def check_history(self):
self.problem = pypesto.Problem(self.obj, self.lb, self.ub)
optimize_options = pypesto.OptimizeOptions(
allow_failed_starts=False
)
history_options = pypesto.HistoryOptions(
trace_record=True,
trace_record_hess=False,
trace_save_iter=1,
storage_file='tmp/traces/conversion_example_{id}.csv',
)
result = pypesto.minimize(
problem=self.problem,
optimizer=self.optimizer,
n_starts=1,
startpoint_method=pypesto.startpoint.uniform,
options=optimize_options,
history_options=history_options
)
# disable trace from here on
self.obj.history.options.trace_record = False
for start in result.optimize_result.list:
trace = start.history._trace
it_final = int(trace[('fval', np.NaN)].idxmin())
it_start = int(np.where(np.logical_not(
np.isnan(trace['fval'].values)
))[0][0])
self.assertTrue(np.isclose(
trace['x'].values[0, :], start.x0
).all())
self.assertTrue(np.isclose(
trace['x'].values[it_final, :], start.x
).all())
self.assertTrue(np.isclose(
trace['fval'].values[it_start, 0], start.fval0
))
funs = {
'fval': self.obj.get_fval,
'grad': self.obj.get_grad,
'hess': self.obj.get_hess,
'res': self.obj.get_res,
'sres': self.obj.get_sres,
'chi2': lambda x: res_to_chi2(self.obj.get_res(x)),
'schi2': lambda x: sres_to_schi2(*self.obj(
x,
(0, 1,),
pypesto.objective.constants.MODE_RES
))
}
for var, fun in funs.items():
for it in range(5):
if var in ['fval', 'chi2']:
if not np.isnan(trace[var].values[it, 0]):
self.assertTrue(np.isclose(
trace[var].values[it, 0],
fun(trace['x'].values[it, :])
))
elif var in ['hess', 'sres', 'res']:
if trace[var].values[it, 0] is not None:
self.assertTrue(np.isclose(
trace[var].values[it, 0],
fun(trace['x'].values[it, :])
).all())
elif self.obj.history.options[f'trace_record_{var}'] \
and not \
np.isnan(trace[var].values[it, :]).all():
self.assertTrue(np.isclose(
trace[var].values[it, :],
fun(trace['x'].values[it, :])
).all())
def check_history_consistency(self, start: pypesto.OptimizerResult):
def xfull(x_trace):
return self.problem.get_full_vector(x_trace,
self.problem.x_fixed_vals)
if isinstance(start.history, (CsvHistory, Hdf5History)):
# get index of optimal parameter
ix_admit = [
ix for ix, x in enumerate(start.history.get_x_trace())
if np.all(x >= self.problem.lb) and np.all(
x <= self.problem.ub)
]
it_final = np.nanargmin(start.history.get_fval_trace(ix_admit))
if isinstance(it_final, np.ndarray):
it_final = it_final[0]
it_final = ix_admit[it_final]
it_start = int(
np.where(
np.logical_not(np.isnan(
start.history.get_fval_trace())))[0][0])
assert np.allclose(xfull(start.history.get_x_trace(it_start)),
start.x0), type(start.history)
assert np.allclose(xfull(start.history.get_x_trace(it_final)),
start.x), type(start.history)
assert np.isclose(start.history.get_fval_trace(it_start),
start.fval0), type(start.history)
funs = {
FVAL:
self.obj.get_fval,
GRAD:
self.obj.get_grad,
HESS:
self.obj.get_hess,
RES:
self.obj.get_res,
SRES:
self.obj.get_sres,
CHI2:
lambda x: res_to_chi2(self.obj.get_res(x)),
SCHI2:
lambda x: sres_to_schi2(*self.obj(
x,
(
0,
1,
),
pypesto.C.MODE_RES,
)),
}
for var, fun in funs.items():
for it in range(5):
x_full = xfull(start.history.get_x_trace(it))
val = getattr(start.history, f'get_{var}_trace')(it)
if not getattr(self.history_options, f'trace_record_{var}',
True):
assert np.isnan(val)
continue
if np.all(np.isnan(val)):
continue
if var in [FVAL, CHI2]:
# note that we can expect slight deviations here since
# this fval/chi2 may be computed without sensitivities
# while the result here may be computed with with
# sensitivies activated. If this fails to often,
# increase atol/rtol
assert np.isclose(val, fun(x_full), rtol=1e-3,
atol=1e-4), var
elif var in [RES]:
# note that we can expect slight deviations here since
# this res is computed without sensitivities while the
# result here may be computed with with sensitivies
# activated. If this fails too often, increase atol/rtol
assert np.allclose(val, fun(x_full), rtol=1e-3,
atol=1e-4), var
elif var in [SRES]:
assert np.allclose(
val,
fun(x_full)[:, self.problem.x_free_indices],
), var
elif var in [GRAD, SCHI2]:
assert np.allclose(
val,
self.problem.get_reduced_vector(fun(x_full)),
), var
elif var in [HESS]:
assert np.allclose(
val,
self.problem.get_reduced_matrix(fun(x_full)),
), var
else:
raise RuntimeError('missing test implementation')