Update other tests, which used implicit downcasting.

theano/compile/tests/test_shared.py

@@ -203,49 +203,52 @@ class Test_SharedVariable(unittest.TestCase):
         # should keep the same bit width / precision as the original value used to create the
         # shared variable.
         #
+
+        # Since downcasting of a value now raises an Exception,
+
         def f(var, val): var.value = val
 
-        b = shared(numpy.int64(7))
+        b = shared(numpy.int64(7), allow_downcast=True)
         assert b.type == theano.tensor.lscalar
         f(b,8.23)
         assert b.value==8
 
-        b = shared(numpy.int32(7))
+        b = shared(numpy.int32(7), allow_downcast=True)
         assert b.type == theano.tensor.iscalar
         f(b,8.23)
         assert b.value==8
 
-        b = shared(numpy.int16(7))
+        b = shared(numpy.int16(7), allow_downcast=True)
         assert b.type == theano.tensor.wscalar
         f(b,8.23)
         assert b.value==8
 
-        b = shared(numpy.int8(7))
+        b = shared(numpy.int8(7), allow_downcast=True)
         assert b.type == theano.tensor.bscalar
         f(b,8.23)
         assert b.value==8
 
-        b = shared(numpy.float64(7.234))
+        b = shared(numpy.float64(7.234), allow_downcast=True)
         assert b.type == theano.tensor.dscalar
         f(b,8)
         assert b.value==8
         
-        b = shared(numpy.float32(7.234))
+        b = shared(numpy.float32(7.234), allow_downcast=True)
         assert b.type == theano.tensor.fscalar
         f(b,8)
         assert b.value==8
 
-        b = shared(numpy.float(7.234))
+        b = shared(numpy.float(7.234), allow_downcast=True)
         assert b.type == theano.tensor.dscalar
         f(b,8)
         assert b.value==8
 
-        b = shared(7.234)
+        b = shared(7.234, allow_downcast=True)
         assert b.type == theano.tensor.dscalar
         f(b,8)
         assert b.value==8
 
-        c = shared(numpy.zeros((5,5), dtype='float32'))
+        c = shared(numpy.zeros((5,5), dtype='float32'), allow_downcast=True)
         self.failUnlessRaises(TypeError, f, b, numpy.random.rand(5,5))
 
 
@@ -253,32 +256,32 @@ class Test_SharedVariable(unittest.TestCase):
     def test_tensor_floatX(self):
         def f(var, val): var.value = val
 
-        b = shared(numpy.int64([7]))
+        b = shared(numpy.int64([7]), allow_downcast=True)
         assert b.type == theano.tensor.lvector
         f(b,[8.23])
         assert b.value == 8
 
-        b = shared(numpy.int32([7]))
+        b = shared(numpy.int32([7]), allow_downcast=True)
         assert b.type == theano.tensor.ivector
         f(b,[8.23])
         assert b.value == 8
 
-        b = shared(numpy.int16([7]))
+        b = shared(numpy.int16([7]), allow_downcast=True)
         assert b.type == theano.tensor.wvector
         f(b,[8.23])
         assert b.value == 8
 
-        b = shared(numpy.int8([7]))
+        b = shared(numpy.int8([7]), allow_downcast=True)
         assert b.type == theano.tensor.bvector
         f(b,[8.23])
         assert b.value == 8
 
-        b = shared(numpy.float64([7.234]))
+        b = shared(numpy.float64([7.234]), allow_downcast=True)
         assert b.type == theano.tensor.dvector
         f(b,[8])
         assert b.value == 8
 
-        b = shared(numpy.float32([7.234]))
+        b = shared(numpy.float32([7.234]), allow_downcast=True)
         assert b.type == theano.tensor.fvector
         f(b,[8])
         assert b.value == 8
@@ -293,12 +296,12 @@ class Test_SharedVariable(unittest.TestCase):
 #        assert b.type == theano.tensor.dvector
 #        f(b,[8])
 
-        b = shared(numpy.asarray([7.234],dtype=theano.config.floatX))
+        b = shared(numpy.asarray([7.234],dtype=theano.config.floatX), allow_downcast=True)
         assert b.dtype == theano.config.floatX
         f(b,[8])
         assert b.value == 8
 
-        c = shared(numpy.zeros((5,5), dtype='float32'))
+        c = shared(numpy.zeros((5,5), dtype='float32'), allow_downcast=True)
         self.failUnlessRaises(TypeError, f, b, numpy.random.rand(5,5))
 
