A new event-grounded benchmark shows LLM agents remain far below saturation on personalized financial memory, with simple retrieval outperforming structured memory systems especially after consequential events.
Emergent phenomena in living systems: a statistical mechanical perspective
1 Pith paper cite this work. Polarity classification is still indexing.
abstract
A natural phenomenon occurring in a living system is an outcome of the dynamics of the specific biological network underlying the phenomenon. The collective dynamics have both deterministic and stochastic components. The stochastic nature of the key processes like gene expression and cell differentiation give rise to fluctuations (noise) in the levels of the biomolecules and this combined with nonlinear interactions give rise to a number of emergent phenomena. In this review, we describe and discuss some of these phenomena which have the character of phase transitions in physical systems. We specifically focus on noise-induced transitions in a stochastic model of gene expression and in a population genetics model which have no analogs when the dynamics are solely deterministic in nature. Some of these transitions exhibit critical-point phenomena belonging to the mean-field Ising universality class of equilibrium phase transitions. A number of other examples, ranging from biofilms to homeostasis in adult tissues, are also discussed which exhibit behavior similar to critical phenomena in equilibrium and nonequilbrium phase transitions. The examples illustrate how the subject of statistical mechanics provides a bridge between theoretical models and experimental observations.
fields
cs.AI 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
FinPerMA: A Theory-Informed, Event-Grounded Personalized-Memory Benchmark for LLM Agents
A new event-grounded benchmark shows LLM agents remain far below saturation on personalized financial memory, with simple retrieval outperforming structured memory systems especially after consequential events.