Indexed text attachment
evolutionary-game-theory-in-higher-categories.tex
Source for Evolutionary Game Theory in Higher Categories
View paper · Download original file
Lines 1129–1228 of 3949. Source text is displayed literally.
1129and let \(b_j(x)\ge0\) be an absolute birth rate. With fitness1130used in these units the normalized equation is1131\begin{equation}1132 \dot x_i=\sum_j x_jb_j(x)R_{ji}1133 -x_i\sum_j x_jb_j(x).1134 \label{eq:birth-mutation}1135\end{equation}1136This follows by normalizing the abundance equation with1137incoming births \(\dot n_i=\sum_jn_jb_j(x)R_{ji}\). Setting1138\(R\) equal to the identity recovers selection, and the sum1139of the velocities on the simplex is zero. Under the same1140local Lipschitz hypothesis, forward invariance follows by1141the incoming-rate argument. For that argument the rates1142can be extended off the simplex by their positive parts so1143that incoming forcing remains nonnegative. For the diagonal1144term the coefficient is \(b_iR_{ii}-\sum_jx_jb_j\).1145 1146A common baseline added to \(b_j\) changes1147\eqref{eq:birth-mutation} by \(c\bigl(R^{\mathsf T}x-x\bigr)\).1148This describes mutation occurring during additional births.1149In \eqref{eq:additive-mutation}, the corresponding baseline1150change to \(F\) has no effect. The two conventions therefore1151require different information about which transformations of1152fitness preserve the model. The formulation of an evolution1153operator should specify where mutation enters before passing1154to equivalence classes of payoff data \cite{hofbauer-sigmund}.1155 1156\section{Time changes and the strength of mutation}1157 1158Multiplying a fitness assignment by a positive constant \(a\)1159changes the replicator solution from \(x(t)\) to \(x(at)\).1160The geometric trajectories are the same and their1161parametrizations differ. The same rule holds for a positive1162continuous scalar \(a(x)\) on the simplex. Along a trajectory1163of the rescaled field, the clock for the original field has1164rate \(a(x(t))\). On the compact simplex, the positive1165minimum of \(a\) ensures that this clock advances without1166bound as physical time grows. Thus this kind of rescaling1167preserves forward limit sets as well as rest points.1168 1169In the additive mutation model, the same clock rescales1170both selection and the generator. Holding \(Q\) fixed while1171rescaling \(F\) can change the stationary states. More1172precisely, if \(x\) is stationary for both fitness \(F\)1173and fitness \(aF\) with the same \(Q\) and a constant1174\(a\ne1\), subtraction of the two stationary equations gives1175\(R_F(x)=0\). The original equation then gives1176\(Q^{\mathsf T}x=0\). Conversely, a common zero of these two1177fields is stationary for every relative scaling. When \(Q\)1178is irreducible this test picks out at most its unique1179stationary distribution \cite{hofbauer-sigmund}. Consequently,1180fitness units and mutation rates carry joint information1181when the model includes both mechanisms.1182\par1183\chapter{Invasion and Lyapunov geometry}1184\label{ch:lyapunov}1185 1186\section{Relative entropy near a resident state}1187 1188A state \(p\in\Delta(S)\) is locally superior for \(F\) if1189\[1190 (p-x)\cdot F(x)>01191\]1192for every distinct \(x\) sufficiently close to \(p\) in the1193simplex. This formulation compares the resident with a1194nearby population in the environment of that population.1195For symmetric matrix games it is equivalent to the usual1196evolutionarily stable strategy condition discussed in1197Chapter~\ref{ch:ess}. The implication to dynamical stability1198is the classical result studied by Hofbauer, Schuster, and1199Sigmund \cite{hofbauer-schuster-sigmund,alexander}.1200 1201Set \(U_p=\{x\in\Delta(S):x_i>0\text{ whenever }p_i>0\}\).1202The quantity1203\[1204 D(p\Vert x)=\sum_{i:p_i>0}p_i\log\frac{p_i}{x_i}1205\]1206is finite on \(U_p\), including when \(p\) lies in a proper1207face. The bound \(-\log z\ge1-z\) at \(z=x_i/p_i\) yields1208\[1209 D(p\Vert x)\ge1-\sum_{i:p_i>0}x_i\ge0.1210\]1211Equality demands \(x_i=p_i\) on the support of \(p\), which1212also forces the remaining entries of \(x\) to vanish.1213The derivative along \eqref{eq:replicator} is1214\begin{equation}1215 \frac{d}{dt}D(p\Vert x(t))1216 = x(t)\cdot F(x(t))-p\cdot F(x(t)).1217 \label{eq:kl-dissipation}1218\end{equation}1219The cancellation uses \(\sum_i p_i=1\). Support preservation1220makes the logarithms legitimate for all finite times1221whenever they are defined initially.1222 1223\begin{theorem}[Local stability from superiority]1224\label{thm:superiority-stability}1225For the replicator equation of Chapter~\ref{ch:replicator},1226a locally superior state \(p\) is a locally asymptotically1227stable rest point relative to \(\Delta(S)\).1228\end{theorem}