 

theano/sandbox/cuda/tests/test_basic_ops.py

@@ -206,7 +206,7 @@ def test_elemwise_empty():
     f2 = pfunc([b], [], updates=[(a, a+b)], mode=mode_without_gpu)
 
     a0 = a.value * 1.0
-    f(numpy.ones((0,0)))
+    f(numpy.ones((0,0), dtype='float32'))
 
     assert numpy.all(a0 + 1.0 == a.value)
 
@@ -225,7 +225,7 @@ def test_elemwise0():
     print 'BEFORE ADD', a.value
     for i, node in enumerate(f.maker.env.toposort()):
         print i, node
-    f(numpy.ones((4,4)))
+    f(numpy.ones((4,4), dtype='float32'))
     print 'AFTER ADD', a.value
 
     assert numpy.all(a0 + 1.0 == a.value)
@@ -796,7 +796,7 @@ def test_duplicate_arg_elemwise():
 
     f = theano.function([A],B, mode = mode_with_gpu)
 
-    Aval = numpy.random.RandomState([1,2,3]).randn(5,5)
+    Aval = numpy.random.RandomState([1,2,3]).randn(5,5).astype('float32')
     Bval = Aval + Aval
 
     assert numpy.allclose(Bval,f(Aval))

theano/sandbox/cuda/tests/test_nnet.py

@@ -42,7 +42,8 @@ def test_GpuCrossentropySoftmaxArgmax1HotWithBias():
     dot_result = T.fmatrix('dot_result')
 
     xx = numpy.asarray(numpy.random.rand(batch_size,n_in),dtype=numpy.float32)
-    yy = numpy.ones((batch_size,),dtype='float32')
+    #?????yy = numpy.ones((batch_size,),dtype='float32')
+    yy = numpy.ones((batch_size,),dtype='int32')
     b_values = numpy.zeros((n_out,),dtype='float32')
     W_values = numpy.asarray(numpy.random.rand(n_in,n_out),dtype='float32')
     
@@ -82,7 +83,7 @@ def test_GpuCrossentropySoftmax1HotWithBiasDx():
     batch_size = 4097
     n_out = 1250
     
-    softmax_output_value = numpy.random.rand(batch_size, n_out)
+    softmax_output_value = numpy.random.rand(batch_size, n_out).astype('float32')
 
     softmax_output = T.fmatrix()
     softmax_output /= softmax_output.sum(axis=1).reshape(softmax_output.shape[1],1)

theano/sandbox/cuda/tests/test_opt.py

@@ -73,7 +73,7 @@ def test_softmax():
     f = theano.function([x],tensor.nnet.nnet.Softmax()(x), mode=mode_with_gpu)
     f2 = theano.function([x],tensor.nnet.nnet.Softmax()(x), mode=mode_without_gpu)
     assert isinstance(f.maker.env.toposort()[1].op,cuda.nnet.GpuSoftmax)
-    xv=numpy.random.rand(7,8)
+    xv=numpy.random.rand(7,8).astype('float32')
     assert numpy.allclose(f(xv),f2(xv))
 
 
@@ -84,8 +84,8 @@ def test_softmax_with_bias():
     f = theano.function([x,b],tensor.nnet.nnet.SoftmaxWithBias()(x,b), mode=mode_with_gpu)
     f2 = theano.function([x,b],tensor.nnet.nnet.SoftmaxWithBias()(x,b), mode=mode_without_gpu)
     assert isinstance(f.maker.env.toposort()[2].op,cuda.nnet.GpuSoftmaxWithBias)
-    xv=numpy.random.rand(7,8)
-    bv=numpy.random.rand(8)
+    xv=numpy.random.rand(7,8).astype('float32')
+    bv=numpy.random.rand(8).astype('float32')
     assert numpy.allclose(f(xv,bv),f2(xv,bv))
 
 def test_opt_gpujoin_onlyajoin():
@@ -153,7 +153,7 @@ def test_print_op():
     assert isinstance(topo[1].op, theano.printing.Print)
     assert isinstance(topo[2].op, cuda.GpuElemwise)
     assert topo[3].op == cuda.host_from_gpu
-    f(numpy.random.random((5,5)))
+    f(numpy.random.random((5,5)).astype('float32'))
 
 def test_elemwise_fusion():
     """ Test the the GpuElemwise fusion work correctly"""

theano/tensor/tests/test_basic.py

@@ -1654,48 +1654,48 @@ class test_comparison(unittest.TestCase):
     def test_gt(self):
         x, y = fvector(), fvector()
         fn = inplace_func([x,y], x > y)
-        l = numpy.asarray([0.,-1.,1.])
-        r = numpy.asarray([0.,1.,-1.])
+        l = numpy.asarray([0.,-1.,1.], dtype='float32')
+        r = numpy.asarray([0.,1.,-1.], dtype='float32')
         v = fn(l, r)
         self.failUnless(numpy.all(v == (l > r)), (v, (l>r)))
 
