Emergent conserved quantities via irreversibility

This paper demonstrates that irreversible reactions in chemical reaction networks and Markov chains generate emergent conservation laws and broken cycles, resolving a recent conundrum regarding non-integer conservation laws by deriving a new law that links conserved quantities, broken cycles, and a "co-production index" to correct existing undercounting methods.

The Gist

Imagine you are trying to understand a complex machine, like a giant factory or a bustling city. You watch the workers (molecules) moving around, turning raw materials into products, and swapping places. To understand how this machine works, scientists look for rules that never change. These are called "conserved quantities."

For example, in a closed room, the total number of people never changes, even if they move from the kitchen to the living room. In chemistry, this might mean the total number of carbon atoms stays the same, no matter how many reactions happen.

For a long time, scientists had a specific formula (a "rulebook") to count how many of these unchangeable rules exist in a chemical system. But recently, computers using Artificial Intelligence (AI) started finding "ghost rules." These were quantities that seemed to stay constant, but the old rulebook said they shouldn't exist. This was a puzzle: Where did these extra rules come from?

This paper solves that puzzle by introducing a new concept called "Co-production."

The "Double-Task" Analogy

Imagine a factory where two different machines, Machine A and Machine B, are working side-by-side.

• Machine A takes a block of wood and turns it into a chair.
• Machine B takes a block of wood and turns it into a table.

Usually, these are two separate jobs. But imagine a scenario where, due to the way the factory is set up, Machine A and Machine B always run at the exact same speed and use the exact same amount of wood. They are "locked in step."

In the old rulebook, scientists counted these as two separate processes. But the authors of this paper say: "If they are locked in step, treat them as one single process."

They call this merging. When you merge these two synchronized processes, you realize they aren't actually creating two independent outcomes; they are creating a specific, fixed mixture of chairs and tables. This new, merged view reveals a hidden rule: The ratio of chairs to tables produced will always stay the same, no matter how long the factory runs.

This hidden rule is the "Emergent Conserved Quantity." It didn't exist in the old view because the old view was looking at the machines separately. It only appears when you realize the machines are "co-producing" in a synchronized way.

Why Does This Happen? (The "One-Way Street")

The paper explains that this "locking in step" happens most often when reactions are irreversible.

Think of a reversible reaction like a two-way street: cars can go from point A to B, and from B back to A.
Think of an irreversible reaction like a one-way street. Once you go down it, you can't come back.

The authors found that when you have a network of one-way streets, it's very common for two different paths to become "collinear" (parallel). If two one-way paths always carry the same amount of traffic, they effectively become a single, wider path.

When you merge these paths, two things can happen:

1. A Broken Cycle: Sometimes, merging paths breaks a loop that used to exist in the system.
2. A New Rule: Sometimes, merging paths creates a new, unbreakable rule (a conserved quantity) that wasn't visible before.

The "Ghost" Rules Explained

The paper specifically addresses a recent mystery where a computer found a "non-integer" rule.

• Normal Rule: "Total number of atoms = 100." (You can't have half an atom).
• The Ghost Rule: "3.5 times the amount of Chemical X plus 2.2 times Chemical Y = Constant."

This looked weird because you can't have 3.5 atoms. But the authors show that this "weird" rule is actually just the result of merging two irreversible reactions that produce a specific, fractional mix of products. The computer found the rule because the physics of the system demanded it, even if the numbers looked strange.

Real-World Examples in the Paper

The authors tested their idea on two specific types of systems:

1. Atmospheric Chemistry: They looked at a model of the air we breathe. A computer had found a mysterious rule about how certain gases (like formaldehyde) behave. The authors showed that two reactions in the atmosphere were "co-producing" (running in lockstep), which created this hidden rule. This confirmed the computer wasn't making a mistake; it had found a real, physical law that the old textbooks missed.

2. Random Adsorption (The "Parking" Game): Imagine a long parking lot where cars (molecules) of a specific length try to park randomly. Once a car parks, it blocks that space forever.

   • The paper shows that in this "one-way" process, there are hidden rules about the average number of empty spaces left between cars.
   • By merging the "parking events" that happen in sync, they found new rules that predict exactly how full the parking lot will get when it's jammed.

The Bottom Line

The paper argues that the old way of counting rules in chemical systems was incomplete because it treated every reaction as unique.

The new insight: If two irreversible reactions are running in perfect sync, they are actually just one reaction in disguise. When you spot these "synchronized pairs" and merge them, you unlock a new set of conservation laws.

This doesn't just fix a math problem; it gives scientists a better toolkit to understand complex systems, from the air we breathe to how molecules stick to surfaces, by revealing the hidden "synchronized dances" that govern them.

