Physicists Solve Longstanding Puzzle of Cellular Noise Using Spin Glass Physics
A novel analytical framework combining field theory and variational methods accurately tracks complex fluctuations in low-copy-number biochemical networks.
Biological cells and small ecological populations are governed by intense random fluctuations that standard mathematical formulas often fail to capture. Researchers have struggled for decades to accurately predict how discrete chemical reactions and population networks behave when molecule counts drop to very low numbers, where traditional mass action laws break down. Now, physicists Moshir Harsh and Peter Sollich have developed a powerful new analytical framework that successfully tracks these complex dynamics by drawing on tools from spin glass physics and quantum field theory.
Why Small Populations Defy Standard Math
When studying biochemical reaction networks inside tiny cellular compartments, atto-litre volumes mean that key signaling molecules or transcription factors might only exist in one or two copies. In this regime, intrinsic noise creates massive variations between identical cells. Traditional approaches like mass action kinetics assume smooth, continuous concentrations, completely ignoring the discrete nature of individual molecules. Other approximation schemes quickly fail when copy numbers are low, while exact numerical simulations such as the Gillespie algorithm demand immense computational power as systems grow more complex.
To overcome these hurdles, the authors built a path integral representation using Doi-Peliti second quantization, mapping discrete population jump processes to continuous fields. They treated simple unary creation and destruction reactions as a baseline Gaussian theory, handling binary interactions through an advanced perturbation expansion. The work is available on the arXiv preprint server, outlining how a constrained free energy approach can derive equations for system observables.
Unlocking Memory Effects Through Variational Theory
By applying the Plefka expansion, originally formulated to understand disordered magnetic systems known as spin glasses, the researchers derived a dynamical variational principle. While the underlying microscopic reaction processes are strictly Markovian, meaning they depend only on the current state, the resulting effective dynamics incorporate self-consistent memory terms. These memory kernels integrate past states and account for the feedback of unresolved higher-order moments onto the retained population means and correlation functions.
To prevent instabilities caused by large interaction rates, the team introduced an innovative technique called generalized Self-consistent Bubble Resummation (gSBR). This approach sums an infinite subseries of diagrams to regularize the memory kernel. As a result, the method provides stable, highly accurate predictions for marginal distributions, response functions, and temporal correlations across transient and stationary regimes.
Benchmarking Gene Regulation and Epidemic Spreading
To demonstrate the versatility of their framework, the researchers tested it on three distinct biological paradigms. In gene regulation models featuring an autorepressive loop, where protein binding inactivates a gene, gSBR accurately captured bursty production and intermittent fluctuations that standard linear noise approximations missed. For Michaelis-Menten enzyme kinetics involving a single average enzyme molecule, the approach successfully reproduced complex cross-correlations and multi-scale dynamics.
Finally, the team applied the framework to Susceptible-Infected-Recovered epidemic models in small populations. While conventional mean-field equations severely exaggerate outbreak sizes and predict that nearly every susceptible individual will become infected, gSBR correctly accounted for extinction probabilities and bimodal steady-state distributions. To make these tools accessible, the authors released a plug-and-play computational package named Achedmy.jl, integrated with the Julia scientific programming language, providing researchers with an efficient resource for exploring strongly fluctuating stochastic systems.
The study was posted to the arXiv preprint server on November 19, 2026.
This article has been fact checked for accuracy, with information verified against reputable sources. Learn more about us and our editorial process.
Last reviewed on .
Article history
- Latest version
Reference(s)
- Harsh, Moshir. “Dynamical Variational Theory of Stochastic Reaction Networks.” arXiv.org <http://arxiv.org/abs/2611.25491v1>.
Cite this page:
- Posted by Omar Mirza