def test_grad(self):
        x = T.vector('x')
        a = np.random.random(50).astype(config.floatX)

        theano.function([x], T.grad(T.sum(diff(x)), x))
        utt.verify_grad(self.op, [a])

        for k in range(TestDiffOp.nb):
            theano.function([x], T.grad(T.sum(diff(x, n=k)), x))
            utt.verify_grad(DiffOp(n=k), [a], eps=7e-3)

    def test_diffOp(self):
        x = T.matrix('x')
        a = np.random.random((30, 50)).astype(config.floatX)

        f = theano.function([x], diff(x))
        assert np.allclose(np.diff(a), f(a))

        for axis in range(len(a.shape)):
            for k in range(TestDiffOp.nb):
                g = theano.function([x], diff(x, n=k, axis=axis))
                assert np.allclose(np.diff(a, n=k, axis=axis), g(a))

    def test_infer_shape(self):
        x = T.matrix("x")
        a = np.random.random((30, 50)).astype(config.floatX)

        self._compile_and_check([x], [self.op(x)], [a], self.op_class)

        for axis in range(len(a.shape)):
            for k in range(TestDiffOp.nb):
                self._compile_and_check([x], [diff(x, n=k, axis=axis)], [a], self.op_class)

 def toChw(self, position):
     samples, targetDim = K.shape(position)
     position = K.reshape(position, (samples, 2, 2))
     centroid = K.sum(position, axis=1) / 2.0
     hw = K.abs(THEO.diff(position, axis=1)[:,0,:])
     chw = K.concatenate((centroid, hw), axis=1)
     chw = chw[:, [0, 1, 3, 2]] # Changing from cwh to chw
     
     return chw