Merge pull request #123 from boschresearch/backport-fix-woehler-warnings

Backport fix woehler warnings

CHANGELOG.md

@@ -9,7 +9,9 @@ In this file noteworthy changes of new releases of pyLife are documented since
 
 * Massive performance improvement when reading element connectivity by
   `odbclient.` Note that also `odbserver` needs to be updated in its
-  environment.
+  environment. (#121)
+
+* Fix some warnings on import
 
 
 ## pylife-2.1.2

setup.cfg

@@ -117,6 +117,8 @@ addopts =
     --cov src/pylife --cov-append -m "not slow_acceptance and not demos"
     --doctest-modules
     --ignore=src/pylife/materialdata/woehler/bayesian.py
+    --ignore=src/pylife/strength/helpers.py
+    --ignore=src/pylife/strength/sn_curve.py
 
 norecursedirs =
     dist

src/pylife/materialdata/woehler/elementary.py

@@ -70,7 +70,7 @@ def analyze(self, **kwargs):
 
         Parameters
         ----------
-        \*\*kwargs : kwargs arguments
+        **kwargs : kwargs arguments
             Arguments to be passed to the derived class
         """
         if len(self._fd.load.unique()) < 2:

src/pylife/materialdata/woehler/maxlike.py

@@ -125,42 +125,39 @@ def warn_and_fix_if_less_than_two_mixed_levels():
         warn_and_fix_if_no_runouts()
         warn_and_fix_if_less_than_two_mixed_levels()
 
-        p_opt = initial_wcurve.to_dict()
+        prms_to_optimize = initial_wcurve.to_dict()
         for k in fixed_prms:
-            p_opt.pop(k)
+            prms_to_optimize.pop(k)
 
-        if not p_opt:
+        if not prms_to_optimize:
             raise AttributeError('You need to leave at least one parameter empty!')
-        var_opt = optimize.fmin(
-            self.__likelihood_wrapper, [*p_opt.values()],
-            args=([*p_opt], fixed_prms),
+        optimized_prms = optimize.fmin(
+            self.__likelihood_wrapper, [*prms_to_optimize.values()],
+            args=([*prms_to_optimize], fixed_prms),
             full_output=True,
             disp=False,
             maxiter=1e4,
             maxfun=1e4,
-        )
-        res = {}
-        res.update(fixed_prms)
-        res.update(zip([*p_opt], var_opt[0]))
+        )[0]
 
-        return self.__make_parameters(res)
+        # TODO: Change to following line when python 3.8 is dropped
+        #r esult = fixed_prms | dict(zip(prms_to_optimize, optimized_prms))
+
+        result = {}
+        result.update(fixed_prms)
+        result.update(zip(prms_to_optimize, optimized_prms))
+
+        return self.__make_parameters(result)
 
     def __make_parameters(self, params):
-        params['SD'] = np.abs(params['SD'])
-        params['TS'] = np.abs(params['TS'])
-        params['k_1'] = np.abs(params['k_1'])
-        params['ND'] = np.abs(params['ND'])
-        params['TN'] = np.abs(params['TN'])
-        return params
+        return {k: np.abs(v) for k, v in params.items()}
 
     def __likelihood_wrapper(self, var_args, var_keys, fix_args):
-        ''' 1) Finds the start values to be optimized. The rest of the paramters are fixed by the user.
-            2) Calls function mali_sum_lolli to calculate the maximum likelihood of the current
-            variable states.
-        '''
+        # TODO: Change to following line when python 3.8 is dropped
+        # args = self.__make_parameters(fix_args | dict(zip(var_keys, var_args)))
+
         args = {}
         args.update(fix_args)
         args.update(zip(var_keys, var_args))
         args = self.__make_parameters(args)
-
-        return -self._lh.likelihood_total(args['SD'], args['TS'], args['k_1'], args['ND'], args['TN'])
+        return -self._lh.likelihood_total(**args)

src/pylife/strength/meanstress.py

@@ -458,6 +458,9 @@ def sum_intervals(iv, ranges, obj):
 
         range_itv_idx, means_itg_idx = resulting_intervals()
 
+        # TODO: remove once we drop python 3.8 support
+        future_stack = {} if pd.__version__ < "2.1" else {'future_stack': True}
+
         if len(self._remaining_names) > 0:
             remaining_idx = self._obj.index.to_frame(index=False).groupby(self._remaining_names).first().index
             result = (
@@ -466,7 +469,9 @@ def sum_intervals(iv, ranges, obj):
             )
             result.index = remaining_idx
             result.columns = pd.MultiIndex.from_arrays([result.columns, means_itg_idx])
-            result = result.stack(['range', 'mean']).reorder_levels(['range', 'mean'] + self._remaining_names)
+            result = result.stack(['range', 'mean'], **future_stack).reorder_levels(
+                ['range', 'mean'] + self._remaining_names
+            )
         else:
             result = range_itv_idx.to_series().apply(lambda iv: sum_intervals(iv, ranges, self._obj))
             result.index = pd.MultiIndex.from_arrays([result.index, means_itg_idx])

src/pylife/stress/rainflow/fourpoint.py

@@ -25,7 +25,7 @@
 
 
 class FourPointDetector(AbstractDetector):
-    """Implements four point rainflow counting algorithm.
+    r"""Implements four point rainflow counting algorithm.
 
     .. jupyter-execute::
 

tests/materialdata/woehler/test_analyzer.py

@@ -514,6 +514,7 @@ def test_woehler_max_likelihood_full_without_fixed_params():
     np.testing.assert_almost_equal(we.bayesian_information_criterion(), bic, decimal=2)
 
 
+@pytest.mark.filterwarnings("ignore:invalid value encountered in subtract")
 def test_woehler_max_likelihood_full_without_fixed_params_no_runouts():
     expected = pd.Series({
         'SD': 0,