turing_21_matrix_operations 0.1.1

Turing college DE 2.1 coursework for Matrix operations: transform, window and cross-correlation using Python standard library.

Contents

• Overview
• Prerequisites
• Installation
• Functions
  • create_random_matrix
  • transpose2d
  • window1d
  • convolution2d
• Custom Generic Types
• Future Improvements

Overview

This library offers transpose2d, window1d and convolution2d functions to perform 2D matrix creation, transposition and cross-correlation, and 1D array windowing.These functions are particularly useful for machine learning applications.

Prerequisites

Python 3.10 or newer. Numpy 2.7 or newer.

Custom Generic Types

Within this library, there are three custom generic types to streamline function signatures and maintain type consistency, which can be enforced by MyPy:

• Matrix: Represents a 2D list of floats and is bound to List[List[float]].
• Array1D: Represents a 1D array, which can either be a Python list or a 1D numpy array. It is defined as TypeVar("Array1D", List[float], np.ndarray).
• Matrix2D: Represents a 2D numpy array and is bound to np.ndarray.

Installation

Package can be reached manually at PyPi.

The library can be installed directly from PyPi:

pip install turing-21-matrix-operations

Functions

Create Random Matrix

The create_random_matrix function generates a matrix filled with random numbers between 0 and 10.

Arguments

• rows (int): Number of rows in the matrix.
• columns (int): Number of columns in the matrix.

Returns

• Matrix: 2D list of random floats.

Usage:

from turing-21-matrix-operations import create_random_matrix
matrix = create_random_matrix(3, 4)
print(matrix)

Transpose2D

The transpose2d function is designed to transpose a 2D matrix by swapping its rows and columns.

Arguments

• input_matrix (Matrix): 2D list to be transposed.

Returns

• Matrix: Transposed 2D list.

from turing-21-matrix-operations import transpose2d
matrix = [[1, 2], [3, 4], [5, 6]]
transposed = transpose2d(input_matrix=matrix)
print(transposed)

Window1D

The window1d function creates overlapping windows of a given size from a 1D list or numpy array.

Arguments

• input_array (Array1D): 1D list or numpy array.
• size (int): Desired window size.
• shift (int, optional): Window shift size. Default is 1.
• stride (int, optional): Window stride size. Default is 1.

Returns

• List[Array1D]: List of 1D windows.

from turing-21-matrix-operations import window1d
input_array = [1, 2, 3, 4, 5, 6, 7, 8, 9, 10]
windows = window1d(input_array=input_array, size=2, shift=2, stride=2)
print(windows)

Convolution2D

The convolution2d function performs 2D convolution using a given kernel.

Arguments

• input_matrix (Matrix2D): 2D numpy array input matrix.
• kernel (Matrix2D): 2D numpy array convolution kernel.
• stride (int, optional): Convolution stride. Default is 1. Returns
• Matrix2D: Convolved 2D numpy array.

from turing-21-matrix-operations import convolution2d
input_matrix = np.array([[1, 2, 3], [4, 5, 6], [7, 8, 9]])
kernel = np.array([[1, 0], [0, -1]])
convolved = convolution2d(input_matrix=input_matrix, kernel=kernel)
print(convolved)