BUG FIX: the stabilization optimization of the grad of log(erfc(x)) was applied…

theano/tensor/opt.py

@@ -2611,6 +2611,7 @@ def local_grad_log_erfc_neg(node):
     if not node.inputs[1].owner or node.inputs[1].owner.op != T.erfc:
         return False
     erfc = node.inputs[1]
+    erfc_x = erfc.owner.inputs[0]
     if not node.inputs[0].owner:
         return False
 
@@ -2642,6 +2643,9 @@ def local_grad_log_erfc_neg(node):
         #We use that flag to don't apply the optimization recursively
         return False
 
+    if erfc_x is not x: 
+        return False
+
     #we move the cst outside the div.
     true_div_no_mul = T.true_div(exp,erfc)
     true_div_no_mul.owner.tag.local_grad_log_erfc_neg=True

theano/tensor/tests/test_opt.py

@@ -1765,6 +1765,7 @@ class T_local_erfc(unittest.TestCase):
             val.remove(10)
         val = numpy.asarray(val)
         x = T.vector()
+        y = T.vector()
 
         #their is some nan that will happear in the graph for the log of the negatives values
         mode = copy.copy(self.mode)
@@ -1791,7 +1792,23 @@ class T_local_erfc(unittest.TestCase):
 
         #test that we work without the mul
         f = theano.function([x],T.exp(T.neg(T.sqr(x)))/T.erfc(x), mode=mode)
-        theano.printing.debugprint(f)
+        #theano.printing.debugprint(f)
+        assert len(f.maker.env.nodes)==21, len(f.maker.env.nodes)
+        assert f.maker.env.outputs[0].dtype==theano.config.floatX
+        assert not any([hasattr(n.op,'scalar_op') and n.op.scalar_op==scal.pow for n in f.maker.env.nodes])
+        assert all(numpy.isfinite(f(val)))
+
+        #test that we don't work if x!=y
+        f = theano.function([x,y],T.exp(T.neg(T.sqr(x)))/T.erfc(y), mode=mode)
+        #theano.printing.debugprint(f)
+        assert len(f.maker.env.nodes)==5, len(f.maker.env.nodes)
+        assert f.maker.env.outputs[0].dtype==theano.config.floatX
+        assert not any([hasattr(n.op,'scalar_op') and n.op.scalar_op==scal.pow for n in f.maker.env.nodes])
+        f(val,val-3)
+
+        #test that we work without the sqr and neg
+        f = theano.function([x],T.exp(T.mul(-1,x,x))/T.erfc(x), mode=mode)
+        #theano.printing.debugprint(f)
         assert len(f.maker.env.nodes)==21, len(f.maker.env.nodes)
         assert f.maker.env.outputs[0].dtype==theano.config.floatX
         assert not any([hasattr(n.op,'scalar_op') and n.op.scalar_op==scal.pow for n in f.maker.env.nodes])