added care wrt non-int,non-float scalar types

scalar.py

@@ -58,14 +58,15 @@ class Scalar(Result):
 
     def dtype_specs(self):
         try:
-            return {'float32': (float, 'npy_float32', 'PyFloat_Check', 'PyFloat_AsDouble', 'PyFloat_FromDouble'),
-                    'float64': (float, 'npy_float64', 'PyFloat_Check', 'PyFloat_AsDouble', 'PyFloat_FromDouble'),
-                    'int8': (int, 'npy_int8', 'PyInt_Check', 'PyInt_AsLong', 'PyInt_FromLong'),
-                    'int16': (int, 'npy_int16', 'PyInt_Check', 'PyInt_AsLong', 'PyInt_FromLong'),
-                    'int32': (int, 'npy_int32', 'PyInt_Check', 'PyInt_AsLong', 'PyInt_FromLong'),
-                    'int64': (int, 'npy_int64', 'PyInt_Check', 'PyInt_AsLong', 'PyInt_FromLong'),
-                    'complex128': (complex, 'theano_complex128', 'PyComplex_Check', 'PyComplex_AsCComplex', 'PyComplex_FromCComplex'),
-                    'complex64': (complex, 'theano_complex64', None, None, None)}[self.dtype]
+            return {'float32': (numpy.float32, 'npy_float32', 'PyFloat_Check', 'PyFloat_AsDouble', 'PyFloat_FromDouble'),
+                    'float64': (numpy.float64, 'npy_float64', 'PyFloat_Check', 'PyFloat_AsDouble', 'PyFloat_FromDouble'),
+                    'complex128': (numpy.complex128, 'theano_complex128', 'PyComplex_Check', 'PyComplex_AsCComplex', 'PyComplex_FromCComplex'),
+                    'complex64': (numpy.complex64, 'theano_complex64', None, None, None),
+                    'int8':  (numpy.int8, 'npy_int8', 'PyInt_Check', 'PyInt_AsLong', 'PyInt_FromLong'),
+                    'int16': (numpy.int16, 'npy_int16', 'PyInt_Check', 'PyInt_AsLong', 'PyInt_FromLong'),
+                    'int32': (numpy.int32, 'npy_int32', 'PyInt_Check', 'PyInt_AsLong', 'PyInt_FromLong'),
+                    'int64': (numpy.int64, 'npy_int64', 'PyInt_Check', 'PyInt_AsLong', 'PyInt_FromLong')
+                    }[self.dtype]
         except KeyError:
             raise TypeError("Unsupported dtype for %s: %s" % (self.__class__.__name__, self.dtype))
 
@@ -148,9 +149,7 @@ class Scalar(Result):
         return template % dict(nbits = 64, half_nbits = 32) + template % dict(nbits = 128, half_nbits = 64)
 
     def __copy__(self):
-        """
-        Return a copy of this instance (with its own attributes)
-        """
+        """Return a copy of this instance (with its own attributes)"""
         cpy = self.__class__(self.dtype, self.name)
         cpy.data = self.data
         return cpy
@@ -207,11 +206,16 @@ class ScalarOp(GuardedOp):
         
         inputs = [as_scalar(input) for input in inputs]
         i_dtypes = [getattr(input, 'dtype', None) for input in inputs]
-        o_dtypes = [upcast(*i_dtypes)] * self.nout
+        o_dtypes = self.output_dtypes(*i_dtypes)
 
         self.inputs = inputs
         self.outputs = [Scalar(dtype) for dtype in o_dtypes]
 
+    def output_dtypes(self, *dtypes):
+        if self.nout != 1:
+            raise NotImplementedError()
+        return upcast(*dtypes),
+
     def impl(self, *inputs):
         raise AbstractFunctionError()
     
@@ -232,6 +236,13 @@ class UnaryScalarOp(ScalarOp):
 class BinaryScalarOp(ScalarOp):
     nin = 2
 
+class FloatUnaryScalarOp(UnaryScalarOp):
+    def output_dtypes(self, input_dtype):
+        if 'int' in input_dtype: return 'float64',
+        if 'float' in input_dtype: return input_dtype,
+        raise NotImplementedError()
+
+
 
 class Add(ScalarOp):
     identity = 0
@@ -318,6 +329,7 @@ class Neg(UnaryScalarOp):
         return "%(z)s = -%(x)s;" % locals()
 
 class Abs(UnaryScalarOp):
+    #TODO: for complex input, output is some flavour of float
     def impl(self, x):
         return numpy.abs(x)
     def grad(self, (x, ), (gz, )):
@@ -333,22 +345,24 @@ class Abs(UnaryScalarOp):
 
 class Sgn(UnaryScalarOp):
     def impl(self, x):
-        return numpy.abs(x) / x
+        #casting to output type is handled by filter
+        return 1.0 if x >= 0 else -1.0
     def grad(self, (x, ), (gz, )):
         return None,
     def c_code(self, (x, ), (z, ), sub):
-        return "%(z)s = %(x)s/%(prefix)sabs(%(x)s);" \
-            % dict(locals(), prefix = 'float' in self.inputs[0].dtype and 'f' or '') # TODO: C use copysign
+        #casting is done by compiler
+        #TODO: use copysign
+        return "%(z)s = (%(x)s >= 0) ? 1.0 : -1.0;" % locals()
 
