fix #76 and modify test to catch it

sncosmo/fitting.py

@@ -533,11 +533,9 @@ def _nest_lc(data, model, vparam_names, modelcov,
              bounds=None, priors=None, ppfs=None, tied=None,
              nobj=100, maxiter=10000, maxcall=1000000, verbose=False):
     """Assumes that data has already been standardized.
-
-    Run `data = standardize_data(data)`"""
-
-    # Order vparam_names the same way it is ordered in the model:
-    vparam_names = [s for s in model.param_names if s in vparam_names]
+    (do `data = standardize_data(data)`).  Output samples and parameter
+    names are not reordered (order wil be the same as input parameter list).
+    """
 
     if ppfs is None:
         ppfs = {}
@@ -596,7 +594,7 @@ def loglikelihood(parameters):
 
     res = nest.nest(loglikelihood, prior, npar, nipar, nobj=nobj,
                     maxiter=maxiter, maxcall=maxcall, verbose=verbose)
-    res.vparam_names = vparam_names
+    res.vparam_names = copy.copy(vparam_names)
     res.ndof = len(data) - len(vparam_names)
     return res
 
@@ -706,6 +704,9 @@ def nest_lc(data, model, vparam_names, bounds, guess_amplitude_bound=False,
     model = copy.copy(model)
     bounds = copy.copy(bounds)  # need to copy this dict b/c we modify it below
 
+    # Order vparam_names the same way it is ordered in the model:
+    vparam_names = [s for s in model.param_names if s in vparam_names]
+
     # Drop data that the model doesn't cover.
     data = cut_bands(data, model, z_bounds=bounds.get('z', None))
 
@@ -747,7 +748,7 @@ def nest_lc(data, model, vparam_names, bounds, guess_amplitude_bound=False,
     return res, model
 
 
-def mcmc_lc(data, model, param_names, errors, bounds=None, nwalkers=10,
+def mcmc_lc(data, model, vparam_names, errors, bounds=None, nwalkers=10,
             nburn=100, nsamples=500, verbose=False):
     """Run an MCMC chain to get model parameter samples.
 
@@ -766,7 +767,7 @@ def mcmc_lc(data, model, param_names, errors, bounds=None, nwalkers=10,
         Table of photometric data. Must include certain column names.
     model : `~sncosmo.Model`
         The model to fit.
-    param_names : iterable
+    vparam_names : iterable
         Model parameters to vary.
     errors : iterable
         The starting positions of the walkers are randomly selected from a
@@ -795,21 +796,25 @@ def mcmc_lc(data, model, param_names, errors, bounds=None, nwalkers=10,
         raise ImportError("mcmc_lc() requires the emcee package.")
 
     data = standardize_data(data)
-    ndim = len(param_names)
-    idx = np.array([model.param_names.index(name) for name in param_names])
+
+    # Order vparam_names the same way it is ordered in the model:
+    vparam_names = [s for s in model.param_names if s in vparam_names]
+
+    ndim = len(vparam_names)
+    idx = np.array([model.param_names.index(name) for name in vparam_names])
 
     # Check that z is bounded if it is being varied.
     if bounds is None:
         bounds = {}
-    if 'z' in param_names:
+    if 'z' in vparam_names:
         if 'z' not in bounds or None in bounds['z']:
             raise ValueError('z must be bounded if fit.')
 
     # Drop data that the model doesn't cover.
     data = cut_bands(data, model, z_bounds=bounds.get('z', None))
 
     # Convert bounds indicies to integers
-    bounds_idx = dict([(param_names.index(name), bounds[name])
+    bounds_idx = dict([(vparam_names.index(name), bounds[name])
                        for name in bounds])
 
     # define likelihood

sncosmo/tests/test_fitting.py

@@ -47,9 +47,12 @@ def setup_class(self):
 
     def test_fit_lc(self):
         """Ensure that fit results match input model parameters (data are
-        noise-free)."""
+        noise-free).
+
+        Pass in parameter names in order different from that stored in
+        model; tests parameter re-ordering."""
         res, fitmodel = sncosmo.fit_lc(self.data, self.model,
-                                       ['z', 't0', 'amplitude'],
+                                       ['amplitude', 'z', 't0'],
                                        bounds={'z': (0., 1.0)})
 
         # set model to true parameters and compare to fit results.
@@ -70,14 +73,18 @@ def test_wrong_param_names(self):
             res, fitmodel = sncosmo.fit_lc(self.data, self.model, [])
 
     def test_nest_lc(self):
-        """Ensure that nested sampling runs."""
+        """Ensure that nested sampling runs.
+
+        Pass in parameter names in order different from that stored in
+        model; tests parameter re-ordering.
+        """
 
         np.random.seed(0)  # seed the RNG for reproducible results.
 
         self.model.set(**self.params)
 
         res, fitmodel = sncosmo.nest_lc(
-            self.data, self.model, ['z', 't0', 'amplitude'],
+            self.data, self.model, ['amplitude', 'z', 't0'],
             bounds={'z': (0., 1.0)}, guess_amplitude_bound=True, nobj=50)
 
         assert_allclose(fitmodel.parameters, self.model.parameters, rtol=0.05)