SciPy Linear Algebra - SciPy Linalg

Linear algebra deals with vectors, matrices and systems of linear equations. SciPy’s scipy.linalg module provides useful tools to perform various linear algebra operations such as solving equations, computing matrix decompositions, finding eigenvalues, matrix inverses etc. It is similar to NumPy’s linalg module but offers extended functionality and better performance for larger problems.

Linear Algebra Operations in SciPy

scipy.linalg module offers efficient functions for common linear algebra tasks like solving equations, inverting matrices, computing decompositions, eigenvalues all using NumPy arrays.

Let’s explore these operations one by one.

1. Solving Systems of Linear Equations

Solving systems of linear equations means finding variable values that satisfy all equations. SciPy’s linalg.solve() finds solution using coefficient matrix and output values.

Syntax:

scipy.linalg.solve(a, b)

Parameter:

• a: Coefficient matrix (square array).
• b: Right-hand side array (output values).

Example: This code solves two linear equations by finding values of variables using linalg.solve()

7x + 2y = 8,
4x + 5y = 10

Output

[0.74074074 1.40740741]

2. Matrix Inversion

Matrix inversion finds another matrix that, when multiplied with the original, gives the identity matrix. SciPy uses the linalg.inv() method to compute the inverse.

Syntax:

scipy.linalg.inv(a)

Parameter:a is square matrix to invert.

Example: This code computes inverse for below 2×2 matrix using scipy.linalg.inv():

Output

[[ 0.18518519 -0.07407407]
[-0.14814815 0.25925926]]

3. Pseudo-Inverse of a Matrix

pseudo-inverse (or Moore-Penrose inverse) is used to find solutions to systems of linear equations that may not have a unique or exact solution. In SciPy, it's computed using scipy.linalg.pinv().

Syntax:

scipy.linalg.pinv(a)

Parameter: a is an input matrix (can be rectangular).

Example: In the code below, scipy.linalg.pinv takes a matrix x and returns its Moore-Penrose pseudo inverse.

Output

[[ 0.14251208 -0.03381643 -0.03864734]
[-0.07487923 0.16183575 0.11352657]]

4. Matrix Determinant

The determinant is a scalar value that describes certain properties of a square matrix, such as whether it is invertible. In SciPy, scipy.linalg.det() is used to compute the determinant.

Syntax:

scipy.linalg.det(a)

Parameter:a is an input square matrix.

Example: The scipy.linalg.det takes a square matrix A and returns D, the determinant of A. The determinant is a specific property of the linear transformation of a matrix. The determinant of a 2×2 matrix is given by:

Output

21.0

From above Python code, the determinant is calculated as:

5. Singular Value Decomposition (SVD)

Singular Value Decomposition (SVD) breaks a matrix into three components: two orthogonal matrices and a diagonal matrix of singular values to make specific subsequent matrix calculations simpler. It is calculated using scipy.linalg.svd.

Syntax:

scipy.linalg.svd(a)

Parameter: a is an input matrix.

Example: This example performs Singular Value Decomposition (SVD) on a 2×2 matrix M. It breaks the matrix into three components:

• U (left singular vectors),
• s (singular values),
• V^H (right singular vectors).

Output

U = [[-0.38684104 -0.92214641]
[-0.92214641 0.38684104]]

Singular values = [12.62901571 1.58365469]

V^H = [[-0.46873958 -0.88333641]
[ 0.88333641 -0.46873958]]

6. Eigenvalues and Eigenvectors

Eigenvalues and eigenvectors help describe how a matrix transforms space.

Let M be an n × n matrix and X a non-zero vector. If M · X = λ · X for some number λ, then λ is called an eigenvalue and X is eigenvector.

Syntax:

scipy.linalg.eig(a, b=None)

Parameter:

• a: Input square matrix.
• b (Optional): Second matrix for generalized eigenvalue problems.

Example: This example calculates eigenvalues and eigenvectors of a 2×2 matrix using scipy.linalg.eig(). Eigenvalues indicate how the matrix scales vectors, while eigenvectors show directions that remain unchanged.

Output

Eigenvalues: [10.+0.j 3.+0.j]

Eigenvectors: [[ 0.9486833 -0.4472136 ]
[ 0.31622777 0.89442719]]

7. Matrix Norms

Matrix norms measure "size" or "length" of a matrix or vector. They help compare matrices, check stability or define distances. SciPy’s linalg.norm() function computes various norms like L1, L2 or Frobenius depending on the need.

Syntax:

scipy.linalg.norm(a, ord=None, axis=None, keepdims=False, check_finite=True)

Parameter:

• a: input array or matrix
• ord: type of norm (e.g., 1, 2, 'fro')

scipy.linalg.norm computes vector or matrix norms.

• L2 norm (Euclidean): Measures straight-line distance from the origin.
• L1 norm (Manhattan): Sum of absolute values like grid-based distance.

Example: This example demonstrates how to compute the L2 (Euclidean) and L1 (Manhattan) norms of a vector using scipy.linalg.norm().

Output

L2 norm: 6.708203932499369
L1 norm: 9.0

8. Other Matrix Functions

Other Matrix Functions in scipy.linalg provide tools for advanced matrix operations like computing matrix square roots, exponentials, and trigonometric functions used in solving differential equations. Below are most common matrix functions:

Let's explore these functions.

Example: This example demonstrates how to apply various matrix functions from scipy.linalg module to a 2x2 matrix.

Output

Matrix square root: [[ 2.49878019 0.62469505]
[15.61737619 3.90434405]]

Matrix exponential: [[2.49695022e+17 6.24237555e+16]
[1.56059389e+18 3.90148472e+17]]

Matrix sine: [[-0.06190153 -0.01547538]
[-0.38688456 -0.09672114]]

Matrix cosine: [[ 0.22445296 -0.19388676]
[-4.84716897 -0.21179224]]

Matrix tangent: [[0.0626953 0.01567382]
[0.39184561 0.0979614 ]]

Related Articles

• How to create a Sparse Matrix in Python
• How to reduce dimensionality on Sparse Matrix in Python
• Sparse Matrix Multiplication
• Spatial Distance Matrix
• Integration SciPy