[MSM] fix dt_model in bayesian msm, added tests for other estimators

appveyor.yml

@@ -32,7 +32,7 @@ install:
 
   - conda config --set always_yes true 
   - conda config --add channels conda-forge
-  - conda install -q conda-build=2.1.4 jinja2
+  - conda install -q conda-build=2.1.15 jinja2
 
   # use agg backend in matplotlib to avoid gui popup, which can not be closed.
   - echo %userprofile%

devtools/conda-recipe/meta.yaml

@@ -51,6 +51,7 @@ test:
   requires:
     - pytest 
     - pytest-cov
+    - pytest-faulthandler
 
 about:
   home: http://emma-project.org

doc/source/CHANGELOG.rst

@@ -19,6 +19,9 @@ Changelog
 - datasets: fixed get_multi_temperature_data and get_umbrella_sampling_data for Python 3. #1102
 - coordinates: fixed StreamingTransformers (TICA, Kmeans, etc.) not respecting the in_memory flag. #1112
 - coordinates: made TrajectoryInfoCache more fail-safe in case of concurrent processes. #1122
+- msm: fix setting of dt_model for BayesianMSM. This bug led to wrongly scaled time units for mean first passage times,
+  correlation and relaxation times as well for timescales for this estimator. #1116
+
 
 2.4 (05-19-2017)
 ----------------

pyemma/msm/estimators/maximum_likelihood_msm.py

@@ -149,6 +149,7 @@ def __init__(self, lag=1, reversible=True, count_mode='sliding', sparse=False,
         # time step
         self.dt_traj = dt_traj
         self.timestep_traj = _TimeUnit(dt_traj)
+        self.dt_model = self.dt_traj
 
         # score
         self.score_method = score_method

pyemma/msm/tests/test_amm.py

@@ -157,6 +157,9 @@ def test_timestep(self):
     def test_transition_matrix(self):
         self._transition_matrix(self.amm)
 
+    def test_dt_model(self):
+        self._dt_model(self.amm)
+
     # ---------------------------------
     # SIMPLE STATISTICS
     # ---------------------------------

pyemma/msm/tests/test_bayesian_msm.py

@@ -51,9 +51,9 @@ def setUpClass(cls):
         cls.nsamples = 100
 
         cls.lag = 100
-        cls.bmsm_rev = bayesian_markov_model(obs_macro, cls.lag,
+        cls.bmsm_rev = bayesian_markov_model(obs_macro, cls.lag, dt_traj='4 fs',
                                              reversible=True, nsamples=cls.nsamples)
-        cls.bmsm_revpi = bayesian_markov_model(obs_macro, cls.lag,
+        cls.bmsm_revpi = bayesian_markov_model(obs_macro, cls.lag, dt_traj='4 fs',
                                                reversible=True, statdist=pi_macro,
                                                     nsamples=cls.nsamples)
 
@@ -309,6 +309,11 @@ def _timescales_stats(self, msm):
 
     # TODO: these tests can be made compact because they are almost the same. can define general functions for testing
     # TODO: samples and stats, only need to implement consistency check individually.
+
+    def test_dt_model(self):
+        from pyemma.util.units import TimeUnit
+        tu = TimeUnit("4 fs").get_scaled(self.bmsm_rev.lag)
+        self.assertEqual(self.bmsm_rev.dt_model, tu)
 
 if __name__ == "__main__":
     unittest.main()

pyemma/msm/tests/test_hmsm.py

@@ -406,6 +406,11 @@ def test_simulate_HMSM(self):
         assert (len(traj) <= N)
         assert (len(np.unique(traj)) <= len(hmsm.transition_matrix))
 
+    def test_dt_model(self):
+        from pyemma.util.units import TimeUnit
+        tu = TimeUnit("1 step").get_scaled(self.hmsm_lag10.lag)
+        self.assertEqual(self.hmsm_lag10.dt_model, tu)
+
     # ----------------------------------
     # MORE COMPLEX TESTS / SANITY CHECKS
     # ----------------------------------

pyemma/msm/tests/test_msm.py

@@ -342,6 +342,19 @@ def test_timestep(self):
         self._timestep(self.msmrevpi_sparse)
         self._timestep(self.msm_sparse)
 
+    def _dt_model(self, msm):
+        from pyemma.util.units import TimeUnit
+        tu = TimeUnit("1 step").get_scaled(self.msm.lag)
+        self.assertEqual(msm.dt_model, tu)
+
+    def test_dt_model(self):
+        self._dt_model(self.msmrev)
+        self._dt_model(self.msmrevpi)
+        self._dt_model(self.msm)
+        self._dt_model(self.msmrev_sparse)
+        self._dt_model(self.msmrevpi_sparse)
+        self._dt_model(self.msm_sparse)
+
     def _transition_matrix(self, msm):
         P = msm.transition_matrix
         # should be ndarray by default

pyemma/util/units.py

@@ -91,6 +91,9 @@ def __str__(self):
         else:
             return str(self._factor)+' '+self._unit_names[self._unit]
 
+    def __repr__(self):
+        return "[TimeUnit {}]".format(self)
+
     @property
     def dt(self):
         return self._factor
@@ -140,6 +143,13 @@ def rescale_around1(self, times):
         # nothing to do
         return times, self._unit
 
+    def __eq__(self, other):
+        if not isinstance(other, TimeUnit):
+            return False
+
+        return self._unit == other._unit and self._factor == other._factor
+
+
 def bytes_to_string(num, suffix='B'):
     """
     Returns the size of num (bytes) in a human readable form up to Yottabytes (YB).