-class Inv(UnaryScalarOp):
+class Inv(FloatUnaryScalarOp):
     def impl(self, x):
-        return 1 / x
+        return 1.0 / x
     def grad(self, (x, ), (gz, )):
         return -gz / (x * x),
     def c_code(self, (x, ), (z, ), sub):
-        return "%(z)s = 1 / %(x)s;" % locals()
+        return "%(z)s = 1.0 / %(x)s;" % locals()
 
-class Log(UnaryScalarOp):
+class Log(FloatUnaryScalarOp):
     def impl(self, x):
         return math.log(x)
     def grad(self, (x, ), (gz, )):
@@ -356,7 +370,7 @@ class Log(UnaryScalarOp):
     def c_code(self, (x, ), (z, ), sub):
         return "%(z)s = log(%(x)s);" % locals()
 
-class Log2(UnaryScalarOp):
+class Log2(FloatUnaryScalarOp):
     def impl(self, x):
         return numpy.log2(x)
     def grad(self, (x, ), (gz, )):
@@ -364,7 +378,7 @@ class Log2(UnaryScalarOp):
     def c_code(self, (x, ), (z, ), sub):
         return "%(z)s = log2(%(x)s);" % locals()
 
-class Exp(UnaryScalarOp):
+class Exp(FloatUnaryScalarOp):
     def impl(self, x):
         return math.exp(x)
     def grad(self, (x, ), (gz, )):
@@ -380,7 +394,7 @@ class Sqr(UnaryScalarOp):
     def c_code(self, (x, ), (z, ), sub):
         return "%(z)s = %(x)s * %(x)s;" % locals()
 
-class Sqrt(UnaryScalarOp):
+class Sqrt(FloatUnaryScalarOp):
     def impl(self, x):
         return math.sqrt(x)
     def grad(self, (x, ), (gz, )):
@@ -388,7 +402,7 @@ class Sqrt(UnaryScalarOp):
     def c_code(self, (x, ), (z, ), sub):
         return "%(z)s = sqrt(%(x)s);" % locals()
 
-class Cos(UnaryScalarOp):
+class Cos(FloatUnaryScalarOp):
     def impl(self, x):
         return math.cos(x)
     def grad(self, (x, ), (gz, )):
@@ -396,7 +410,7 @@ class Cos(UnaryScalarOp):
     def c_code(self, (x, ), (z, ), sub):
         return "%(z)s = cos(%(x)s);" % locals()
 
-class Sin(UnaryScalarOp):
+class Sin(FloatUnaryScalarOp):
     def impl(self, x):
         return math.sin(x)
     def grad(self, (x, ), (gz, )):
@@ -404,15 +418,15 @@ class Sin(UnaryScalarOp):
     def c_code(self, (x, ), (z, ), sub):
         return "%(z)s = sin(%(x)s);" % locals()
 
-class Tan(UnaryScalarOp):
+class Tan(FloatUnaryScalarOp):
     def impl(self, x):
         return math.tan(x)
     def grad(self, (x, ), (gz, )):
-        raise NotImplementedError('lazy')
+        raise NotImplementedError()
     def c_code(self, (x, ), (z, ), sub):
         return "%(z)s = tan(%(x)s);" % locals()
 
-class Cosh(UnaryScalarOp):
+class Cosh(FloatUnaryScalarOp):
     def impl(self, x):
         return math.cosh(x)
     def grad(self, (x, ), (gz, )):
@@ -420,15 +434,15 @@ class Cosh(UnaryScalarOp):
     def c_code(self, (x, ), (z, ), sub):
         return "%(z)s = cosh(%(x)s);" % locals()
 
-class Sinh(UnaryScalarOp):
+class Sinh(FloatUnaryScalarOp):
     def impl(self, x):
-        return math.sin(x)
+        return math.sinh(x)
     def grad(self, (x, ), (gz, )):
-        return -gz * cos(x),
+        raise NotImplementedError()
     def c_code(self, (x, ), (z, ), sub):
         return "%(z)s = sin(%(x)s);" % locals()
 
-class Tanh(UnaryScalarOp):
+class Tanh(FloatUnaryScalarOp):
     def impl(self, x):
         return math.tanh(x)
     def grad(self, (x, ), (gz, )):
@@ -496,9 +510,6 @@ def composite(inputs, outputs):
                 i += 1
                 name = "V%%(id)s_tmp%i" % i
                 subd[output] = name
-                # the c code is not robust to any other dtypes than those of the specified inputs
-                # a solution would be to require Composite.c_code to fill in the dtypes using
-                # a proper upcast
                 _c_code += "%s %s;\n" % (output.dtype_specs()[1], name)
         _c_code += op.c_code([subd[input] for input in op.inputs],
                              [subd[output] for output in op.outputs],
@@ -529,7 +540,9 @@ def composite(inputs, outputs):
         nin = len(inputs)
         nout = len(outputs)
 
-        # todo: propagate_dtypes?
+        def output_dtypes(self, *input_dtypes):
+            assert input_dtypes == tuple([input.dtype for input in inputs])
+            return [output.dtype for dtype in outputs]
 
         def perform(self):
             inputs = [input.data for input in self.inputs]