TST: Add (failing) unittests for LBA.

hddm/tests/test_likelihoods.py

@@ -263,38 +263,53 @@ def test_failure_mode(self):
             st = 0.1
 
 
-def test_lba():
-    try:
-        import rpy2
-        import rpy2.robjects as robjects
-        from rpy2.robjects.packages import importr
-
-    except ImportError:
-        return
-    import hddm
-    import numpy as np
-
-    robjects.r("source('lba-math.r')")
-    for i in range(100):
-        x = np.random.randn() * 3
-        A = np.random.rand() * 2
-        b = A + np.random.rand() * 2
-        v1 = np.random.rand() * 2
-        v2 = np.random.rand() * 2
-        sv = np.random.rand() * 2 + .5
-        pdf = robjects.r("n1PDF") #t,x0max,chi,drift,sdI
-        if x > 0:
-            drifts = rpy2.robjects.vectors.FloatVector([v1, v2])
-        else:
-            drifts = rpy2.robjects.vectors.FloatVector([v2, v1])
-
-        r_result = pdf(abs(x), A, b, drifts, sv)
-        hddm_result = np.exp(hddm.lba.lba_like(np.array([x]), A, b, 0, sv, v1, v2))
-
-        np.testing.assert_almost_equal(r_result, hddm_result, 6,
-                                       "Parameters x=%f, A=%f, b=%f, v1=%f, v2=%f, sv=%f" %(x, A, b, v1, v2, sv))
+class TestLBA(unittest.TestCase):
+    def test_lba(self):
+        try:
+            import rpy2
+            import rpy2.robjects as robjects
+            from rpy2.robjects.packages import importr
 
+        except ImportError:
+            return
+        import hddm
+        import numpy as np
+
+        robjects.r("source('lba-math.r')")
+        for i in range(100):
+            x = np.random.randn() * 3
+            A = np.random.rand() * 2
+            b = A + np.random.rand() * 2
+            v1 = np.random.rand() * 2
+            v2 = np.random.rand() * 2
+            sv = np.random.rand() * 2 + .5
+            pdf = robjects.r("n1PDF") #t,x0max,chi,drift,sdI
+            if x > 0:
+                drifts = rpy2.robjects.vectors.FloatVector([v1, v2])
+            else:
+                drifts = rpy2.robjects.vectors.FloatVector([v2, v1])
+
+            r_result = pdf(abs(x), A, b, drifts, sv)
+            hddm_result = np.exp(hddm.lba.lba_like(np.array([x]), A, b, 0, sv, v1, v2))
+
+            np.testing.assert_almost_equal(r_result, hddm_result, 6,
+                                           "Parameters x=%f, A=%f, b=%f, v1=%f, v2=%f, sv=%f" %(x, A, b, v1, v2, sv))
+
+
+    @SkipTest
+    def test_pdf_integrate_to_one(self):
+        np.random.seed(123)
+        for tests in range(5):
+            x = np.random.randn() * 3
+            A = np.random.rand() * 2
+            b = A + np.random.rand() * 2
+            v1 = np.random.rand() * 2
+            v2 = np.random.rand() * 2
+            sv = np.random.rand() * 2 + .5
+            func = lambda x: np.exp(hddm.lba.lba_like(np.array([x]), A, b, 0, sv, v1, v2))
+            integ, error = sp.integrate.quad(func, a=-np.inf, b=np.inf)
 
+            np.testing.assert_almost_equal(integ, 1, 2)
 
 if __name__=='__main__':
     print "Run nosetest."

hddm/tests/test_models.py

@@ -348,5 +348,59 @@ def test_posterior_plots_breakdown():
     ppc = hddm.utils.post_pred_gen(m, samples=10)
     hddm.utils.post_pred_stats(data, ppc)
 
+
+class TestRecovery(unittest.TestCase):
+    def __init__(self, *args, **kwargs):
+        super(TestRecovery, self).__init__(*args, **kwargs)
+        self.iter = 2000
+        self.burn = 20
+        np.random.seed(1)
+
+    def runTest(self):
+        return
+
+    @SkipTest
+    def test_HLBA_flat(self):
+        params_true = {'A': .5, 'b': 1, 't': .3, 's': .5, 'v': [.5, .6, .7, .8]}
+        extended_params, merged_params = extend_params(params_true)
+
+        data, params_true = hddm.models.hlba_truncated.gen_rand_data(extended_params, size=100, subjs=1)
+        model = hddm.models.HLBA(data, depends_on={'v': 'condition'})
+        model.find_starting_values()
+        model.sample(self.iter, burn=self.burn)
+        model.gen_stats()
+        model.print_stats()
+        for param, true_val in merged_params.iteritems():
+            np.testing.assert_almost_equal(true_val, model.nodes_db.ix[param]['mean'])
+
+    @SkipTest
+    def test_HLBA_hierarchical(self):
+        params = hddm.models.hlba_truncated.gen_rand_params(cond_dict={'v': [.5, .6, .75, .8]})
+        data, params_true = hddm.models.hlba_truncated.gen_rand_data(params, size=100, subjs=10)
+        model = hddm.models.HLBA(data, depends_on={'v': 'condition'})
+        model.find_starting_values()
+        model.sample(self.iter, burn=self.burn)
+
+
+def extend_params(params):
+    # Find list
+    extend_param = [param for param, val in params.iteritems() if isinstance(val, (list, tuple))]
+    if len(extend_param) > 1:
+        raise ValueError('Only one parameter can be extended')
+    extend_param = extend_param[0]
+
+    fixed_params = [param for param, val in params.iteritems() if not isinstance(val, (list, tuple))]
+
+    out_extended = {}
+    out_merged = {k: params[k] for k in fixed_params}
+    for i_cond, extend_val in enumerate(params[extend_param]):
+         cond_params = {k: params[k] for k in fixed_params}
+         cond_params[extend_param] = extend_val
+         out_extended['cond%i' % i_cond] = cond_params
+
+         out_merged['%s(cond%i)' % (extend_param, i_cond)] = extend_val
+
+    return out_extended, out_merged
+
 if __name__=='__main__':
     print "Run nosetest.py"