#1294 allow setting of hyper parameters

pints/functionaltests/_problems.py

@@ -49,7 +49,7 @@ class RunMcmcMethodOnTwoDimGaussian(RunMcmcMethodOnProblem):
     distribution.
     """
 
-    def __init__(self, method, n_chains, n_iterations, n_warmup):
+    def __init__(self, method, n_chains, n_iterations, n_warmup, method_hyper_parameters=None):
         pdf = pints.toy.GaussianLogPDF(mean=[0, 0], sigma=[1, 1])
 
         # Get initial parameters
@@ -59,13 +59,21 @@ def __init__(self, method, n_chains, n_iterations, n_warmup):
         initial_parameters = log_prior.sample(n=n_chains)
 
         # Set up sampler
-        sampler = pints.MCMCController(
+        controller = pints.MCMCController(
             pdf, n_chains, initial_parameters, method=method)
-        sampler.set_max_iterations(n_iterations)
-        sampler.set_log_to_screen(False)
+        controller.set_max_iterations(n_iterations)
+        controller.set_log_to_screen(False)
+
+        # Set hyper parameters, if required. This is different based on single/multi chain
+        if method_hyper_parameters is not None:
+            if issubclass(method, pints.MultiChainMCMC):
+                controller.sampler().set_hyper_parameters(method_hyper_parameters)
+            else:
+                for sampler in controller.samplers():
+                    sampler.set_hyper_parameters(method_hyper_parameters)
 
         # Infer posterior and throw away warm-up
-        chains = sampler.run()
+        chains = controller.run()
         chains = chains[:, n_warmup:]
 
         super().__init__(pdf, chains)