High performance Multi-Criterion Decision Making Algorithms (MCDMs) written in Rust for 🚀 and built for the web with WASM.
Currently implements Promethee I/II.
Parallelism This project leverages rayon and web-workers to take advantage of every core.
Low-Memory These algorithms often require pair-wise comparisons which can rapidly exceed the memory available if care is not taken. In this demonstration we can analyze 10k alternatives, which requires over 100,000,000 comparisons per criteria and there are 7 criteria (thats 700M comparisons!) -- all solved with under 20MB of memory.
High Performance The combination of the above features means that this library is very high performance. Users can calculate 700 million weighted pair-wise comparisons in under 1 second. Users can also explore the impact of adjusting the weights and see updated rankings for all 10k alternatives in under 0.5 milliseconds.
Serverless This implementation is purely client-side.
Works on Mobile The memory and computation are so well optimized that the project runs on mobile phones just as responsively as on desktop machines.
All of the following are possible with no changes to the internal mcdmrs-web library:
- editable criteria weights, preference functions, indifferences and preferences.
- upload user alternatives (csv) and criteria (csv).
- upload user alternatives as geojson and display scores as choropleth.