disco-sparse-array 0.1.0

Sparse vector and matrix classes with fast operations for Disco simulations

🧮 sparse-array

Sparse vector and matrix classes with fast C++/NumPy-backed operations
for use in Disco simulation programs and other large-scale discrete event models.

---

Overview

disco-sparse-array provides compact sparse vector and sparse matrix types that operate efficiently on large, structured numerical data.
The package is optimized for repeated arithmetic and transformation operations that occur during Monte Carlo and discrete-event simulations, as used in the Disco simulation framework.

It combines:

• Lightweight Python classes for usability.
• C++/pybind11 kernels for compute-intensive operations.
• NumPy interoperability for zero-copy exchange of data.

---

✨ Features

• Sparse Vector and Matrix classes with explicit indices and values.
• Fast arithmetic:
  • a + b, a @ m, and in-place variants.
• Change detection and subset validation (diff, changes).
• Conversion utilities:
  • Sparse → dense arrays (to_dense).
  • Dense → sparse (via index filtering).
• Consistent API and predictable semantics across all operations.
• MIT-licensed, pure Python/C++ — no external runtime dependencies beyond NumPy.

---

🚀 Installation

pip install disco-sparse-array