawFastCallKeywords () at Objects/call.c:698
#4  0x00002aaaaae47050 in _PyCFunction_FastCallKeywords (func=0x2aaabd86cf50, args=args@entry=0x2aaaaabd35c8, nargs=<optimized out>, kwnames=kwnames@entry=0x0)
    at Objects/call.c:734
#5  0x00002aaaaaea49ec in call_function (kwnames=0x0, oparg=<optimized out>, pp_stack=<synthetic pointer>) at Python/ceval.c:4568
#6  _PyEval_EvalFrameDefault () at Python/ceval.c:3093
#7  0x00002aaaaadfd471 in _PyEval_EvalCodeWithName () at Python/ceval.c:3930
#8  0x00002aaaaadfe483 in PyEval_EvalCodeEx () at Python/ceval.c:3959
#9  0x00002aaaaadfe4ab in PyEval_EvalCode (co=co@entry=0x2aaaaab894b0, globals=globals@entry=0x2aaaaabb9d70, locals=locals@entry=0x2aaaaabb9d70) at Python/ceval.c:524
#10 0x00002aaaaaf1c542 in run_mod () at Python/pythonrun.c:1035
#11 0x00002aaaaaf1dd67 in PyRun_FileExFlags () at Python/pythonrun.c:988
#12 0x00002aaaaaf1ff58 in PyRun_SimpleFileExFlags () at Python/pythonrun.c:429
#13 0x00002aaaaaf20109 in PyRun_AnyFileExFlags () at Python/pythonrun.c:84
#14 0x00002aaaaaf21eaa in pymain_run_file (p_cf=0x7fffffff015c, filename=<optimized out>, fp=0x4298d0) at Modules/main.c:427
#15 pymain_run_filename (cf=0x7fffffff015c, pymain=0x7fffffff0230) at Modules/main.c:1606
#16 pymain_run_python (pymain=0x7fffffff0230) at Modules/main.c:2867
#17 pymain_main () at Modules/main.c:3028
#18 0x00002aaaaaf220ce in _Py_UnixMain (argc=<optimized out>, argv=<optimized out>) at Modules/main.c:3063
#19 0x00002aaaabc2e555 in __libc_start_main () from /lib64/libc.so.6
#20 0x000000000040079e in _start ()

I've confirmed that reverting to commit acc20d3 gets rid of the seg fault, so it must be related to the changes in the neighbour list, but I haven't yet dug deep enough to figure out exactly what changed.

@pastewka or @chcltchunk, any idea what might be causing this? Did ring statistics make some assumptions about the neighbour list ordering which are no longer valid?

Hi,

Error

In matscipy.dislocation.plot_vitek(), the call to the atomman package function differential_displacement provokes the following error :

TypeError: differential_displacement() got an unexpected keyword argument 'show'
with atomman version >1.2.4.

Piece of code involved

Below, the call to atomman in matscipy :

am.defect.differential_displacement(base_system, disl_system,
                                        burgers,
                                        cutoff=neighborListCutoff,
                                        xlim=plot_range[0],
                                        ylim=plot_range[1],
                                        zlim=plot_range[2],
                                        plot_scale=plot_scale,
                                        show=show, save_file=save_file,
plot_axes=plot_axes)

Reason

It is because we changed the arguments passed to the function to make it more general (see here).

Proposed solution

The fix needs to account that differential_displacement now returns a matplotlib figure object, that can be saved later directly in plot_vitek.
When plot_vitek is called in dislocation.show_NEB_configurations(), there is a little change to make to be able to add new vitek plots on top of each other.

In the meantime, one can still use versions of atomman that are <=1.2.3.

EAM tables of kind 'eam' (DYNAMO funcfl format) contain the effective charge function :math:z_i, rather
than the pair potential phi_ij. The pair potential for elements i and j can be calculated from the charge functions as r*phi_ij=z_i z_j, see https://lammps.sandia.gov/doc/pair_eam.html. An additional prefactor 27.2*0.529 may be
necessary in order to convert to units of electronvolt.

I can't see where phi_ij is calculated from z_i in the matscipy/calculators/eam/io.py or
matscipy/calculators/eam/calculator.py. It seems that z_i is directly
assigned to rep. So the pair potential energy is calculated from z_i, and not from phi_ij.
Therefore, I believe that the current implementation of the eam style is wrong. The other styles
should be OK, because they store the pair potentials, and not the charge function.

The new implementation of the Hessian does not yet work for atomic system with multiple species.

For some reason writing and re-reading standard EAM files (kind="eam") fails.

It would be good to install Atomistica in the Travis-CI environment @pastewka do you think this is possible?

We should as a pyproject.toml file and at least switch to declarative setuptools build. My suggestion would be to additionally switch to Meson as a build environment, see here: https://thiblahute.gitlab.io/mesonpep517/pyproject.html

hi Guys!

