Data folks and math-inclined devs. I'm looking for an Open Source library for Optimization and Linear Algebra. Kotlin preferred, but can be in any other JVM language compatible with Kotlin (Java, Clojure, Scala, etc). It needs to support arbitrary-precision math (e.g., BigDecimal) and be able to scale horizontally to solve millions of problems in parallel for short bursts (not necessarily HPC scale). Ideally something I can AOT-compile with GraalVM and trigger tens of thousands of serverless functions in parallel, or quickly spin up a small k8s cluster, schedule a few million jobs, and then tear the cluster down. Throughput at scale beats raw performance for my use case. So far I’ve come across Klan and Multik for linear algebra; Kotpim and Kotlin-MIP for optimization. I have no practical experience with any of them and no idea about AOT compilation performance and throughput (this is a must, because after the MVP the plan is to distribute those computations across a gazillion personal devices, e.g., mobile phones, Raspberry Pis, and even embedded devices that may not easily run a JVM and some with less than 1 GB RAM). Any pointers? I’m also fine with non-permissive licenses (e.g., AGPL). #FOSS #Kotlin #devstr #asknostr