Technical Summary

Technical Summary: Emergent Conserved Quantities via Irreversibility

Problem Statement
The inference and analysis of complex chemical reaction networks (CRNs) and Markov chains rely heavily on identifying conserved quantities. While classical structural laws, specifically the "first structural law" (SL1) derived from the rank of the stoichiometric matrix, provide a method to count standard conserved quantities (left nullvectors of the stoichiometric matrix), these methods are known to undercount the total number of conserved quantities in certain systems. This discrepancy poses a significant obstacle for machine learning (ML) efforts aimed at discovering physical conservation laws from data. A specific instance of this problem involves a recent ML study that identified a candidate non-integer conserved quantity in an atmospheric chemistry model which could not be theoretically explained by existing dynamical frameworks. The paper addresses the gap between the number of conserved quantities predicted by classical linear algebra and those actually observed in systems containing irreversible reactions.

Methodology
The authors propose a structural extension to CRN theory based on the concept of "co-production." The methodology involves:

1. Reaction Merging: Identifying pairs of irreversible reactions that possess pairwise proportional rates (collinear currents). These reactions are mathematically merged into a single effective reaction with a weighted stoichiometry and a single effective current.
2. Effective Stoichiometric Matrix ($S^*$): Constructing a new stoichiometric matrix, $S^*$, representing the system after all possible pairwise proportional reactions have been merged. This matrix contains only independent currents.
3. Derivation of the Co-production Law: Applying the fundamental theorem of linear algebra to $S^*$ to derive a relationship between the number of merged reactions ($\Upsilon$), the change in the number of conserved quantities ($\Delta \ell$), and the change in the number of cycles ($\Delta c$). The law is formulated as:
   $$\Upsilon = \Delta \ell - \Delta c$$
   where $\Upsilon$ is the number of pairwise mergers performed.
4. Application to Specific Models: The framework is applied to two distinct domains:
   • Atmospheric Chemistry: Re-analyzing a specific atmospheric CRN model where an unexplained conserved quantity (CQ3) was previously detected numerically.
   • Markov Processes: Applying the theory to stochastic processes, specifically $n$-mer random sequential adsorption (RSA) and biased diffusion models, to derive emergent conservation laws governing population means.

Key Contributions and Results

• The Co-production Law: The paper establishes that irreversible reactions can lead to "emergent" conservation laws that are not visible in the original stoichiometric matrix $S$ but appear in the merged matrix $S^*$. These laws arise because linearly dependent currents (collinear reactions) reduce the effective dimensionality of the reaction space.
• Resolution of the Atmospheric Conundrum: The authors demonstrate that the mysterious non-integer conserved quantity (CQ3) found in the atmospheric model is a direct consequence of co-production. By merging two collinear reactions involving formaldehyde ($HCHO$) photolysis, the resulting effective stoichiometric matrix yields a left nullvector with non-integer coefficients that exactly matches the numerically discovered quantity. This confirms CQ3 is a genuine conservation law, not an artifact or approximation.
• Emergent Laws in Markov Chains: The study shows that co-production is a generic feature of Markov models with irreversible transitions. In the context of $n$-mer adsorption on a lattice, the authors derive $n$ distinct co-production conservation laws. These laws constrain the ensemble averages of gap populations and allow for the analytical calculation of stationary gap distributions and jamming coverages without solving the full kinetic equations.
• Extension of Structural Laws: The work extends the widely used index law for CRNs. It clarifies that the standard count of conserved quantities ($\ell$) is a lower bound when irreversible reactions with proportional rates exist. The true number of linear conserved quantities is $\ell + \Delta \ell$, where $\Delta \ell$ is determined by the co-production index.

Significance and Claims
The paper claims to provide a theoretical foundation for conservation laws that have previously been discovered only through numerical or machine learning methods but lacked a structural interpretation. By introducing the concept of co-production, the authors:

• Resolve the discrepancy between classical structural laws and observed conservation laws in systems with irreversible reactions.
• Furnish new tools for the inference and analysis of models based on conservation laws, particularly in fields like atmospheric chemistry, photochemistry, and radiochemistry where irreversible processes are prevalent.
• Extend the corpus of current-current relations in stochastic thermodynamics beyond stationary states and Markov processes, offering a structural explanation for emergent conservation laws arising from the irreversibility of transitions.
• Demonstrate that these emergent laws apply to the ensemble averages of stochastic processes (expectation values) rather than individual stochastic trajectories, providing a deterministic constraint on the statistical behavior of the system.

The authors conclude that the inclusion of co-production is essential for correctly counting linear conserved quantities and that this framework bridges the gap between data-driven discovery and first-principles physical interpretation.