y = lab.map(x,f) returns a new Tensor f(x): f() is applied successively to each element of x. 

> x = lab.rand(5,5)
> print(x)

 0.2828  0.2709  0.0191  0.1217  0.0161
 0.4847  0.6586  0.3027  0.6749  0.4945
 0.2003  0.9536  0.3208  0.9612  0.9265
 0.5415  0.1827  0.6799  0.3222  0.7285
 0.7835  0.5920  0.8516  0.2312  0.1857
[torch.Tensor of dimension 5x5]

> = lab.map(x, math.cos)

 0.9603  0.9635  0.9998  0.9926  0.9999
 0.8848  0.7909  0.9545  0.7807  0.8802
 0.9800  0.5788  0.9490  0.5725  0.6007
 0.8569  0.9834  0.7776  0.9486  0.7462
 0.7084  0.8298  0.6588  0.9734  0.9828
[torch.Tensor of dimension 5x5]

f must return a number or nil. If it returns nil, the corresponding value in the returned Tensor will be 0. This behavior is slightly different from the behavior of map functions provided in the torch package, which do not modify the destination tensor if the function returns nil. 

> x = lab.rand(5,5)
> print(x)

 0.0446  0.5119  0.4007  0.9743  0.0957
 0.4649  0.7575  0.4963  0.5474  0.5009
 0.5523  0.8003  0.2841  0.0048  0.9853
 0.5080  0.1222  0.3049  0.8647  0.2566
 0.6863  0.8243  0.3053  0.9196  0.6050
[torch.Tensor of dimension 5x5]

> = lab.map(x, function(z)
>> if z > 0.5 then return 1 end -- nothing returned otherwise!
>> end
>> )

 0  1  0  1  0
 0  1  0  1  1
 1  1  0  0  1
 1  0  0  1  0
 1  1  0  1  1
[torch.Tensor of dimension 5x5]