added file new_ops where we should put ops that are not in the core of theano.…

new_ops.py

+## This file contain new ops that are not currently integrated in the core of threano. 
+## Not all of those ops have been throughtfully tested.
+
+from theano.scalar import *
+import theano.tensor as T
+import theano
+from numpy import *
+class Prepend_scalar_constant_to_each_row(Op):
+    def __init__(self, val = 0):
+        if isinstance(val, float):
+            val = constant(val)
+        self.val = val
+
+    def make_node(self, mat):
+        #check type of input
+        if not isinstance(mat,Result) or not mat.type==T.matrix().type:
+            raise TypeError("Expected a matrix as input")
+        x = T.as_tensor(mat)
+        y = T.as_tensor(self.val)
+        if x.type.dtype != y.type.dtype:
+            TypeError("the value to prepend don't have the same type as the matrix")
+        
+        node = Apply(op=self, inputs=[mat], outputs=[T.matrix()])
+        return node
+
+    def perform(self, node, (mat, ), (output, )):
+        if output[0] == None:
+            output[0]=numpy.empty((mat.shape[0],mat.shape[1]+1),dtype=mat.dtype)
+            out=output[0]
+        else:
+            out=output[0]
+            assert out.shape==(mat.shape[0],mat.shape[1]+1)
+
+        out[:,0].fill(self.val.data)
+        out[:,1:]=mat
+
+    def grad(self, (mat,), (goutput,)):
+        return goutput[:,1:]
+
+class Prepend_scalar_to_each_row(Op):        
+    def make_node(self, val, mat):
+        #check type of input
+        if isinstance(val, float):
+            val = constant(val)
+        if not isinstance(mat,Result) or not mat.type==T.matrix().type:
+            raise TypeError("Expected a matrix as input")
+        x = T.as_tensor(mat)
+        y = T.as_tensor(val)
+        if x.type.dtype != y.type.dtype:
+            TypeError("the value to prepend don't have the same type as the matrix")
+        
+        node = Apply(op=self, inputs=[val,mat], outputs=[T.matrix()])
+        return node
+
+    def perform(self, node, (val,mat), (output, )):
+        if output[0] == None:
+            output[0]=numpy.empty((mat.shape[0],mat.shape[1]+1),dtype=mat.dtype)
+            out=output[0]
+        else:
+            out=output[0]
+            assert out.shape==(mat.shape[0],mat.shape[1]+1)
+        out[:,0].fill(val)
+        out[:,1:]=mat
+
+    def grad(self, (val, mat), (goutput,)):
+        return goutput[:,0], goutput[:,1:]
+
+if __name__ == '__main__':
+
+    x=T.matrix('x')
+    y=Prepend_scalar_constant_to_each_row(4.)(x)
+    f=theano.function([x],[y])
+    mat=numpy.random.rand(3,5)
+    gradient(f(mat))
+    print f(mat)
+
+    x=T.matrix('x')
+    y=Prepend_scalar_to_each_row()(5.,x)
+    f=theano.function([x],[y])
+    mat=numpy.ones((3,5),dtype="float32")
+    print f(mat)
+