Downhill from here
Hybrid Meetup #57 took place 2026-01-27 19:00 at Basislager Leipzig and we had a great presentation by Karl, mixing mathematical optimization, Go and webassembly in his open source OnlyOffice NLP Solver plugin:
The project can be found on GitHub.
Some bits from the talk:
- nonlinear optimization can be made approachable to a larger user base by embedding it into a common (data) environment, like a spreadsheet application
- the plugin handles constrained parameters and side conditions
- parsing arbitrary arithmetic expressions requires nesting and recursion, handled well by context-free grammars
- wiring WASM to the onlyoffice spreadsheet api is another step to consider
- for linear optimization, you can use the simpler simplex (one of the top ten algorithms of the 20th century)
An example nonlinear optimization problem could be optimal warehouse placement (FACILITY LOCATION PROBLEM): Given a number of customers with fixed locations and different demands, where should a warehouse be optimally placed to minimize the total transport distances (if you are a delivery service and you want to open a dark store, you may want to use this).
In the visualization, blue dots represent customers (their size is proportional to their demand) and the red dot is the optimal location, as discovered by the Nelder-Mead algorithm; more specific algorithms exist for this problem).
Misc
- Will open source survive AI? When it becomes cheaper to generate code than to search for an open source solution, that may or may not cover all the requirements? A recent personal example is gh-repos.
- Go has Web Assembly support since 1.11, released 2018-08-24
Thanks again Karl for the nice presentation and project!
Join our meetup to get notified of upcoming events.
