Recently, Google DeepMind made waves with a blog post about AlphaEvolve, a new LLM-based tool that writes its own algorithms to crack notoriously difficult math problems.
Just weeks after Google published new best-known solutions for 13 previously unsolved problems using AlphaEvolve, a team from the MODAL research campus developed formulations that were solved with FICO Xpress, outperforming AlphaEvolve’s results and setting new records in solution quality on several challenging benchmarks, including circle packing and the minimization of distance ratios.
This result impressively demonstrates that established optimization methods with solid mathematical foundations continue to play a central role in solving complex problems, even in the age of generative AI.
For more information, see:
🔗 Full FICO blog post:
https://www.fico.com/blogs/best-global-optimization-solver
🔗 Full Google DeepMind blog post:
https://deepmind.google/discover/blog/alphaevolve-a-gemini-powered-coding-agent-for-designing-advanced-algorithms/