🧬 biology

Asymmetry Demystified: Strict CLFs and Feedbacks for Predator-Prey Interconnections

This paper addresses the challenges of globally stabilizing predator-prey population dynamics with positive states and controls by introducing a novel framework for designing clean, strict Control Lyapunov Functions (CLFs) that generalize classical Volterra-style constructions and enable concurrent feedback design via customized forwarding and backstepping techniques.

Original authors: Miroslav Krstic

Published 2026-02-26

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Miroslav Krstic

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ⚕️ This is an AI-generated explanation of a preprint that has not been peer-reviewed. It is not medical advice. Do not make health decisions based on this content. Read full disclaimer

The Big Picture: Taming the Wild Garden

Imagine you are the gardener of a very delicate ecosystem containing two types of plants: Prey (let's call them "Rabbits") and Predators (let's call them "Foxes").

Your goal is to keep the population stable. You don't want the Rabbits to eat everything and die out, and you don't want the Foxes to starve. You have a tool: a Harvesting Rate (a trap or a net). You can catch and remove animals, but there's a catch: You can only remove animals; you cannot magically add them back. Your "control" must always be positive (you can't "un-catch" a fox).

The problem is that nature is messy.

Extinction is scary: If the population gets too low, the math gets wild and unpredictable.
Overpopulation is chaotic: If there are too many, the dynamics change completely.
The "Strict" Problem: Most mathematicians can prove the system might stabilize using a "fuzzy" proof (like saying, "It will eventually settle down, but I can't tell you how fast or if a tiny mistake will ruin it"). This paper wants a "Strict" proof: a guarantee that it will stabilize, exactly how fast, and that it's robust against small errors.

The author, Miroslav Krstic, is saying: "I found a new, elegant way to build a mathematical 'safety net' (called a Strict CLF) that proves this system will stabilize, even with the tricky rule that you can only harvest, not add."

The Two Scenarios

The paper looks at two different ways you can manage this garden:

Scenario A: The "Fox-Only" Farmer

You can only catch Foxes.

The Challenge: If you catch too many Foxes, the Rabbits explode. If you catch too few, the Foxes eat all the Rabbits.
The Old Way: Previous math used complex, "heavy-handed" techniques (like Matrosov methods) that were hard to understand and didn't give clear insights.
The New Way: The author invents a new formula. Think of it like a specialized scale.
- Standard scales just weigh the total number of animals.
- This new scale weighs the ratio of Foxes to Rabbits.
- It also adds a "barrier" function. Imagine a wall that gets infinitely high if the Rabbits get close to zero. This forces the math to respect the rule that "animals can't be negative."
- Result: A clean, simple formula that proves the garden will stabilize, no matter where you start.

Scenario B: The "Blind" Farmer

You have to catch both Rabbits and Foxes indiscriminately (maybe you're just clearing land).

The Challenge: This is much harder. If you catch a Rabbit, you might starve a Fox. If you catch a Fox, the Rabbits might overpopulate.
The New Way: The author creates a more complex "safety net." It's like a customized shock absorber for a car. It has a specific shape that handles the "bumps" (the asymmetry between catching rabbits vs. foxes) perfectly.
Result: Even though it's harder, the author proves a strict safety net exists here too.

The "Bonus": Designing the Trap and the Map Together

Usually, mathematicians first pick a trap (controller) and then try to find a map (CLF) to prove it works. Sometimes the map doesn't exist.

In the "Bonus" section, the author tries to design the trap and the map at the same time.

