def test_vasp_relax(vasp_cmd, ZrO_vasp_rundir):
""" Test vasp.Relax.run() (Work in progress) """
rundir = str(ZrO_vasp_rundir)
# print("rundir:", rundir)
# additional vasp.Relax settings
settings = {"vasm_cmd": vasp_cmd}
with captured_output() as (sout, serr):
subprocess.run("ls", cwd=rundir)
def test_run(self):
"""Test vasp.run()"""
for i, case in enumerate(self.cases["run"]):
input = join(self.classdir, 'input_data', case['input_data'])
output = join(self.classdir, 'output_data', 'vasp', 'run',
'test_case_' + str(i))
cp_input(input, output)
with captured_output() as (sout, serr):
if self.has_vasp:
vasp.run(output, **case["settings"])
self.assertTrue(True) # Todo: add tests
else:
self.assertRaisesRegexp(OSError,
"No such file or directory",
vasp.run, output,
**case["settings"])
def test_checkhull(self):
"""Test casm-learn --exRFE"""
if self.has_projects:
# ZrO test project construction
proj = project.Project(self.ZrO_dir, verbose=False)
# files and directories
tmp_dir = join(proj.path, '.casm', 'tmp')
fit_dir = join(proj.path, 'fit')
fit_RFE = join(fit_dir, 'fit_RFE.json')
default_bset = join(self.ZrO_dir, 'basis_sets', 'bset.default')
default_bspecs = join(default_bset, 'bspecs.json')
test_bset = join(self.ZrO_dir, 'basis_sets', 'bset.test')
test_bspecs = join(test_bset, 'bspecs.json')
test_clex = join(self.ZrO_dir, 'cluster_expansions',
'clex.formation_energy', 'calctype.default',
'ref.default', 'bset.test')
test_eci = join(test_clex, 'eci.test', 'eci.json')
# setup
def _clean():
if os.path.exists(tmp_dir):
shutil.rmtree(tmp_dir)
if os.path.exists(fit_dir):
shutil.rmtree(fit_dir)
if os.path.exists(test_bset):
shutil.rmtree(test_bset)
if os.path.exists(test_clex):
shutil.rmtree(test_clex)
_clean()
os.mkdir(fit_dir)
# test 'casm-learn --exREF'
testargs = ["casm-learn", "--exRFE"]
with contexts.captured_output(wd=fit_dir) as (sout, serr):
with contexts.patch.object(sys, 'argv', testargs):
casm_learn.main()
input = json.loads(sout.getvalue())
self.assertEqual(input['feature_selection']['method'], 'RFE')
self.assertEqual(input['estimator']['method'], 'LinearRegression')
# set random_state for CV score
input['problem_specs']['cv']['kwargs']['random_state'] = 0
# have 'checkhull' only checking training configs
input['checkhull'] = {'selection': 'train', 'write_results': True}
# save 'fit_RFE.json'
with open(fit_RFE, 'wb') as f:
f.write(six.u(json.dumps(input, indent=2)).encode('utf-8'))
# create 'test' bset and eci
proj.capture('settings --new-bset test')
proj.capture('settings --new-eci test')
shutil.copyfile(default_bspecs, test_bspecs)
stdout, stderr, returncode = proj.capture('bset -u')
#print("OK 1")
# create 'train' selection
with contexts.captured_output(wd=fit_dir) as (sout, serr):
proj.capture(
"select --set 'and(is_calculated,lt(comp(a),0.695))' -o train"
)
#print("OK 2")
# fit
testargs = 'casm-learn -s fit_RFE.json --quiet'.split()
#print testargs
with contexts.captured_output(wd=fit_dir) as (sout, serr):
with contexts.patch.object(sys, 'argv', testargs):
casm_learn.main()
self.assertTrue(
os.path.exists(join(fit_dir, 'fit_RFE_halloffame.pkl')))
self.assertTrue(os.path.exists(join(fit_dir, 'fit_RFE_specs.pkl')))
#print("OK 3")
# view hall of fame
testargs = 'casm-learn -s fit_RFE.json --hall --quiet'.split()
#print testargs
with contexts.captured_output(wd=fit_dir) as (sout, serr):
with contexts.patch.object(sys, 'argv', testargs):
casm_learn.main()
lines = sout.getvalue().splitlines()
res = dict(zip(lines[0].split(), lines[2].split()))
self.assertEqual(len(lines), 3)
self.assertEqual(res['Selected'],
'0111001101000000001101001001000100010010...')
self.assertEqual(int(res['#Selected']), 25)
self.assertAlmostEqual(float(res['CV']),
0.016540883) # depends on 'random_state'
self.assertAlmostEqual(float(res['RMS']), 0.014697276)
self.assertAlmostEqual(float(res['wRMS']), 0.014697276)
self.assertEqual(res['Estimator'], 'LinearRegression')
self.assertEqual(res['FeatureSelection'], 'RFE')
#print("OK 4")
# check data in pickle file
hall = learn.open_halloffame(
join(fit_dir, 'fit_RFE_halloffame.pkl'))
self.assertEqual(len(hall), 1)
# open input file and set default values
input = learn.open_input(fit_RFE)
#testargs = 'casm-learn -s fit_RFE.json --checkhull --indiv 0 --quiet'.split()
#print testargs
with contexts.working_dir(wd=fit_dir):
learn.checkhull(input, hall, indices=[0], verbose=False)
self.assertTrue(
os.path.exists(join(fit_dir, 'checkhull_fit_RFE_0_dft_gs')))
self.assertTrue(
os.path.exists(join(fit_dir, 'checkhull_fit_RFE_0_clex_gs')))