     def test_lt(self):
         x, y = fvector(), fvector()
         fn = inplace_func([x,y], x < y)
-        l = numpy.asarray([0.,-1.,1.])
-        r = numpy.asarray([0.,1.,-1.])
+        l = numpy.asarray([0.,-1.,1.], dtype='float32')
+        r = numpy.asarray([0.,1.,-1.], dtype='float32')
         v = fn(l, r)
         self.failUnless(numpy.all(v == (l < r)), (v, (l<r)))
 
     def test_le(self):
         x, y = fvector(), fvector()
         fn = inplace_func([x,y], x <= y)
-        l = numpy.asarray([0.,-1.,1.])
-        r = numpy.asarray([0.,1.,-1.])
+        l = numpy.asarray([0.,-1.,1.], dtype='float32')
+        r = numpy.asarray([0.,1.,-1.], dtype='float32')
         v = fn(l, r)
         self.failUnless(numpy.all(v == (l <= r)), (v, (l<=r)))
 
     def test_ge(self):
         x, y = fvector(), fvector()
         fn = inplace_func([x,y], x >= y)
-        l = numpy.asarray([0.,-1.,1.])
-        r = numpy.asarray([0.,1.,-1.])
+        l = numpy.asarray([0.,-1.,1.], dtype='float32')
+        r = numpy.asarray([0.,1.,-1.], dtype='float32')
         v = fn(l, r)
         self.failUnless(numpy.all(v == (l >= r)), (v, (l>=r)))
 
     def test_eq(self):
         x, y = fvector(), fvector()
         fn = inplace_func([x,y], eq(x,y))
-        l = numpy.asarray([0.,-1.,1.])
-        r = numpy.asarray([0.,1.,-1.])
+        l = numpy.asarray([0.,-1.,1.], dtype='float32')
+        r = numpy.asarray([0.,1.,-1.], dtype='float32')
         v = fn(l, r)
         self.failUnless(numpy.all(v == (l == r)), (v, (l==r)))
 
     def test_neq(self):
         x, y = fvector(), fvector()
         fn = inplace_func([x,y], neq(x, y))
-        l = numpy.asarray([0.,-1.,1.])
-        r = numpy.asarray([0.,1.,-1.])
+        l = numpy.asarray([0.,-1.,1.], dtype='float32')
+        r = numpy.asarray([0.,1.,-1.], dtype='float32')
         v = fn(l, r)
         self.failUnless(numpy.all(v == (l != r)), (v, (l!=r)))
 
@@ -2607,30 +2607,32 @@ class TestARange(unittest.TestCase):
         assert numpy.all(f(0,0,1) == numpy.arange(0,0,1))
 
     def test_float32(self):
-        """Test arange constructor, on integer outputs"""
+        """Test arange constructor, on float32 outputs"""
         start, stop, step = fscalars('start', 'stop', 'step')
         out = arange(start, stop, step)
         f = function([start, stop, step], out)
 
         assert out.dtype == start.type.dtype
-        assert numpy.all(f(0,5,1) == numpy.arange(0,5,1, dtype=start.type.dtype))
-        assert numpy.all(f(2,11,4) == numpy.arange(2,11,4, dtype=start.type.dtype))
-        assert numpy.all(f(-5,1.1,1.2) == numpy.arange(-5,1.1,1.2, dtype=start.type.dtype))
-        assert numpy.all(f(1.3,2,-2.1) == numpy.arange(1.3,2,-2.1, dtype=start.type.dtype))
-        assert numpy.all(f(10,2,2) == numpy.arange(10,2,2, dtype=start.type.dtype))
+        arg_vals = [ (0,5,1), (2,11,4), (-5,1.1,1.2), (1.3,2,-2.1), (10,2,2) ]
+        for arg_v in arg_vals:
+            start_v, stop_v, step_v = arg_v
+            start_v_, stop_v_, step_v_ = numpy.asarray(arg_v, dtype=start.type.dtype)
+            assert numpy.all(f(start_v_, stop_v_, step_v_) == \
+                    numpy.arange(start_v, stop_v, step_v, dtype=start.type.dtype))
 
     def test_float64(self):
-        """Test arange constructor, on integer outputs"""
+        """Test arange constructor, on float64 outputs"""
         start, stop, step = dscalars('start', 'stop', 'step')
         out = arange(start, stop, step)
         f = function([start, stop, step], out)
 
