Here’s what it looks (and sounds) like when you let the “Auction algorithm” do its job and unravel your data. File under: oddly satisfying.
The Auction algorithm was developed by Dimitri Bertsekas (1942-2026), a Greek-American scientist and professor at the Massachusetts Institute of Technology, to provide an efficient solution to the assignment problem. He passed away recently, on June 3, 2026, leaving behind a legacy of major contributions to the fields of graph optimization, large-scale computing, reinforcement learning, and artificial intelligence.
For those who talk in math, the assignment problem consists in “finding, in a weighted bipartite graph, a matching of maximum size, in which the sum of weights of the edges is minimum“… Precisely what I needed in the Brain Orchestra project, to map clusters of neurons to their corresponding atlas labels. I used to use the Hungarian algorithm, and it worked well, but the Auction algorithm, which a fellow Max user told me about recently, is good enough and much faster. (thanks, Gussi!)
If this sounds a bit abstract to you, just know that this kind of algorithm is used in virtually every large-scale data management application to assign you the closest cab, the best dating match, the next song in your tracklist, the right time slot for your appointment, the next ad you’ll see on your social media, the optimal timetable for that new school year, etc.—or on the battlefield, to deploy the most effective military assets against imminent threats, or firefighting teams in burning forests.
