theano - how to efficiently replicate and add tensors? -

i have 1 tensor in shape (2, g) , in shape (n, 2).

i need add them in such way output (n, 2, g), meaning first tensor replicated (n, 2, g) , second tensor added each matrix along third dimension. (or vice versa: second tensor replicated (n, 2, g) , first 1 added every sub-tensor along first dimension).

how can done efficiently in theano? thanks.

in attempt understand problem, following example assumed representative.

if

a = [[1, 2, 3],      [4, 5, 6]] 

and

b = [[1, 2],      [3, 4],      [5, 6],      [7, 8]] 

then result should be

c = [[[  2.   3.   4.]       [  6.   7.   8.]]      [[  4.   5.   6.]       [  8.   9.  10.]]      [[  6.   7.   8.]       [ 10.  11.  12.]]      [[  8.   9.  10.]       [ 12.  13.  14.]]] 

here g=3 , n=4.

to achieve in theano, 1 need add new broadcastable dimensions , rely on broadcasting desired result.

import numpy import theano import theano.tensor tt  x = tt.matrix() y = tt.matrix() z = x.dimshuffle('x', 0, 1) + y.dimshuffle(0, 1, 'x') f = theano.function([x, y], outputs=z) print f(numpy.array([[1, 2, 3], [4, 5, 6]]), numpy.array([[1, 2], [3, 4], [5, 6], [7, 8]]))