GeomScale Project

Twelve months of core development, maintenance, and documentation for GeomScale and its key libraries (volesti, Rvolesti, dingo). Funding covers senior maintainer time, infrastructure, docs, community support, and contingency. Total request: $85,000/year. This support sustains a niche but critical open-source stack of algorithms, keeping them available for computational biology, finance, statistics, AI, and machine learning.

Six months of focused development on algorithmic improvements and language bindings. Focus areas: - Parallel and SIMD implementations in volesti for large-scale problems - New sampling algorithms (e.g., improved walks and advanced proposals) and better defaults - Python bindings for wider adoption - Performance engineering and benchmarking against alternative toolchains Requested amount: $50,000 (one-time).

A comprehensive documentation and learning-materials push for GeomScale. Deliverables: - Unified API documentation for volesti, Rvolesti, and dingo - End-to-end tutorials for computational biology (metabolic networks) and finance (portfolio optimization, risk scenarios) - Jupyter notebooks for teaching high-dimensional sampling and volume computation - Talks and videos explaining use cases for data science and ML practitioners Requested amount: $25,000 (one-time).

Any amount of unrestricted support helps keep GeomScale sustainable and responsive to user needs. Funds are used for maintenance, bug fixes, and incremental improvements that are hard to fund via traditional research grants.