Abstract:This paper studies the rational synthesis problem for multi-player games played on graphs when rational players are following subgame perfect equilibria. In these games, one player, the system, declares his strategy upfront, and the other players, composing the environment, then rationally respond by playing strategies forming a subgame perfect equilibrium. We study the complexity of the rational synthesis problem when the players have {\omega}-regular objectives encoded as parity objectives. Our algorithm is based on an encoding into a three-player game with imperfect information, showing that the problem is in 2ExpTime. When the number of environment players is fixed, the problem is in ExpTime and is NP- and coNP-hard. Moreover, for a fixed number of players and reachability objectives, we get a polynomial algorithm.
What problem does this paper attempt to address?
This paper attempts to solve whether in a multi - player non - zero - sum game, the system (i.e., player 0) can find a strategy such that when the other players in the environment (players 1 to players n) respond rationally, the system can achieve its goals. Specifically, the paper studies the rational synthesis problem when rational players follow Subgame Perfect Equilibrium (SPE). Here, the environmental players will form SPE strategies according to the system's strategy selection.
### Problem Background
In multi - player infinite - duration graph games, each player has their own ω - regular goals (such as parity goals). The paper focuses on how to determine whether there exists a strategy σ₀ for the system player (player 0) such that no matter how the environmental players respond rationally, the system's ω - regular goal can be satisfied. In this case, the environmental players will not be completely adversarial to the system player but will act rationally according to their own goals.
### Main Problem
The core problem of the paper is the **Non - Cooperative Rational Synthesis Problem (NCRS)**, especially using Subgame Perfect Equilibrium (SPE) to define rational behavior. Specifically, the NCRS problem can be described as:
Given a multi - player non - zero - sum game \(G=(G,(W_i)_{i\in\Pi})\), where each player \(i\) has an ω - regular goal \(W_i\), ask whether there exists a strategy \(\sigma_0\) of the system player (player 0) such that for any Subgame Perfect Equilibrium \(\bar{\sigma}_{- 0}\) formed by the environmental players (players 1 to players n), the combined strategy \((\sigma_0,\bar{\sigma}_{-0})\) can ensure that the system player's goal \(W_0\) is satisfied.
### Research Methods
To study this problem, the paper proposes the following methods:
1. **Transform the problem into a Prover - Challenger game**: By introducing two provers (Prover 1 and Prover 2) and a challenger (Challenger), the NCRS problem is transformed into a three - player incomplete - information game (P1CP2 game).
2. **Eliminate incomplete information**: Through specific technical means, the three - player incomplete - information game is converted into a two - player zero - sum game, and finally simplified into a two - player zero - sum perfect - information game with parity goals.
3. **Complexity analysis**: Analyze in detail the computational complexity of each step, prove that the time complexity of solving the SPE - NCRS problem is double - exponential time (2ExpTime), and in the case of a fixed number of players, it belongs to exponential time (ExpTime).
### Conclusion
The main contribution of the paper is to provide a novel algorithm framework for solving the non - cooperative rational synthesis problem and achieve significant improvement in computational complexity. This method not only improves algorithm efficiency but also provides a new perspective for understanding and dealing with rational behavior in multi - player non - zero - sum games.