← BACK TO PROJECTS

Sparse Matrix Research

C++ Algorithm Design HPC Memory Optimization

Project Overview

Research conducted at XIM University. I identified bottlenecks in traditional sparse matrix storage formats (CSR, CDS) when dealing with diagonally dominant matrices. I invented a new format, 'Contiguous Clustering' (CC), and wrote a custom C++ engine to benchmark it, proving massive gains in SpMV (Sparse Matrix-Vector Multiplication).

Engineering Architecture

// System Flow

1. User Input → Validation Layer (Pydantic)

2. Core Processing → AI Inference (C++)

3. Data Persistence → Database / File System

4. Response → Client UI

Key Metrics

30-50% Memory Save
10x Speedup
Novel Algorithm

Source Code

View Repository