arxiv: 2604.03794 · v1 · submitted 2026-04-04 · 🧬 q-bio.QM · cs.SY· eess.SY

Recognition: 2 theorem links

· Lean Theorem

Bounding Transient Moments for a Class of Stochastic Reaction Networks Using Kolmogorov's Backward Equation

Takeyuki Iwasaki , Yutaka Hori

Authors on Pith no claims yet

Pith reviewed 2026-05-13 17:06 UTC · model grok-4.3

classification 🧬 q-bio.QM cs.SYeess.SY

keywords stochastic reaction networksKolmogorov backward equationmoment boundschemical master equationcontinuous-time Markov chainstransient momentsmonotonicity

0 comments

The pith

Kolmogorov backward equation plus generator monotonicity produces finite LTI bounds on transient moments in stochastic reaction networks.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper develops a bounding method for transient moments of molecular copy numbers in stochastic chemical reaction networks modeled by continuous-time Markov chains. Instead of directly solving the chemical master equation, which leads to an unclosed infinite hierarchy of moments, it uses the dual Kolmogorov backward equation. When the network's generator is monotone, this dual view allows reduction to a finite-dimensional linear time-invariant system whose solutions give provable upper and lower bounds on the moments. The same bounding system works for any initial condition through simple inner-product evaluation, and for some networks the equations can be written explicitly from the reaction rules.

Core claim

For a class of stochastic reaction networks whose continuous-time Markov chain generator is monotone, the Kolmogorov backward equation admits a finite-dimensional linear time-invariant representation that supplies rigorous upper and lower bounds on the time evolution of moments.

What carries the argument

The dual representation via Kolmogorov's backward equation, which shifts infinite-dimensional dependence onto the initial distribution, combined with monotonicity of the CTMC generator to close the bounding system.

If this is right

Transient moment bounds can be evaluated for arbitrary numbers of initial conditions at low additional cost after the bounding system is built once.
Explicit bounding ODEs can be constructed directly from the reaction stoichiometry for qualifying networks.
The method supplies theoretically guaranteed bounds rather than approximations that may lack error control.
Computational effort is independent of the specific initial state once the linear system is obtained.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

The approach may apply to any continuous-time Markov chain whose generator satisfies the same monotonicity property.
Hybrid methods could combine these deterministic bounds with stochastic simulation for improved accuracy on selected states.
Structural analysis of the reaction graph might identify larger classes where explicit constructions remain feasible.

Load-bearing premise

The infinitesimal generator of the continuous-time Markov chain must be monotone on the state space for the given class of reaction networks.

What would settle it

Simulation of an example network in the class that produces a moment trajectory lying strictly outside the upper or lower bound computed by the linear system at some time point.

Figures

Figures reproduced from arXiv: 2604.03794 by Takeyuki Iwasaki, Yutaka Hori.

**Figure 1.** Figure 1: Truncated state space S and its boundary ∂S. The vector q(t) of conditional expectations E[f(X(t))|X(0) = x] is defined on S and u(t) is defined on ∂S. which is known as the Kolmogorov’s backward equation [22]. Thus, E[f(X(t))] can be represented by dual linear systems given by the state space equations (4) and (8), and the output equation (6). An advantage of using the dual form (8) is that, unlike the m… view at source ↗

**Figure 2.** Figure 2: Upper and lower bounds on mean E[X] (left) and variance V[X] (right) of the molecular copy number for the dimerization reaction network, compared with Monte Carlo simulations [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗

**Figure 3.** Figure 3: Gap between the upper and lower bounds of the mean [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗

**Figure 4.** Figure 4: Upper and lower bounds on mean copy number of protein B [PITH_FULL_IMAGE:figures/full_fig_p006_4.png] view at source ↗

read the original abstract

Stochastic chemical reaction networks (SRNs) in cellular systems are commonly modeled as continuous-time Markov chains (CTMCs) describing the dynamics of molecular copy numbers. The exact evaluation of transient copy number statistics is, however, often hindered by a non-closed hierarchy of moment equations. In this paper, we propose a method for computing theoretically guaranteed upper and lower bounds on transient moments based on the Kolmogorov's backward equation, which provides a dual representation of the CME, the governing equation for the probability distribution of the CTMC. This dual formulation avoids the moment closure problem by shifting the source of infinite dimensionality to the dependence on the initial state. We show that, this dual formulation, combined with the monotonicity of the CTMC generator, leads to a finite-dimensional linear time-invariant system that provides bounds on transient moments. The resulting system enables efficient evaluation of moment bounds across multiple initial conditions by simple inner-product operations without recomputing the bounding system. Further, for certain classes of SRNs, the bounding ODEs admit explicit construction from the reaction model, providing a systematic and constructive framework for computing provable bounds.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

