added casting Ops

_test_tensor.py

@@ -586,6 +586,19 @@ class T_Shape(unittest.TestCase):
         s = shape(numpy.ones((5, 3, 10)))
         self.failUnless((eval_outputs([s]) == [5, 3, 10]).all())
 
+class T_Cast(unittest.TestCase):
+    def test_basic(self):
+        for type1 in ['int8', 'int16', 'int32', 'int64', 'float32', 'float64']:
+            x = Tensor(dtype = type1, broadcastable = (False, )).make_result()
+            for type2, converter in zip(['int8', 'int16', 'int32', 'int64', 'float32', 'float64'],
+                                        [convert_to_int8, convert_to_int16, convert_to_int32, convert_to_int64,
+                                         convert_to_float32, convert_to_float64]):
+                y = converter(x)
+                f = function([x], [y], strict = True, linker = 'c&py')
+                a = numpy.arange(10, dtype = type1)
+                b = f(a)
+                self.failUnless(numpy.all(b == numpy.arange(10, dtype = type2)))
+
 class T_argmax(unittest.TestCase):
     def setUp(self):
         numpy.random.seed(123784)
@@ -1626,5 +1639,5 @@ class _test_grad(unittest.TestCase):
 if __name__ == '__main__':
    unittest.main()
 #    suite = unittest.TestLoader()
-#    suite = suite.loadTestsFromTestCase(T_subtensor)
+#    suite = suite.loadTestsFromTestCase(T_Cast)
 #    unittest.TextTestRunner(verbosity=2).run(suite)

tensor.py

@@ -390,6 +390,19 @@ s2t.TensorConstant = TensorConstant
 s2t.TensorValue = TensorValue
 
 
+
+#########################
+# Utilities
+#########################
+
+def _elemwise(scalar_op, name):
+    straight = s2t.Elemwise(scalar_op)
+    inplace_scalar_op = scalar_op.__class__(scal.transfer_type(0))
+    inplace = s2t.Elemwise(inplace_scalar_op, {0: 0})
+    return straight, inplace
+
+
+
 #########################
 # Casting Operations
 #########################
@@ -421,6 +434,28 @@ class ScalarFromTensor(Op):
 scalar_from_tensor = ScalarFromTensor()
 
 
+def cast(t, dtype):
+    mapping = {'int8': convert_to_int8,
+               'int16': convert_to_int16,
+               'int32': convert_to_int32,
+               'int64': convert_to_int64,
+               'float32': convert_to_float32,
+               'float64': convert_to_float64,
+               'complex64': convert_to_complex64,
+               'complex128': convert_to_complex128}
+    return mapping[dtype](t)
+
+convert_to_int8  = s2t.Elemwise(scal.Identity(scal.specific_out(scal.int8)))
+convert_to_int16 = s2t.Elemwise(scal.Identity(scal.specific_out(scal.int16)))
+convert_to_int32 = s2t.Elemwise(scal.Identity(scal.specific_out(scal.int32)))
+convert_to_int64 = s2t.Elemwise(scal.Identity(scal.specific_out(scal.int64)))
+convert_to_float32 = s2t.Elemwise(scal.Identity(scal.specific_out(scal.float32)))
+convert_to_float64 = s2t.Elemwise(scal.Identity(scal.specific_out(scal.float64)))
+convert_to_complex64  = s2t.Elemwise(scal.Identity(scal.specific_out(scal.complex64)))
+convert_to_complex128 = s2t.Elemwise(scal.Identity(scal.specific_out(scal.complex128)))
+
+
+
 ##########################
 # Unary Operations
 ##########################
@@ -471,12 +506,6 @@ def max(x, axis=None):
     return argmax(x,axis)[0]
 
 
-def _elemwise(scalar_op, name):
-    straight = s2t.Elemwise(scalar_op)
-    inplace_scalar_op = scalar_op.__class__(scal.transfer_type(0))
-    inplace = s2t.Elemwise(inplace_scalar_op, {0: 0})
-    return straight, inplace
-
 ##########################
 # Comparison
 ##########################