Indexed text attachment
evolutionary-game-theory-in-higher-categories.tex
Source for Evolutionary Game Theory in Higher Categories
View paper · Download original file
Lines 2521–2620 of 3949. Source text is displayed literally.
2521and identity maps preserve all the specified data. Products of2522these maps preserve the sequential and parallel response2523formulas. For the sequential case, play preservation gives2524the same intermediate incumbent state and coplay preservation2525gives the same transported continuation. One can then apply2526response preservation separately in the two games. The parallel2527argument uses preservation of the two incumbent outputs.2528This yields a bicategorical description after retaining the2529canonical product reassociations as invertible cells.2530For a more general morphism allowing lenses between the2531boundary interfaces, see the double category of open games2532in \cite{hedges-morphisms}.2533 2534For sequential composition, the diagram below holds at every2535upstream state \(x\). Set \(E_x(\sigma,\tau)=2536P_{\mathcal H}(\tau,P_{\mathcal G}(\sigma,x))\), where2537\(Z\) is the final forward output space. Primes denote the2538corresponding maps for the target games.2539\[2540\begin{tikzcd}[column sep=large,row sep=large]2541\Sigma_{\mathcal G}\times\Sigma_{\mathcal H}2542 \arrow[r,"E_x"{above}]\arrow[d,"\phi\times\psi"']2543 & Z\arrow[d,equal]\\2544\Sigma_{\mathcal G'}\times\Sigma_{\mathcal H'}2545 \arrow[r,"E'_x"{above}]2546 & Z2547\end{tikzcd}2548\]2549The backward returns commute by coplay preservation. These2550two identities compare the boundary behavior independently of2551whether a pair of strategies satisfies the response tests.2552 2553Response-preserving maps carry equilibria to equilibria:2554one applies the implication to \((\sigma,\sigma)\).2555A noninvertible map can identify equilibria or fail to reach2556some equilibria of the target. An isomorphism of games2557instead preserves and reflects the response tests in every2558context. The geometric information carried by a morphism is2559therefore different from a claim about the size or structure2560of its image on equilibrium sets. This distinction matters2561when strategy representations are quotiented or multiple2562agents share an interface.2563 2564\section{Probability and continuation transport}2565 2566For a fixed strategy, one may model forward play by a2567probability kernel. For finitely supported probabilities2568\(P(y\mid x)\) and \(Q(z\mid y)\), their forward composite is2569\[2570 (QP)(z\mid x)=\sum_y P(y\mid x)\,Q(z\mid y).2571\]2572This has unit given by a point mass at the input. Associativity2573is the reassociation of the joint sum over intermediate values.2574If a continuation takes values in \(\mathbb R\), this kernel2575transports it by expectation. Indeed2576\[2577 \sum_y P(y\mid x)\sum_z Q(z\mid y)k(z)2578 =\sum_z (QP)(z\mid x)k(z).2579\]2580Thus a component that compares expected returns can use the2581transported function of its possible inputs. Existence of a2582maximizing strategy is again an additional hypothesis.2583 2584Stochastic parallel composition requires a decision about joint2585laws. Tensoring two kernels using independent randomness gives2586the product law conditional on their inputs. That tensor does2587not have the universal property of a cartesian product in2588stochastic maps. A joint distribution is generally not determined2589by its two marginals. For example, a nondegenerate finite law2590paired with itself by sharing its sample and the product of2591two independent copies of the law have the same marginals.2592The continuation may detect their dependence. Consequently2593continual substitution at the level of marginal expected2594payoffs alone can lose information relevant for later decisions.2595Bayesian open games supply additional structure for stochastic2596play, hidden information and best responses2597\cite{bolt-bayesian}. The calculation above describes only the2598forward-kernel and expected-return part of that framework.2599 2600\section{An equilibrium interface and a revision rule}2601 2602Response relations are evaluated at a context rather than at a2603time. To extract a trajectory one supplies a protocol for when2604revisions occur, which states agents observe, and how allowed2605revisions become a state update. For a fixed context, let2606\(B(\sigma)\) denote the set of responses to \(\sigma\).2607If it is nonempty at every incumbent, a deterministic revision2608rule can choose a single \(U(\sigma)\in B(\sigma)\) at each2609incumbent. A fixed point of \(U\) is an equilibrium because its2610selected response is itself. The converse holds if the rule2611retains the incumbent whenever \(\sigma\in B(\sigma)\).2612The relation alone leaves this decision open, as well as2613the choice among other responses and the possible regularity2614of the selector. A protocol can instead attach a probability2615law to responses, which adds another layer of dynamical data.2616 2617In continuous time a smooth population2618vector field \(F\) and its positive rescaling \(aF\) by a smooth2619function \(a\) have the same rest points. On any interval over2620which a solution of one field exists and the scalar remains