Sparse diagonal matrix

November 8, 2022

---

The diags function from SciPy can be used to create a sparse diagonal matrix. The matrix is constructed from lists or NumPy arrays that represent the diagonals of the matrix. The offsets argument sets the placement of the diagonals within the matrix where k = 0 is the main diagonal, k < 0 is the kth lower diagonal, and k > 0 is the kth upper diagonal.

from scipy.sparse import diags

# Diagonals as lists or NumPy arrays
a = [33, 33, 33, 33]
b = [44, 44, 44, 44, 44]
c = [55, 55, 55, 55]

# Create a sparse diagonal matrix
z = diags([a, b, c], offsets=[1, 0, -1]).toarray()

Where z is a NumPy array containing the diagonals

>>> z
array([[44, 33,  0,  0,  0],
       [55, 44, 33,  0,  0],
       [ 0, 55, 44, 33,  0],
       [ 0,  0, 55, 44, 33],
       [ 0,  0,  0, 55, 44]])