Vikram and I were reviewing the code in get_delta_plus_epsilon(), because we wanted to understand exactly what was being calculated. Here are some suggestions:

1. you compute the deformation matrix, which is indeed delta+epsilon, as suggested by the name of the function, but then the name of the variable into which it is accumulated is just called epsilon, which is confusing.

2. The deformation matrix that is thus computed is not symmetric, so can include rotations. we suggest to take those out, using M.T*M

3. To actually get a strain, one has to subtract the identity, so we suggest that is done, so the final thing returned should be 0.5 (M.TM - I) where M is the thing accumulated in get_delta_plus_epsilon()

So far, while citing matscipy in our papers I had to use a web citation to the repository that always feels a bit ad-hoc. Thus I was thinking that it would be nice to have a paper published to be able to formally cite the code. Moreover, it can help all of us to get more formal recognition of our efforts. This issue is essentially to discuss how we can achieve this.

After a short discussion about it wit @jameskermode we decided that Journal of Open Source Software looks like a nice option. It requires only a very short (250-1000 words) high level description of the software, but the review is more focused on the software itself: installation, tests coverage and documentation. You can find more details on submission requirements and reviewer checklist pages.

We should host the paper.md file as a part of the repository, that can be in a short lived branch that is never merged to the main repository. Then everyone can add a small bit with the description of the parts of the code that they developed together with more general description. I am happy to volunteer to do the bookkeeping for this branch and the submission itself.

It seems to me that in the current shape matscipy already passes most of the requirements and writing a short summary should not take too long. However, before starting working on the submission it makes sense to understand if there are parts of matscipy that require a significant rework before the submission. I would suggest using the discussion on this issue to provide your opinions. On my site, I probably should enable building the documentation for the dislocation.py module as well as to think about solving a dependency on atomman package.

I am happy to discuss other options and suggestions as well.

In the function elastic_moduli(), defined in elasticity.py, a rotation of the vector u_b (based on the function argument l) into u_a = np.array([1, 0, 0]) is performed. That is, the argument vector is rotated into the x direction. A crucial bit here is the cross product

u_v = np.cross(u_a, u_b)

If I am not mistaken, a rotation starting with this operation corresponds to a rotation of u_a into u_b, i.e., a rotation of the x-vector into the chosen direction. Therefore, I believe it is correcter to write

u_v = np.cross(u_b, u_a)

or swap the definitions of u_a and u_b. I spotted there is an inconsistency with the Young's moduli as predicted by ELATE, if directions such as [101] and [-101] are not equivalent. So in many higher-symmetry systems this issue won't show up. Making the proposed changes makes the issue go away.

I got the following segfault during the test, which it looks like it is cause by theneighbourlist?

julia(98489,0x7fff72062000) malloc: *** mach_vm_map(size=18446744072217231360) failed (error code=3)
*** error: can't allocate region
*** set a breakpoint in malloc_error_break to debug
julia(98489,0x7fff72062000) malloc: *** mach_vm_map(size=18446744072217231360) failed (error code=3)
*** error: can't allocate region
*** set a breakpoint in malloc_error_break to debug

signal (11): Segmentation fault: 11
while loading /Users/ortner/.julia/v0.5/JuLIP/test/testvarcell.jl, in expression starting on line 68
py_neighbour_list at /Users/ortner/anaconda/lib/python2.7/site-packages/_matscipy.so (unknown line)

This is reproducible: it happens when I put too much (negative??) pressure on a cell so that it breaks all bonds and the cell starts expanding without bounds. Here is the sequence of cell matrices before the segfault:

F = [0.0 4.05 4.05; 4.05 0.0 4.05; 4.05 4.05 0.0]
F = [-344.689 348.739 348.739; 348.739 -344.689 348.739; 348.739 348.739 -344.689]
F = [-1723.45 1727.5 1727.5; 1727.5 -1723.45 1727.5; 1727.5 1727.5 -1723.45]
F = [-8617.23 8621.28 8621.28; 8621.28 -8617.23 8621.28; 8621.28 8621.28 -8617.23]

Maybe you don't want to treat it as a bug at all, but I was just wondering whether a more gentle mode of failure (e.g. throw an exception) would be possible.

Dear authors,

The provided paper is comprehensive and well-written. However, these comments may help improve it:

• As mentioned in the paper, there exists other packages for multi-scale coupling such as LibMultiScale and MultiBench, but the necessity of the proposed package is not well elaborated. Since it’s the most important paragraph of the statement of need, I appreciate it if the authors elaborate on this.
• I see duplicate references. For example, there are two identical Seidel et al. (2021a and 2021b). Can you please check the paper for this?
• The authors have referred to 2 papers using the coupling of the numerical models of the Poisson–Nernst–Planck equation in electrochemistry (Hörmann et al., 2023; Seidl et al. 2021b), but it doesn't seem that those papers use this coupling. Can you please elaborate on the mentioned usage?

