tensor_inc_subtensor_test: added numerical tests, alloc test not finished

theano/tensor/tests/test_opt.py

@@ -1946,6 +1946,11 @@ def test_subtensor_inc_subtensor():
     prog = f.maker.fgraph.toposort()
     assert len(prog) == 1
     assert isinstance(prog[0].op, DeepCopyOp)
+    # basic test, numerical check
+    x_ = numpy.random.uniform(size=[3, 4])
+    v_ = numpy.random.uniform(size=[4,])
+    i_ = 1
+    assert f(x_, i_, v_) == v_
 
     # complicated test
     x = tensor.tensor4('x')
@@ -1961,6 +1966,11 @@ def test_subtensor_inc_subtensor():
     prog = f.maker.fgraph.toposort()
     assert len(prog) == 1
     assert isinstance(prog[0].op, DeepCopyOp)
+    # complicated test, numerical check
+    x_ = numpy.random.uniform(size=[3, 4, 5, 6])
+    v_ = numpy.random.uniform(size=[2, 3, 2])
+    i1_, i2_, i3_, i4_ = 1, 2, 3, 4
+    assert f(x_, i1_, i2_, i3_, i4_, v_) == v_
 
     # case not use this optimization
     z = y[i1, :i3, i2:, ::i4]
@@ -1970,6 +1980,31 @@ def test_subtensor_inc_subtensor():
     assert len(prog) != 1
     assert any(isinstance(x.op, tensor.IncSubtensor) for x in prog)
     assert any(isinstance(x.op, tensor.Subtensor) for x in prog)
+    # case not use this optimization, numerical check
+    x_ = numpy.random.uniform(size=[3, 4, 5, 6])
+    v_ = numpy.random.uniform(size=[2, 3, 2])
+    i1_, i2_, i3_, i4_ = 1, 2, 3, 4
+    x_[i1, :i2, i3:, ::i4] = v_
+    assert f(x_, i1_, i2_, i3_, i4_, v_) == x_[i1, :i3, i2:, ::i4]
+
+    # case when v is broadcastable
+    x = tensor.matrix('x')
+    i1 = tensor.iscalar('i')
+    i2 = tensor.iscalar('i')
+    v = tensor.vector('v')
+    y = tensor.set_subtensor(x[:i1, :i2], v)
+    z = y[:i1, :i2]
+    mode = theano.compile.mode.get_default_mode().including('local_subtensor_inc_subtensor')
+    f = theano.function([x, i1, i2, v], z, mode=mode)
+    prog = f.maker.fgraph.toposort()
+    import pdb
+    pdb.set_trace()
+    assert len(prog) == 1
+    # basic test, numerical check
+    x_ = numpy.random.uniform(size=[3, 4])
+    v_ = numpy.random.uniform(size=[2,])
+    i1_, i2_ = 1, 2
+    assert f(x_, i_, v_) == v_
 
 
 class test_local_subtensor_make_vector(unittest.TestCase):