The "Forwarding" Attempt (The Wrong Turn):
The author tries a standard engineering trick called "Forwarding." It works great for normal systems, but here, it suggests a trap that sometimes requires negative harvesting.
- Analogy: It's like a recipe that says, "To make the soup perfect, you must add -2 cups of salt." In the real world, you can't add negative salt (you can't un-salt the soup). This solution failed the "positive-only" rule.
The "Backstepping" Success (The Unconventional Genius):
The author tries another trick called "Backstepping." Usually, this method tries to make the system behave like a simple, straight line.
- The Twist: The author realized that for this specific ecosystem, trying to make it "linear" (straight) was the wrong idea. Instead, they embraced the wild, non-linear nature of the predator-prey relationship.
- They designed a trap where the number of animals you catch depends on the square of the Foxes divided by the Rabbits ( $Y^2/X$ ).
- Why it works: This formula is naturally positive (you can't have a negative number of animals squared). It acts like a self-correcting thermostat. If Foxes get too high, the trap opens wider automatically. If Rabbits get low, the trap closes.
- Result: This is the "Small Triumph." It's a simple, elegant rule that guarantees the garden stabilizes, and it never asks you to do the impossible (add animals back).

Why Does This Matter? (The "So What?")

It's Not Just About Animals: While the paper uses Rabbits and Foxes as an example, the math applies to anything where you have two things interacting, you can only remove them (not add them), and they behave very differently when they are rare vs. common.
- Examples: Managing fish stocks, controlling the spread of a virus (where you can only quarantine, not "un-infect"), or managing chemical reactions in a factory.
From "Maybe" to "Definitely": Before this, engineers often had to say, "It probably works, but we can't prove it won't crash if there's a tiny error." This paper provides a Strict proof. It's the difference between saying "The bridge looks sturdy" and "Here is the exact calculation proving the bridge will hold 100 tons."
Elegance over Brute Force: The author shows that you don't need complex, ugly math to solve hard problems. Sometimes, by looking at the problem from a different angle (like looking at the ratio of animals instead of the total), the solution becomes simple and beautiful.

In a Nutshell

The author took a notoriously difficult problem in population control—proving that a system will stabilize when you can only remove, not add, things—and solved it by inventing new, "asymmetric" mathematical tools. These tools act like custom-fitted safety nets that respect the natural limits of the system, proving that stability is possible without needing to break the laws of nature (like adding negative animals).

1. Problem Formulation

The paper addresses the challenge of global asymptotic stabilization for population dynamics systems, specifically predator-prey models, under the constraints of positive states (population concentrations $X, Y \geq 0$ ) and positive inputs (harvesting/removal rates $U \geq 0$ ).

The author identifies two specific benchmark models:

Predator-Only Harvesting:
$\dot{X} = (1 - Y)X, \quad \dot{Y} = (X - U)Y$
Here, the control $U$ only affects the predator population.
Simultaneous Harvesting:
$\dot{X} = (2 - Y - U)X, \quad \dot{Y} = (X - U)Y$
Here, the control $U$ affects both populations simultaneously.

The Core Difficulty:
While global stabilization is achievable using non-strict Lyapunov functions (where $\dot{V} \leq 0$ ), such functions rely on LaSalle's invariance principle. This approach offers:

No quantitative measure of convergence rates.
No robustness guarantees against small modeling errors.
Difficulty in handling the extreme asymmetry of dynamics near extinction (states approaching 0) versus overpopulation.

The goal is to construct Strict Control Lyapunov Functions (CLFs) (where $\dot{V} < 0$ everywhere except equilibrium) that are non-Matrosov (avoiding complex, conservative strictification techniques), non-separable, and capable of handling the positivity constraints of the control input.

2. Methodology

The paper employs a combination of structural analysis and novel Lyapunov construction techniques:

Volterra-Style Functions: The foundation is the classical Volterra function $\Psi(S) = S - 1 - \ln(S)$ , which is positive definite and radially unbounded for $S > 0$ .
Non-Separable Augmentation: Instead of summing independent functions of $X$ and $Y$ , the author constructs CLFs using ratios (e.g., $Y/X$ ) and inverse terms (e.g., $1/X$ ). This creates "barrier" properties that handle the singularity at extinction.
Strictification via Augmentation: The method involves taking a known non-strict CLF ( $V_1$ ) and adding a second term ( $V_2$ ) designed to cancel indefinite cross-terms in the derivative $\dot{V}$ , thereby rendering the total derivative negative definite.
Concurrent Design: In Section V, the paper explores designing the feedback control $U$ and the CLF simultaneously using modified Forwarding and Backstepping techniques. Crucially, these methods are adapted to respect the $U \geq 0$ constraint, deviating from their standard linear-target-system forms.