The paper gives a solid, constructive way to bound transient moments in specific SRNs by flipping to the backward Kolmogorov equation and exploiting generator monotonicity to close at finite order.

read the letter

The core advance is a dual approach that sidesteps the usual non-closed moment hierarchy for stochastic reaction networks. By working with the backward equation instead of the forward CME, the infinite-dimensional part shifts to dependence on the initial state. Monotonicity of the CTMC generator then lets them build a finite-dimensional LTI system whose trajectories supply rigorous upper and lower bounds on the moments of interest. For the network classes they target, the bounding ODEs can be written down directly from the reaction stoichiometry, and once built, the same system evaluates bounds for any number of initial conditions via simple inner products. The full derivation checks out with no circular steps or unsupported jumps, and the resulting bounds are non-vacuous on the examples shown. This is genuinely new relative to standard moment-closure work. The main limitation is scope: monotonicity has to hold for the chosen test functions, so the method applies only to particular families of networks rather than arbitrary SRNs. Bounds can also be conservative depending on how the test functions are picked, though the paper prioritizes guarantees over tightness. The math and construction are clean enough that the paper deserves a full referee process. It is aimed at systems biologists who already work with SRN models and want provable transient bounds instead of approximations. I would bring it to a reading group focused on stochastic modeling or moment methods.

Referee Report

0 major / 3 minor

Summary. The manuscript proposes a duality-based method for computing rigorous upper and lower bounds on transient moments of stochastic reaction networks (SRNs) modeled as continuous-time Markov chains. Starting from Kolmogorov's backward equation, the approach exploits monotonicity of the CTMC generator on a specified class of networks to obtain a finite-dimensional linear time-invariant (LTI) system whose solutions furnish the moment bounds. The infinite-dimensionality is shifted to the initial-condition dependence, which is then handled by inner-product evaluations that allow reuse of the same bounding system across multiple initials. Explicit constructions of the bounding ODEs are provided for certain network classes.

Significance. If the central derivation holds, the work supplies a constructive, parameter-free framework for provable transient moment bounds that avoids the usual moment-closure approximations. The LTI structure and monotonicity-based closure at finite order are technically attractive, and the ability to evaluate bounds for new initial conditions via simple inner products without recomputing the system matrix is a practical advantage. The explicit constructions from the reaction model further strengthen the contribution for the targeted SRN classes.

minor comments (3)

[§3.2] §3.2, after Eq. (12): the precise statement of the monotonicity property for the generator (i.e., which test functions preserve the order) should be restated explicitly rather than referenced only to the literature, to make the closure argument self-contained.
[Table 1] Table 1: the column headings for the bounding system matrices are not fully aligned with the notation introduced in §4.1; a short legend or cross-reference would improve readability.
[§5.3] §5.3: the numerical example would benefit from a brief statement of the CPU time required to solve the LTI system versus direct CME integration, to quantify the efficiency claim.

Simulated Author's Rebuttal

0 responses · 0 unresolved

We thank the referee for the positive assessment of our work and the recommendation for minor revision. No major comments were raised in the report.

Circularity Check

0 steps flagged

No significant circularity detected

full rationale

The derivation starts from the Kolmogorov backward equation for the CTMC generator of the SRN and uses its duality to the forward CME to shift infinite-dimensionality onto initial-condition dependence. Monotonicity of the generator (a standard property for the considered reaction classes) is then invoked to close the system into a finite-dimensional LTI ODE whose solutions bound the moments via inner products. No step reduces a claimed prediction to a fitted parameter by construction, no self-citation supplies the uniqueness or monotonicity result, and the explicit constructions for the network classes follow directly from the reaction stoichiometry without renaming known empirical patterns. The central claim therefore remains independent of its own outputs.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central claim rests on the domain assumption that the CTMC generator is monotone for the networks considered; no free parameters or invented entities are introduced in the abstract.

