Is there any way to increase the size during memory sharing between process in PyTorch

Question

My current code is like this:

import torch
import torch.multiprocessing as mp

t = torch.zeros([10,10])
t.share_memory_()

processes = []

for i in range(3):
    p = mp.Process(target=some_function, args=(i, ))
    p.start()
    processes.append(p)

for p in processes:
    p.join()

I will increase t size in process[0], and hope that process[1] and process[2] could also receive this update. But I found that if just simply assign a new tensor:

# in process 0
temp = torch.zeros([20,10])
t = temp 
t.share_memory_()

The size of t in process 1 and process 2 are still the old 10x10.

I think it is some kind of memory and pointer issues, is there any possible way to solve it?