Random Block Solution

---

import torch
import random

random.seed(123)

foo = torch.zeros(size=(10,10), dtype=torch.int32)
i, j = random.randint(0, 7), random.randint(0, 7)
foo[i:(i+3), j:(j+3)] = 1

Explanation

1. Instantiate a 10x10 tensor of 32-bit integer 0s.

   foo = torch.zeros(size=(10,10), dtype=torch.int32)
   
   print(foo)
   # tensor([[0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
   #         [0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
   #         [0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
   #         [0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
   #         [0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
   #         [0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
   #         [0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
   #         [0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
   #         [0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
   #         [0, 0, 0, 0, 0, 0, 0, 0, 0, 0]], dtype=torch.int32)
   

   By default, torch.zeros() creates floats, so we explicitly tell it to use 32-bit integers with dtype=torch.int32.

2. Randomly choose the top-left cell of the 3x3 block.

   We choose random a random (i,j) element such that the entire 3x3 block will fit inside foo.

   import random
   
   random.seed(123)
   i, j = random.randint(0, 7), random.randint(0, 7)
   print(i)  # 0
   print(j)  # 4
   

3. Select the 3x3 block and update 0s to 1s.

   foo[i:(i+3), j:(j+3)] = 1
   
   print(foo)
   # tensor([[0, 0, 0, 0, 1, 1, 1, 0, 0, 0],
   #         [0, 0, 0, 0, 1, 1, 1, 0, 0, 0],
   #         [0, 0, 0, 0, 1, 1, 1, 0, 0, 0],
   #         [0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
   #         [0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
   #         [0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
   #         [0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
   #         [0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
   #         [0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
   #         [0, 0, 0, 0, 0, 0, 0, 0, 0, 0]], dtype=torch.int32)
   

Bonus
We can plot the array as an image using matplotlib.pyplot.imshow() with cmap='gray'.

import matplotlib.pyplot as plt
plt.imshow(foo, cmap='gray')

---

Back Next