P.S. This issue is related to openjournals/joss-reviews#5668

I'd like to setup cibuildwheel to build binary wheels using GitHub Actions that we can release on PyPI. This would also help with issues like #57 requesting support on a wider range of platforms. I've had good success doing this with f90wrap and quippy, and matscipy should be simpler as there's no Fortran (although there is C++, so #96 would still be valuable).

Greetings!
I noticed that method alpha_num_order referred a StackOverflow post, while a recent comment to it noticed that it doesn't work for sorting sets of version numbers alphanumerically; e.g., v1.0.1, v3.5.3, v3.2.4.
I'm trying to learn the usefulness of such small updates on StackOverflow. Would this comment help improve your code? I understand that might not happen in real life situation. In that case, do you think this comment can help prevent future defect (if the code were reused somwhere else)?
I'll really appreciate it if you could kindly give me some feedback or suggestions. Thank you very much for your time.
Have a nice day!

Following merge of #117 we need to update the installation instructions in the README

Dear authors,

The software has standard and comprehensive automated tests implemented using pytest. However, these minor points can be taken into account in this regard:

• I cannot find the FEniCS functionality being tested via the pytest tests. Is it because of the complicated installation procedure? I care about this functionality not only due to its usefulness but also because it's mentioned in the paper.
• All tests pass on my local machine, but some of them produce unclear characters. This is the output of the pytest run:

test_angle_distribution.py ....                                                                                             [  0%]
test_atomic_strain.py .                                                                                                     [  0%]
test_bop.py ............................................................................................................... [ 17%]
......                                                                                                                      [ 18%]
test_bulk_properties.py .............                                                                                       [ 20%]
test_c2d.py ......                                                                                                          [ 21%]
test_cauchy_born_corrector.py ........                                                                                      [ 22%]
test_committee.py ............                                                                                              [ 24%]
test_cubic_crystal_crack.py ......                                                                                          [ 25%]
test_dislocation.py ssssssssssssss.sssssssss.ss.ssssssss.s                                                                  [ 30%]
test_eam_average_atom.py ..                                                                                                 [ 30%]
test_eam_calculator.py .......                                                                                              [ 32%]
test_eam_calculator_forces_and_hessian.py ......                                                                            [ 32%]
test_eam_io.py ....                                                                                                         [ 33%]
test_elastic_moduli.py ..                                                                                                   [ 33%]
test_energy_release.py s                                                                                                    [ 33%]
test_ewald.py .................................                                                                             [ 38%]
test_ffi.py .                                                                                                               [ 39%]
test_full_to_Voigt.py ....                                                                                                  [ 39%]
test_hessian_finite_differences.py ...                                                                                      [ 40%]
test_hessian_precon.py ...                                                                                                  [ 40%]
test_hydrogenate.py .                                                                                                       [ 40%]
test_idealbrittlesolid.py ...                                                                                               [ 41%]
test_invariants.py .                                                                                                        [ 41%]
test_io.py .....                                                                                                            [ 42%]
test_mcfm.py ...                                                                                                            [ 42%]
test_neighbours.py ...................                                                                                      [ 45%]
test_numpy_tricks.py .                                                                                                      [ 45%]
test_opls.py .....                                                                                                          [ 46%]
test_opls_io.py .......                                                                                                     [ 47%]
test_pair_potential_calculator.py ......................                                                                    [ 50%]
test_poisson_nernst_planck_solver.py .                                                                                      [ 50%]
test_polydisperse_calculator.py ............                                                                                [ 52%]
test_pressurecoupling.py ....                                                                                               [ 53%]
test_ring_statistics.py .....                                                                                               [ 53%]
test_rotation_of_elastic_constants.py .                                                                                     [ 54%]
test_spatial_correlation_function.py ..                                                                                     [ 54%]
test_supercell_calculator.py .                                                                                              [ 54%]
test_surface_reconstruction.py ..                                                                                           [ 54%]
manybody/test_manybody_molecules.py .........xxxxxxxxx                                                                      [ 57%]
manybody/test_manybody_potentials.py x.x..x..x..........                                                                    [ 60%]
manybody/test_newmb.py XxXx.x............XXXX.X............XxXx.x............XXXX.X............XxXx.x............XXXX.X.... [ 75%]
........xxxx.x............xxxx.x............xxxx.x............xxxx.x............XxXx.x............XXXX.X............xxxx.x. [ 93%]

P.S. This issue is related to openjournals/joss-reviews#5668

I'm having trouble with the atomic_strain() calculation. I am passing in a crack slab that is thin in the z direction, and then the relaxed version of it, and computing the atomic strain. Some atoms come back with eps_zz = -1, there are domains of these, and the domain walls have strange strains (and large residuals in the least squares fit). The image below shows the hydrostatic strain. The zz strain essentially looks the same.