         assert out.dtype == start.type.dtype
-        assert numpy.all(f(0,5,1) == numpy.arange(0,5,1, dtype=start.type.dtype))
-        assert numpy.all(f(2,11,4) == numpy.arange(2,11,4, dtype=start.type.dtype))
-        assert numpy.all(f(-5,1.1,1.2) == numpy.arange(-5,1.1,1.2, dtype=start.type.dtype))
-        assert numpy.all(f(1.3,2,-2.1) == numpy.arange(1.3,2,-2.1, dtype=start.type.dtype))
-        assert numpy.all(f(10,2,2) == numpy.arange(10,2,2, dtype=start.type.dtype))
+        arg_vals = [ (0,5,1), (2,11,4), (-5,1.1,1.2), (1.3,2,-2.1), (10,2,2) ]
+        for arg_v in arg_vals:
+            start_v, stop_v, step_v = arg_v
+            start_v_, stop_v_, step_v_ = numpy.asarray(arg_v, dtype=start.type.dtype)
+            assert numpy.all(f(start_v_, stop_v_, step_v_) == \
+                    numpy.arange(start_v, stop_v, step_v, dtype=start.type.dtype))
 
     def test_default_step(self):
         """Test that arange constructor uses the correct default step"""
@@ -2669,9 +2671,10 @@ class TestARange(unittest.TestCase):
         ff = function([fstop], fout)
 
         assert fout.dtype == fstop.type.dtype
-        assert numpy.all(ff(0.2) == numpy.arange(0.2))
-        assert numpy.all(ff(-0.7) == numpy.arange(-0.7))
-        assert numpy.all(ff(8.5) == numpy.arange(8.5))
+        fstop_values = [0.2, -0.7, 8.5]
+        for fstop_v in fstop_values:
+            fstop_v32 = numpy.float32(fstop_v)
+            assert numpy.all(ff(fstop_v32) == numpy.arange(fstop_v))
 
     def test_upcast(self):
         """Test that arange compute output type adequately"""
@@ -2756,7 +2759,7 @@ class TestInversePermutation(unittest.TestCase):
 
         # Generate a random permutation
         rng = numpy.random.RandomState(utt.fetch_seed())
-        p_val = rng.permutation(10)
+        p_val = rng.permutation(10).astype('int32')
         inv_val = f_inverse(p_val)
 
         # Check that the inverse of the inverse is the original permutation
@@ -2774,7 +2777,8 @@ class TestInversePermutation(unittest.TestCase):
 
         rng = numpy.random.RandomState(utt.fetch_seed())
         # Generate 10 random permutations
-        p_val = numpy.asarray([rng.permutation(10) for i in range(7)])
+        p_val = numpy.asarray([rng.permutation(10) for i in range(7)],
+                              dtype='int32')
         inv_val = f_inverse(p_val)
 
         # Check that the inverse of the inverse is the original permutation list
@@ -2799,7 +2803,7 @@ class TestPermuteRowElements(unittest.TestCase):
 
         rng = numpy.random.RandomState(utt.fetch_seed())
         input_val = rng.uniform(size=(5,))
-        p_val = rng.permutation(5)
+        p_val = rng.permutation(5).astype('int32')
         out_val = permute(input_val, p_val)
 
         # Should be equivalent to advanced indexing
@@ -2821,7 +2825,7 @@ class TestPermuteRowElements(unittest.TestCase):
 
         rng = numpy.random.RandomState(utt.fetch_seed())
         input_val = rng.uniform(size=(3,5))
-        p_val = rng.permutation(5)
+        p_val = rng.permutation(5).astype('int32')
         out_val = permute(input_val, p_val)
 
         # The same permutation should be applied to every row of the input matrix.
@@ -2843,7 +2847,8 @@ class TestPermuteRowElements(unittest.TestCase):
 
         rng = numpy.random.RandomState(utt.fetch_seed())
         input_val = rng.uniform(size=(3,5))
-        p_val = numpy.asarray([rng.permutation(5) for i in range(3)])
+        p_val = numpy.asarray([rng.permutation(5) for i in range(3)],
+                              dtype='int32')
         out_val = permute(input_val, p_val)
 
         # Each row of p contains a permutation to apply to the corresponding
@@ -2867,7 +2872,7 @@ class TestPermuteRowElements(unittest.TestCase):
 
         rng = numpy.random.RandomState(utt.fetch_seed())
         input_val = rng.uniform(size=(5,))
-        p_val = numpy.asarray([rng.permutation(5) for i in range(3)])
+        p_val = numpy.asarray([rng.permutation(5) for i in range(3)], dtype='int32')
         out_val = permute(input_val, p_val)
 