3. Key Contributions

A. New Strict CLF Constructions (Analysis)

The paper provides explicit, non-Matrosov strict CLFs for two harvesting scenarios:

Predator-Only Harvesting (Theorem 1):
- Controller: $U = Y$ .
- CLF: $V = \Psi(X) + \Psi(Y) + \Psi(1/X) + \Psi(Y/X)$ .
- Result: The derivative is strictly negative: $\dot{V} = -\frac{(X-1)^2}{X} - (Y-1)^2$ . The term $\Psi(1/X)$ handles the asymmetry near prey extinction.
Simultaneous Harvesting (Theorem 2):
- Controller: $U = 1 + \epsilon(Y-1)$ .
- CLF: A complex non-separable function involving $\Psi(X)$ , $\Psi(Y)$ , a new function $\Pi(X)$ , and a ratio term $\Psi(Y/X^{\frac{\epsilon}{1+\epsilon}})$ .
- Result: Proves global asymptotic stability with a strict derivative, overcoming the difficulty of simultaneous harvesting.

B. Concurrent Feedback and CLF Design (Synthesis)

The paper demonstrates that finding a positive control law that works with a global strict CLF is significantly harder than finding a control that works with a non-strict CLF.

Forwarding Approach (Theorem 3): A forwarding-based controller $U = X + Y - X/Y$ is derived. While it yields a strict CLF, the control $U$ becomes negative in certain regions (when prey dominates), making it physically unrealizable for harvesting.
Backstepping Approach (Theorem 4): The author proposes an "unconventional backstepping" controller:
$U = \frac{Y^2}{X}$
- Unlike standard backstepping which targets linear systems, this targets a highly nonlinear system with asymmetric dissipation.
- Significance: This controller is globally positive ( $U > 0$ for all $X, Y > 0$ ) and renders the CLF derivative strictly negative. This is a rare example of a global strict CLF compatible with a positive control constraint.

4. Key Results

Asymmetry Handling: The proposed CLFs explicitly account for the asymmetry between extinction ( $X \to 0$ ) and overpopulation. The functions grow rapidly near zero (barrier functions) and logarithmically at infinity.
Non-Separability: The CLFs are not of the form $V(X) + V(Y)$ . They rely on cross-terms like $Y/X$ and $1/X$ , which are essential for strictification in these bilinear systems.
Positive Control Constraint: The paper highlights that while global stabilization is possible, ensuring the control remains positive ( $U \geq 0$ ) while maintaining a strict CLF is a major hurdle. Most standard designs (like the forwarding example) fail the positivity test. The successful backstepping design ( $U=Y^2/X$ ) is a "small triumph" that overcomes this.
Simplicity: The proofs for the strictness of the new CLFs are concise (often just a few lines), contrasting with the complex Matrosov-based techniques found in previous literature.

5. Significance and Impact

Beyond Population Dynamics: While using predator-prey models as a benchmark, the methodology is intended for any positive-state, positive-input, bilinear system with severe dissipation asymmetries (e.g., chemical reactors, epidemiological models, economic systems).
Robustness: By providing strict CLFs, the paper enables the design of controllers with guaranteed robustness margins and convergence rates, which non-strict CLFs cannot offer.
Methodological Shift: The paper moves away from "separable" Volterra functions and "Matrosov" strictification. It advocates for non-separable, insight-bearing, and explicit constructions that leverage the specific structure of the system (ratios and inverses).
Future Directions: The work suggests a path toward Input-to-State Stability (ISS) CLFs for population dynamics, bridging the gap between theoretical control design and practical ecological management where harvesting cannot be negative.

In summary, Krstic demystifies the construction of strict Lyapunov functions for asymmetric population systems, providing elegant mathematical tools that respect physical constraints (positivity) and offer superior stability guarantees compared to traditional methods.