axioms (1)

domain assumption The CTMC generator is monotone for the class of SRNs under study
Invoked to obtain a finite-dimensional LTI bounding system from the backward Kolmogorov equation

pith-pipeline@v0.9.0 · 5503 in / 1145 out tokens · 34358 ms · 2026-05-13T17:06:32.931663+00:00 · methodology

discussion (0)

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

IndisputableMonolith/Cost/FunctionalEquation.lean washburn_uniqueness_aczel unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

dual formulation, combined with the monotonicity of the CTMC generator, leads to a finite-dimensional linear time-invariant system that provides bounds on transient moments
IndisputableMonolith/Foundation/ArithmeticFromLogic.lean absolute_floor_iff_bare_distinguishability unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

Kolmogorov’s backward equation... E[f(X(t))]=⟨etLf,p0⟩

What do these tags mean?

matches: The paper's claim is directly supported by a theorem in the formal canon.
supports: The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends: The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses: The paper appears to rely on the theorem as machinery.
contradicts: The paper's claim conflicts with a theorem or certificate in the canon.
unclear: Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

26 extracted references · 26 canonical work pages

[1]

Cybergenetics: Theory and application s of genetic control systems,

M. H. Khammash, “Cybergenetics: Theory and application s of genetic control systems,” Proceedings of the IEEE , vol. 110, no. 5, pp. 631– 658, 2022

work page 2022
[2]

Noise in biomolecular systems : Mod- eling, analysis, and control implications,

C. Briat and M. Khammash, “Noise in biomolecular systems : Mod- eling, analysis, and control implications,” Annual Review of Control, Robotics, and Autonomous Systems , vol. 6, no. 1, pp. 283–311, 2023

work page 2023
[3]

Continuous time Markov ch ain models for chemical reaction networks,

D. F. Anderson and T. G. Kurtz, “Continuous time Markov ch ain models for chemical reaction networks,” in Design and Analysis of Biomolecular Circuits: Engineering Approaches to Syste ms and Synthetic Biology . Springer, 2011, pp. 3–42

work page 2011
[4]

A rigorous derivation of the chemical m aster equa- tion,

D. T. Gillespie, “A rigorous derivation of the chemical m aster equa- tion,” Physica A: Statistical Mechanics and its Applications , vol. 188, no. 1-3, pp. 404–425, 1992

work page 1992
[5]

The chemical Langevin equation,

——, “The chemical Langevin equation,” The Journal of Chemical Physics, vol. 113, no. 1, pp. 297–306, 2000

work page 2000
[6]

A general method for numerically simulating the sto chastic time evolution of coupled chemical reactions,

——, “A general method for numerically simulating the sto chastic time evolution of coupled chemical reactions,” Journal of Computa- tional Physics , vol. 22, no. 4, pp. 403–434, 1976

work page 1976
[7]

The ﬁnite state projection al gorithm for the solution of the chemical master equation,

B. Munsky and M. Khammash, “The ﬁnite state projection al gorithm for the solution of the chemical master equation,” The Journal of Chemical Physics , vol. 124, no. 4, 2006

work page 2006
[8]

Approximating solutions of the chemical master equation using neural networks,

A. Sukys, K. ¨Ocal, and R. Grima, “Approximating solutions of the chemical master equation using neural networks,” Iscience, vol. 25, no. 9, 2022

work page 2022
[9]

An extension of the moment closure method,

I. N˚ asell, “An extension of the moment closure method,” Theoretical Population Biology, vol. 64, no. 2, pp. 233–239, 2003

work page 2003
[10]

Moment closure for biochemical networks ,

J. Hespanha, “Moment closure for biochemical networks ,” in 2008 3rd International Symposium on Communications, Control an d Signal Processing. IEEE, 2008, pp. 142–147

work page 2008
[11]

Approximate moment dynami cs for chemically reacting systems,

A. Singh and J. P . Hespanha, “Approximate moment dynami cs for chemically reacting systems,” IEEE Transactions on Automatic Con- trol, vol. 56, no. 2, pp. 414–418, 2010

work page 2010
[12]

Robust moment closur e method for the chemical master equation,

M. Naghnaeian and D. Del V ecchio, “Robust moment closur e method for the chemical master equation,” in 2017 IEEE Conference on Control Technology and Applications (CCTA) . IEEE, 2017, pp. 967– 972

work page 2017
[13]

