arXiv:quant-ph/0106135
Equilibria of Replicator Dynamics in Quantum Games
Azhar Iqbal, A. H. Toor
Abstract
We analyse two-population replicator dynamics for asymmetric $2\times2$ games implemented with the Marinatto--Weber (MW) identity/bit-flip protocol. The effective payoff matrices are explicit convex combinations of relabelled classical matrices. Consequently, the dynamics depends on the initial state only through its four computational-basis probabilities: quantum phases, coherences and entanglement cannot be distinguished from classical shared randomness by the payoffs or trajectories in this protocol. At each pure-strategy vertex, the local eigenvalues equal the negatives of the corresponding unilateral payoff advantages; a vertex is asymptotically stable precisely when it is a strict Nash equilibrium, and thus an evolutionarily stable strategy pair in the asymmetric-game sense. Explicit examples show how varying the initial distribution changes which vertex is stable. When an isolated interior rest point exists, it is a saddle or a nonlinear centre according to the sign of the product of the classical payoff cross-differences; the sign cannot change under the MW mapping, although the point can enter or leave the strategy square. We give all four vertex-stability conditions explicitly in a single table, obtain a closed-form logistic solution on the parity-balanced hyperplane $S=0$, classify the degenerate cases, and give a first integral for the nondegenerate flow. All results concern the MW payoff protocol and its classically simulable family of bimatrix games; none constitutes evidence of an inherently quantum evolutionary advantage.
- DOI
- 10.48550/arxiv.quant-ph/0106135
- Comments
- Revised, 14 pages, one figure