Tweet: 1/ Matrix-aware optimizers like Muon feel like algorithmic black magic. But it turns out they aren't heuristics at all. They are exact continuous-time gradient flows operating in a completely different geometric space. 🧵 @che_shr_cat: "1/ Matrix-aware optimizers like Muon feel like algorithmic black magic. But it turns out they aren't heuristics at all. They are exact continuous-time gradient flows operating in a completely different geometric space. 🧵 https://x.com/che_shr_cat/status/2044807483852571132/photo/1"