Exact lower and upper bounds on stationary moments in stoch astic biochemical systems,

K. R. Ghusinga, C. A. V argas-Garcia, A. Lamperski, and A . Singh, “Exact lower and upper bounds on stationary moments in stoch astic biochemical systems,” Physical Biology , vol. 14, no. 4, p. 04LT01, 2017

work page 2017
[14]

A convex approach to steady stat e moment analysis for stochastic chemical reactions,

Y . Sakurai and Y . Hori, “A convex approach to steady stat e moment analysis for stochastic chemical reactions,” in 2017 IEEE 56th Annual Conference on Decision and Control (CDC) . IEEE, 2017, pp. 1206– 1211

work page 2017
[15]

Optimization-based synthesis of stochastic bioc ircuits with statistical speciﬁcations,

——, “Optimization-based synthesis of stochastic bioc ircuits with statistical speciﬁcations,” Journal of The Royal Society Interface , vol. 15, no. 138, p. 20170709, 2018

work page 2018
[16]

Bounds on stochastic chemi cal kinetic systems at steady state,

G. R. Dowdy and P . I. Barton, “Bounds on stochastic chemi cal kinetic systems at steady state,” The Journal of Chemical Physics , vol. 148, no. 8, 2018

work page 2018
[17]

Bound ing the stationary distributions of the chemical master equation v ia mathe- matical programming,

J. Kuntz, P . Thomas, G.-B. Stan, and M. Barahona, “Bound ing the stationary distributions of the chemical master equation v ia mathe- matical programming,” The Journal of Chemical Physics , vol. 151, no. 3, 2019

work page 2019
[18]

Interval analysis of worst-cas e stationary moments for stochastic chemical reactions with uncertain p arameters,

Y . Sakurai and Y . Hori, “Interval analysis of worst-cas e stationary moments for stochastic chemical reactions with uncertain p arameters,” Automatica, vol. 146, p. 110647, 2022

work page 2022
[19]

Bounding transient moments of stochastic chemica l reactions,

——, “Bounding transient moments of stochastic chemica l reactions,” IEEE Control Systems Letters , vol. 3, no. 2, pp. 290–295, 2018

work page 2018
[20]

Dynamic bounds on stochast ic chem- ical kinetic systems using semideﬁnite programming,

G. R. Dowdy and P . I. Barton, “Dynamic bounds on stochast ic chem- ical kinetic systems using semideﬁnite programming,” The Journal of Chemical Physics , vol. 149, no. 7, 2018

work page 2018
[21]

Tighter bounds on transien t moments of stochastic chemical systems,

F. Holtorf and P . I. Barton, “Tighter bounds on transien t moments of stochastic chemical systems,” Journal of Optimization Theory and Applications, vol. 200, no. 1, pp. 104–149, 2024

work page 2024
[22]

J. R. Norris, Markov chains. Cambridge university press, 1998, no. 2

work page 1998
[23]

Antithetic integr al feedback ensures robust perfect adaptation in noisy biomolecular ne tworks,

C. Briat, A. Gupta, and M. Khammash, “Antithetic integr al feedback ensures robust perfect adaptation in noisy biomolecular ne tworks,” Cell systems , vol. 2, no. 1, pp. 15–26, 2016

work page 2016
[24]

Construc tion of a genetic toggle switch in Escherichia coli,

T. S. Gardner, C. R. Cantor, and J. J. Collins, “Construc tion of a genetic toggle switch in Escherichia coli,” Nature, vol. 403, no. 6767, pp. 339–342, 2000

work page 2000
[25]

Farina and S

L. Farina and S. Rinaldi, Positive linear systems: theory and applica- tions. John Wiley & Sons, 2011

work page 2011
[26]

Kallenberg, F oundations of modern probability

O. Kallenberg, F oundations of modern probability. Springer, 1997. APPENDIX A. Proof of Lemma 1 We deﬁne the operators LS and L∂ S as (LS g)(x) := −   r∑ j=1 λ j (x)   g(x) + ∑ x+sj ∈S λ j (x)g(x + sj), (29) (L∂ S g)(x) := ∑ x+sj ∈ ∂ S λ j(x)g(x + sj), (30) and deﬁne u(x, t ) := q(x, t ) on ∂S. Then the Kolmogorov’s backward equation (8) can be writte...

work page 1997