Independent scholarship · Open accessThe monthly digest

Indexed text attachment

evolutionary-game-theory-in-higher-categories.tex

Source for Evolutionary Game Theory in Higher Categories

View paper · Download original file

Lines 2249–2348 of 3949. Source text is displayed literally.

2249The lens \((\operatorname{id}_X,(x,s)\mapsto s)\) is a two-sided identity.2250This gives an ordinary category. The product of two lenses uses2251cartesian products on both types and applies the two maps separately.2252 2253A continuation on the target is an arbitrary function \(k:Y\to R\).2254An input at the source is an element \(x\in X\). A lens sends an2255input forward by \(v\) and transports a continuation backward by2256\[2257           k\longmapsto \bigl(x\longmapsto u(x,k(v(x)))\bigr).2258\]2259The backward transport reverses composition by direct substitution.2260Keeping \(k\) as a function matters. A single value \(k(v(x))\) gives2261a return only at the chosen output. A component deciding whether to2262deviate may need the value at a different output.2263 2264\section{Strategies and contextual responses}2265 2266An open game \(\mathcal G:(X,S)\to(Y,R)\) consists of a strategy set2267\(\Sigma_{\mathcal G}\), a play map \(P_{\mathcal G}\), a coplay map2268\(C_{\mathcal G}\), and best-response relations:2269\[2270\begin{split}2271 P_{\mathcal G}&:\Sigma_{\mathcal G}\times X\to Y,\\2272 C_{\mathcal G}&:\Sigma_{\mathcal G}\times X\times R\to S,\\2273 B_{\mathcal G}(x,k)&\subseteq \Sigma_{\mathcal G}\times2274                              \Sigma_{\mathcal G}.2275\end{split}2276\]2277Here \(k:Y\to R\). The strategies are chosen for the whole2278component. Fixing a strategy \(\sigma\) gives a lens with maps2279\(P_{\mathcal G}(\sigma,-)\) and \(C_{\mathcal G}(\sigma,-,-)\).2280A pair \((\sigma,\sigma')\in B_{\mathcal G}(x,k)\) means that2281\(\sigma'\) is an admissible response to the incumbent \(\sigma\)2282in this context. The equilibrium set consists of the strategies2283\(\sigma\) with \((\sigma,\sigma)\in B_{\mathcal G}(x,k)\).2284An open game in this generality specifies responses as data.2285Its response relation can be constructed from an optimization2286problem, a restricted revision protocol, or another stated selection2287rule \cite{ghani-open-games}.2288 2289For a decision component with observed type \(X\) and choice type2290\(Y\), take \(\Sigma=Y^X\) and \(P(\sigma,x)=\sigma(x)\). With real2291return type and backward source type \(1\), the standard maximizing2292relation allows \((\sigma,\sigma')\) if2293\[2294        k(\sigma'(x))\ \ge\ k(y)\qquad\text{for every }y\in Y.2295\]2296The map \(C\) is the unique map to \(1\). Existence of a response2297requires attainment of a maximum. At a given observed \(x\) this2298relation constrains the strategy only at \(x\). A stronger requirement2299at all possible observations needs another equilibrium test, as in2300subgame analysis. The distinction is visible in the typing of2301\(B(x,k)\).2302 2303\section{Sequential composition}2304 2305Let \(\mathcal H:(Y,R)\to(Z,Q)\) follow \(\mathcal G\), with outer2306continuation \(k:Z\to Q\). An incumbent \((\sigma,\tau)\) determines2307the intermediate forward value \(y=P_{\mathcal G}(\sigma,x)\).2308The downstream strategy determines the continuation seen upstream:2309\[2310  k_{\tau}(y')=2311  C_{\mathcal H}\bigl(\tau,y',k(P_{\mathcal H}(\tau,y'))\bigr),2312  \qquad y'\in Y.2313\]2314The upper path carries states forward. The lower arrow carries2315the continuation back across the downstream component.2316\[2317\begin{tikzcd}[column sep=large]2318 X \arrow[r,"{P_{\mathcal G}(\sigma,-)}"{above}] & Y2319       \arrow[r,"{P_{\mathcal H}(\tau,-)}"{above}] & Z\\2320 & R^Y & Q^Z \arrow[l,"k\mapsto k_\tau"{above}]2321\end{tikzcd}2322\]2323Here \(Q^Z\) and \(R^Y\) denote sets of continuations.2324A composite strategy is a pair. Its play and coplay are the2325composition of the fixed-strategy lenses, and its response relation is2326\[2327\begin{aligned}2328 &\bigl((\sigma,\tau),(\sigma',\tau')\bigr)2329                     \in B_{\mathcal H\circ\mathcal G}(x,k)\\2330 &\quad\Longleftrightarrow\2331   (\sigma,\sigma')\in B_{\mathcal G}(x,k_\tau)2332   \quad\text{and}\quad2333   (\tau,\tau')\in B_{\mathcal H}(y,k).2334\end{aligned}2335\]2336The two tests use the incumbent context. The second component sees2337the output generated by \(\sigma\). The first component evaluates2338alternatives using \(\tau\) as a fixed strategy, including its play2339and coplay at changed inputs. This is a prescription for testing2340responses with the other components' strategies held fixed.2341 2342\begin{proposition}[Associativity of the response construction]2343Sequential composition is associative under the canonical2344reassociation of strategy products. Singleton-strategy identity2345lenses with the sole identity response give identities.2346\end{proposition}2347 2348\begin{proof}