如何更改阻止循环的theano张量中的子集值？

Question

layer_Fmaps of size (1,96,236,236) sitwches of size (1,96,708,708)

in each 3 by 3 matrix in layer_Fmaps there is only cell having a value 1 which should be replace by the opposite value of sitwches

I can't find a way to solve the problem by assigning a direct value into a certain location using a loop

def switchs(layer_Fmaps, step=2, switches):
        for idx in range(96):
            for i in range(0, 708, step):
                for j in range(0, 708, step):
                    val = layer_Fmaps[0][idx][i/2,j/2]
                    switches = T.set_subtensor(switches[0][idx][i:i + step, j:j + step],val)
        return  switches

knowing that switchs and layer_Fmaps are tensor4

img = np.zeros((1,96,236,236))
sswitchs =  np.zeros((1,96,708,708))

inp = T.tensor4('img')
SW = T.tensor4('SW')

tester = switchs(inp,3,SW)

f = theano.function([inp, SW], tester)    

d = f(img,sswitchs)

Any suggestion would be appreciated.

Answer 1

def switchs(layer_Fmaps, switches, step=2):
     for idx in range(96):     
         for i in range(0, 708, step):
             for j in range(0, 708, step):
                 val = layer_Fmaps[0,idx,i/2,j/2]
                 switches = T.set_subtensor(switches[0, idx, i:i + step, j:j + step],val)
     return  switches

And:

img = np.zeros((1,96,236,236))
sswitchs =  np.zeros((1,96,708,708))

inp = T.tensor4('img')
SW = T.tensor4('SW')

tester = switchs(inp,SW, 3)

f = theano.function([inp, SW], tester)    

d = f(img,sswitchs)

I modified several things:

• Replaced 96 for 69. I guess the "96" was a typo.
• Uniform indexing
• Key arguments should be at the end.