Differential Equation Approximations for Population Games using Elementary Probability

Semih Kara,Nuno C. Martins

2023-12-11

Abstract:Population games model the evolution of strategic interactions among a large number of uniform agents. Due to the agents' uniformity and quantity, their aggregate strategic choices can be approximated by the solutions of a class of ordinary differential equations. This mean-field approach has found to be an effective tool of analysis. However its current proofs rely on advanced mathematical techniques, making them less accessible. In this article, we present a simpler derivation, using only undergraduate-level probability.

Computer Science and Game Theory,Multiagent Systems,Systems and Control,Dynamical Systems,Probability,Applications

What problem does this paper attempt to address?

The problem that this paper attempts to solve is to simplify the proof of the mean - field approximation in Population Games and provide explicit bounds for the approximation error in the finite case. Specifically, the author aims to re - derive and explain this complex problem using methods that only involve probability theory at the undergraduate level, so that more researchers can understand and apply these results.

### Background and Problem Description

Population games are a framework for modeling strategic interactions among a large number of homogeneous agents. Due to the large number and identical nature of the agents, their collective strategic choices can be approximated as the solution of a class of ordinary differential equations (ODEs). This mean - field - based method has proven to be an effective analytical tool. However, current proofs rely on advanced mathematical techniques, which makes them difficult to be widely understood and applied.

### Main Contributions of the Paper

1. **Simplified Proof**: The author proposes a simpler derivation method, using only probability theory knowledge at the undergraduate level, enabling more people to understand and apply these results.
2. **Providing Explicit Bounds**: For a finite number of agents, the author provides explicit approximation error bounds, rather than just the result that the error approaches zero in the limit case.
3. **Educational Significance**: By simplifying the proof and providing explicit bounds, the author hopes to make this field more popular and easier to understand, especially for researchers without a strong mathematical background.

### Mathematical Expressions

The key formulas mentioned in the paper include:

- Convergence of the mean - field approximation:
\[
\lim_{N \to \infty} P\left( \sup_{t \in [0, T]} \|X^N(t) - x(t)\|_2 \geq \epsilon \right) = 0
\]

- The stronger conclusion proposed by Benaïm and Weibull:
\[
P\left( \sup_{t \in [0, T]} \|X^N(t) - x(t)\|_\infty \geq \epsilon \mid X^N(0) = x_0 \right) \leq 2e^{-\epsilon^2 c N}
\]

- The new error bound proposed in this paper:
\[
P\left( \sup_{t \in [0, T]} \|X^N(t) - x(t)\|_2 \geq \epsilon \mid X^N(0) = x_0 \right) \leq \frac{9 \lambda \ell^*}{4 N \epsilon^2 L_\Phi (e^{2 L_\Phi T} - 1)} + \ell^* H_N\left(\frac{T}{\ell^*}\right)
\]
where:
\[
b := \left( \frac{2 T e^{L_\Phi T}}{\epsilon \left( L_\Phi + \frac{2 \phi_{\max} \lambda}{L_\Phi (e^{2 L_\Phi T} - 1)} \right)} \right)^3
\]
\[
\ell^* := 2 \left\lceil \frac{T e^\lambda}{\epsilon} \right\rceil
\]
\[
H_N(\tau) := e^{-\lambda N \tau} + \frac{1}{2 \left\lceil N \epsilon \right\rceil - 1}
\]

Through these formulas, the author not only simplifies the proof process but also provides specific bounds for the approximation error in the finite case, thereby enhancing the practicality and interpretability of the results.