A nice video on the Mercator map projection from @drjoehanson.bsky.social and @pbs.org Be Smart youtu.be/p-9wxYdJing?...
Posts by Austin Buchanan
TIL that gerrymandering is part of the Orban story
For all the LP and MIP enthusiasts out there #orms
…and the final update(?) bsky.app/profile/rick...
Current mood
To all the linear programming enthusiasts out there, how are you gearing up for dual variables day? #orms
“An electric lizard killed my process” is the new “my dog ate my homework”
Also RIP to the 3-day computational experiments I had been running when Lizard caused my PC to reboot
Not the Onion. RIP Lizard www.news9.com/oklahoma-cit...
Okay, who’s gonna MIP it first? And make a web interface? #orms
📢 Abstract submission for the EUROYoung Workshop 2026 in Lugano (CH) opens soon!
📅 Workshop: 14–16 October 2026
📝 Abstracts: 10 March – 15 April 2026
Join young researchers in OR for talks, networking & social events.
💻 euroyoung.eu/lugano2026
📧 euroyoung2026@usi-supsi.ch
@euroyoung.bsky.social
And the vibe coding tool for mathematical optimization would be A Math Programmer’s Lollipop, or AMPL for short
APL = Alan Perlis Lollipop. Ok I’ll stop
Or maybe “A Programmer’s Lollipop”. APL for short
If I were creating a GenAI tool for vibe coding, I’d call it ALollipop
Breaking: Supreme Court Reverses Lower Court Order That Redrew Congressional Map in Staten Island, with Ominous Implications for Voting Rights Act electionlawblog.org?p=154596
bsky.app/profile/stev...
And another… bsky.app/profile/stev...
In 2019 I created a Story Map explainer of NC's distinctive county grouping requirement for legislative redistricting. Soon that resource will vanish (Esri is retiring that old platform). If you want a refresher on the so-called "Stephenson" process, now's the time! #NCPol
1/3
arcg.is/1DnDO
In 2020, Stephenson did quite well in terms of minimizing splits. The state house map is optimal, and the state senate map is suboptimal by only two splits. pubsonline.informs.org/doi/abs/10.1... But the method can fail spectacularly if the populations are chosen just so… if I ever write that up…
Essentially, a heuristic method to minimize county splitting, as proposed by NC courts
The CDS formulation is especially neat. One type of constraint, but exponentially many of them:
\sum_{ i in C } x_i >= 1 for every vertex cut C in the graph
github.com/AustinLBucha...
I have had more success with lazy cuts over user cuts, particularly for problems with connectivity constraints (e.g., TSP, connected dominating set, districting). But you can’t quite compare “cuts” versus “no cuts” because the “no cuts” model is incomplete. But you can compare against flow models
Question for #orms people: do you know any problem-specific cutting planes that are not too hard to explain and separate and would improve performance compared to a baseline MIP solver (e.g. SCIP, Gurobi)? Asking because its more fun for students if they implement something and it helps performance
There’s a new paper on arXiv saying the matching polytope has extension complexity O*( 1.5^n ), beating the Edmonds bound O( 2^n ).
Usually 1, but sometimes 2. The trouble is things like { x | |x| <= 5 } vs { x : |x| <= 5 }