Topological building blocks of nonequilibrium response
arXiv:2607.12096
2026
Dynamics
2 ideas extracted · analyzed Aug 30, 2026
What the math gives to ML
The paper proposes a geometric characterization of nonequilibrium response: maximally sensitive response models are associated with topological structures of a state-space graph, while general responses are conjectured to be convex combinations of these extremal models. The transferable asset is a response-basis construction in which topology determines primitive input-output curves and mixture weights describe less-sensitive systems. A neural implementation can use an input-conditioned Markov latent layer or a response regularizer that projects learned sensitivity curves onto a library of topological extremals, yielding falsifiable convex-hull and sensitivity-bound tests.
Ideas from this paper
△ Mechanism confirmed, baseline not beaten
2026
Insert a small continuous-time Markov latent module between a neural encoder and decoder, with input-dependent transition rates and a fixed library of graph topologies such as directed cycles, reversible chains, and branching motifs. The output is an observable of the stationary distribution, while a learned convex mixture over topology-specific response curves constrains the network to represent responses as combinations of interpretable nonequilibrium mechanisms.
Useful7/10
Difficulty6/10
Novelty8/10
Unverified
2026
Add a response-sensitive regularizer to networks whose outputs should react predictably to a control input, using the stationary Markov sensitivity equation as a certificate. Instead of only penalizing large neural gradients, the method attributes amplification to the generator resolvent and can distinguish amplification caused by a nearly slow latent mode from amplification caused by uncontrolled parameter growth.
Useful6/10
Difficulty5/10
Novelty7/10