MatrixKit: Synthetic Matrix Generation for Machine Learning and Scientific Computing

Link to the initial project repository

GitHub repo "opencampus-preconditioner-ai-project"

Overview

MatrixKit is a sophisticated Python library designed for generating synthetic matrix data, primarily focused on machine learning applications. It was created as part of a machine learning project at OpenCampus Kiel, where my project partner and I faced the challenge of finding labelled real-world matrices to train our models. MatrixKit offers powerful tools for creating custom matrices that simulate real-world data structures and patterns.

Additionally, the library contains a variety of functions to create and apply block jacobi preconditioners.

Features

• Flexible Matrix Generation: Create matrices of various sizes and shapes with customizable properties.
• Realistic Noise Simulation: Add controlled background noise to matrices.
• Complex Block Structures: Generate matrices with intricate block patterns using truncated normal distributions.
• Fine-Tuned Control: Adjust parameters like matrix dimensions, noise levels, block sizes, and densities.
• Comprehensive Metadata: Maintain detailed information about generated matrices, including block positions and user-defined parameters.
• Versatile Applications: Suitable for machine learning, data analysis, scientific computing, and more.

Installation

Install MatrixKit easily using pip:

pip install matrixkit