Exact Discrete Stochastic Simulation With Deep‐Learning‐Scale Gradient Optimization
Open Access
- 10 July 2026
- journal article
- Published by Wiley in Advanced Science
- Vol. 13 (52) , e76297
Abstract
Exact stochastic simulation of continuous‐time Markov chains (CTMCs) is essential when discreteness and noise drive system behavior, but the hard categorical event selection in Gillespie‐type algorithms blocks gradient‐based learning. We eliminate this constraint by decoupling forward simulation from backward differentiation, with hard categorical sampling generating exact trajectories and gradients propagating through a continuous massively‐parallel Gumbel‐Softmax straight‐through surrogate. Our approach enables accurate optimization at parameter scales over four orders of magnitude beyond existing simulators. We validate for accuracy, scalability, and reliability on a reversible dimerization model (0.09% error), a genetic oscillator (1.2% error), a 203,796‐parameter gene regulatory network achieving 98.4% accuracy on the MNIST handwritten‐digit dataset (a prototypical deep‐learning multilayer perceptron benchmark), and experimental patch‐clamp recordings of ion channel gating (R 2 = 0.987) in the single‐channel regime. Our GPU implementation delivers 1.9 billion steps per second, matching the scale of non‐differentiable simulators. By making exact stochastic simulation massively parallel and autodiff‐compatible, our results enable high‐dimensional parameter inference and inverse design across systems biology, chemical kinetics, physics, and related CTMC‐governed domains.Keywords
All Related Versions
Funding Information
- Ministerio de Ciencia, Innovación y Universidades (PID2024‐160016NB‐I00)
- Ministerio de Ciencia, Innovación y Universidades (MICIU/AEI/10.13039/ 501100011033)
- Ministerio de Ciencia, Innovación y Universidades (ERDF/EU)
This publication has 33 references indexed in Scilit:
- Nature, Nurture, or Chance: Stochastic Gene Expression and Its ConsequencesCell, 2008
- Universally Sloppy Parameter Sensitivities in Systems Biology ModelsPLOS Computational Biology, 2007
- Classical Nucleation Theory of Virus CapsidsBiophysical Journal, 2006
- Efficient step size selection for the tau-leaping simulation methodThe Journal of Chemical Physics, 2006
- Real-Time Kinetics of Gene Activity in Individual BacteriaCell, 2005
- From molecular noise to behavioural variability in a single bacteriumNature, 2004
- Stochastic Gene Expression in a Single CellScience, 2002
- Mechanisms of noise-resistance in genetic oscillatorsProceedings of the National Academy of Sciences, 2002
- Stochastic mechanisms in gene expressionProceedings of the National Academy of Sciences, 1997
- On the stochastic properties of single ion channelsProceedings of the Royal Society of London. B. Biological Sciences, 1981