         # Each row of p contains a permutation to apply to the input vector
@@ -2892,7 +2897,8 @@ class TestPermuteRowElements(unittest.TestCase):
 
         rng = numpy.random.RandomState(utt.fetch_seed())
         input_val = rng.uniform(size=(4,1,5))
-        p_val = numpy.asarray([rng.permutation(5) for i in range(3)])
+        p_val = numpy.asarray([rng.permutation(5) for i in range(3)],
+                              dtype='int32')
         out_val = permute(input_val, p_val)
 
         # Each row of p contains a permutation to apply to each row
@@ -3039,7 +3045,7 @@ class test_tensordot(unittest.TestCase):
         amat = fmatrix()
         bmat = dmatrix()#we let at float64 to test mix of float32 and float64.
         axes = 1
-        aval = self.rand(4,5)
+        aval = self.rand(4,5).astype('float32')
         bval = numpy.random.rand(5,3)
         c = tensordot(amat, bmat, axes)
         f3 = inplace_func([amat,bmat],c)

theano/tensor/tests/test_complex.py

@@ -23,7 +23,7 @@ class TestRealImag(unittest.TestCase):
 
 
         x= imatrix()
-        xval = numpy.asarray(rng.randn(3,3)*100, dtype='int')
+        xval = numpy.asarray(rng.randn(3,3)*100, dtype='int32')
         numpy.all( 0 == theano.function([x], imag(x))(xval))
         numpy.all( xval == theano.function([x], real(x))(xval))
 

theano/tensor/tests/test_gc.py

@@ -12,7 +12,7 @@ def test_no_reuse():
     f = theano.function([x, y], x + y)
 
     #provide both inputs in the first call
-    f(numpy.ones(10), numpy.ones(10))
+    f(numpy.ones(10, dtype='int64'), numpy.ones(10, dtype='int64'))
 
     try:
         f(numpy.ones(10))

theano/tensor/tests/test_misc.py

@@ -10,7 +10,7 @@ def test_bug_2009_06_02_trac_387():
     #f = theano.function([y], tensor.join(0,tensor.shape_padleft(y[0] / 2,1)))
     f = theano.function([y], tensor.int_div(tensor.DimShuffle(y[0].broadcastable, ['x'])(y[0]), 2))
     sys.stdout.flush()
-    print f(numpy.ones(1) * 3)
+    print f(numpy.ones(1, dtype='int64') * 3)
     #z = tensor.lscalar('z')
     #f = theano.function([z], tensor.DimShuffle([], ['x'])(z) / 2)
 

theano/tensor/tests/test_opt.py

@@ -2094,6 +2094,6 @@ def test_local_mul_to_neg():
     a = T.imatrix()
     f1 = theano.function([a], -1*a)
     f2 = theano.function([a], -1.0*a)
-    aval = numpy.random.randint(0,10,(2,2))
+    aval = numpy.random.randint(0,10,(2,2)).astype('int32')
     assert f1(aval).dtype == a.dtype
     assert f2(aval).dtype == 'float64'

theano/tensor/tests/test_randomstreams.py

@@ -581,8 +581,8 @@ class T_RandomStreams(unittest.TestCase):
         made = m.make()
         made.random.initialize()
 
-        low_val = [.1, .2, .3]
-        high_val = [1.1, 2.2, 3.3]
+        low_val = [100, 200, 300]
+        high_val = [110, 220, 330]
         seed_gen = numpy.random.RandomState(utt.fetch_seed())
         numpy_rng = numpy.random.RandomState(int(seed_gen.randint(2**30)))
 

theano/tensor/tests/test_raw_random.py

@@ -665,8 +665,8 @@ class T_random_function(unittest.TestCase):
         assert out.ndim == 1
         f = compile.function([rng_R, low, high], [post_r, out], accept_inplace=True)
 
-        low_val = [.1, .2, .3]
-        high_val = [1.1, 2.2, 3.3]
+        low_val = [100, 200, 300]
+        high_val = [110, 220, 330]
         rng = numpy.random.RandomState(utt.fetch_seed())
         numpy_rng = numpy.random.RandomState(utt.fetch_seed())
 

theano/tensor/tests/test_shared_randomstreams.py

@@ -527,8 +527,8 @@ class T_SharedRandomStreams(unittest.TestCase):
         assert out.ndim == 1
         f = function([low, high], out)
 
-        low_val = [.1, .2, .3]
-        high_val = [1.1, 2.2, 3.3]
+        low_val = [100, 200, 300]
+        high_val = [110, 220, 330]
         seed_gen = numpy.random.RandomState(utt.fetch_seed())
         numpy_rng = numpy.random.RandomState(int(seed_gen.randint(2**30)))