def test_fit_constraint_simple_inequality():
""" Check "smaller than" relation during fitting.
"""
dfile = data_file_dl.get_data_file(examplefile)
data = pcf.readfiles.openAny(dfile)
corr = pcf.Correlation(
correlation=data["Correlation"][0],
traces=data["Trace"][0],
corr_type=data["Type"][0],
filename=os.path.basename(dfile),
title="test correlation",
fit_model=6035 # confocal 3D+3D)
)
corr.fit_parameters_variable = [
True, True, True, True, False, False, False
]
# crop triplet data
corr.fit_ival[0] = 8
pcf.Fit(corr)
assert corr.fit_parameters[1] <= corr.fit_parameters[2]
# -> deliberately reverse everything and try again
corr.fit_parameters[1], corr.fit_parameters[2] = corr.fit_parameters[
2], corr.fit_parameters[1]
corr.fit_parameters[3] = 1 - corr.fit_parameters[3]
pcf.Fit(corr)
# This tests also for equality
assert corr.fit_parameters[1] <= corr.fit_parameters[2]
if corr.fit_parameters[1] == corr.fit_parameters[2]:
print(
"found identity of fit parameters - multiplying by two to see if relation holds"
)
corr.fit_parameters[2] *= 2
pcf.Fit(corr)
assert corr.fit_parameters[1] < corr.fit_parameters[2]
def test_fit_constraint_sum_smaller_one():
""" Check "a+b<c" relation during fitting.
"""
dfile = data_file_dl.get_data_file(examplefile)
data = pcf.readfiles.openAny(dfile)
corr = pcf.Correlation(
correlation=data["Correlation"][0],
traces=data["Trace"][0],
corr_type=data["Type"][0],
filename=os.path.basename(dfile),
title="test correlation",
fit_model=6081 # confocal 3D+3D)
)
pcf.Fit(corr)
assert corr.fit_parameters[4] + corr.fit_parameters[5] < 1
parms0 = np.array([
1.13827592342, # n
3.0918704e-05, # τ₁ [ms]
1.98835792339, # τ₂ [ms]
2000.0, # τ₃ [ms]
0.972264423555, # F₁
0.021400173882, # F₂
5.0, # SP
1.0, # α₂₁
1.0, # α₃₁
1e-08, # τ_trip [ms]
0.0, # T
0.0, # offset
])
corr.fit_parameters = parms0
vary = [False] * 12
vary[4] = vary[5] = True
corr.fit_parameters_variable = vary
# crop triplet data
pcf.Fit(corr)
assert corr.fit_parameters[4] + corr.fit_parameters[5] < 1
# -> deliberately reverse everything and try again
corr.fit_parameters[4], corr.fit_parameters[5] = corr.fit_parameters[
5], corr.fit_parameters[4]
pcf.Fit(corr)
# This tests also for equality
assert corr.fit_parameters[4] + corr.fit_parameters[5] < 1
def test_basic():
"""This is a very rudimentary test for the session handling"""
dfile = data_file_dl.get_data_file(examplefile)
data = pcf.readfiles.open_any(dfile)
corr = pcf.Correlation(
correlation=data["Correlation"][0],
traces=data["Trace"][0],
corr_type=data["Type"][0],
filename=os.path.basename(dfile),
title="test correlation",
fit_model=6035 # confocal 3D+3D)
)
corr.fit_parameters_variable = [
True, True, True, True, False, False, False
]
# crop triplet data
corr.fit_ival[0] = 8
pcf.Fit(corr)
tmpdir = tempfile.mkdtemp(prefix="pycorrfit_tests_")
path = pathlib.Path(tmpdir) / "session.pcfs"
fiterr = []
for ii, fitpid in enumerate(corr.fit_results["fit parameters"]):
fiterr.append(
[int(fitpid),
float(corr.fit_results["fit error estimation"][ii])])
Infodict = {
"Correlations": {
1: [corr.lag_time, corr.correlation]
},
"Parameters": {
1: [
"#1:", corr.fit_model.id, corr.fit_parameters,
corr.fit_parameters_variable, corr.fit_ival,
[3, 3, 5, corr.fit_algorithm], [None, None], True, None,
[[0.0, np.inf], [0.0, np.inf], [0.0, np.inf],
[0.0, 0.9999999999999], [-np.inf, np.inf]]
]
},
"Supplements": {
1: {
"FitErr": fiterr,
"Chi sq": float(corr.fit_results["chi2"]),
"Global Share": [],
}
},
"External Functions": {},
"Traces": {},
"Comments": {
"Session": "No comment."
},
"Backgrounds": {},
"External Weights": {},
"Preferences": {},
}
pcf.openfile.SaveSessionData(sessionfile=str(path), Infodict=Infodict)
ldt = pcf.openfile.LoadSessionData(str(path))
# lag time only, shape (N,)
assert np.allclose(data["Correlation"][0][:, 0], ldt["Correlations"][1][0])
# lag time and correlation, shape (N, 2)
assert np.allclose(corr.correlation, ldt["Correlations"][1][1])
# parameters
assert corr.fit_model.id == ldt["Parameters"][0][1]
assert np.allclose(corr.fit_parameters, ldt["Parameters"][0][2])
assert np.allclose(corr.fit_parameters_variable, ldt["Parameters"][0][3])
assert np.allclose(corr.fit_ival, ldt["Parameters"][0][4])
shutil.rmtree(tmpdir, ignore_errors=True)