Pachter presses on: "disproving" Hou-Jin-Liu conjecture is only numerical evidence
lpachter · x · 2026-09-09
Continuing his critique of the Anandkumar group's Euler singularity paper, Lior Pachter points out that the paper claims its approximate profiles "disprove" the Hou-Jin-Liu conjecture — but until an exact singular solution and a stability argument are established, the accurate claim would be "numerical evidence against the conjecture." The thread's core point: AI-plus-math papers shouldn't dress computational evidence up as rigorous disproof.
Related event: Researcher Questions Anandkumar's Fluid Singularity Paper Claims(4 posts)→
More from Research
- Strogatz: two overlooked mathematicians cracked the key to the AI Navier-Stokes proof — stevenstrogatz · 2026-09-09
- Weekly must-read papers: MoE looped transformer scaling laws, world model physics evals — TheTuringPost · 2026-09-09
- Three people solve 90-year-old math problem in the same week — same AI model the common thread? — cephaloform · 2026-09-09
- Personalized skin cancer vaccine hailed as first major immunization win credited to AI — ColbyHawker · 2026-09-09
- PHA team teases final PHA-4 update within months, cites proof linked to OpenAI — jasondeanlee · 2026-09-09
- Self-trained model switches to weekly Sunday weight updates for richer reward gradients — cephaloform · 2026-09-09