The day in conversation for 2026-10-26

discussion

Transcript

The discussion: Tom: That convergence proof using the Horn-Jackson identity sounds incredibly powerful for establishing uniqueness, but I’m struggling to see how that specific exponential family mapping translates directly into a robust Lyapunov function for systems where the parameters themselves might be subject to stochastic fluctuation. Jane: The key difficulty, if I’m reading this right, is whether that descent argument holds when the stoichiometric compatibility class itself changes over time due to external inputs rather than just internal system dynamics. What assumptions are they making about the bounded speed constraint in those scenarios? Lu: From a distribution standpoint, I'm less worried about stochastic fluctuations and more concerned with the sheer computational overhead of calculating that exponential family mapping for large systems; how does the complexity scale relative to the size of the state space being analyzed? Meng: I don’t know much about the Horn-Jackson identity specifics, but if we look at

More episodes

← Home