Indexed text attachment
evolutionary-game-theory-in-higher-categories.tex
Source for Evolutionary Game Theory in Higher Categories
View paper · Download original file
Lines 1521–1620 of 2037. Source text is displayed literally.
1521chosen in proportion to fitness, places its offspring at $j$ with probability $w_{ij}$, where $W$ is a1522stochastic matrix.1523 1524\begin{theorem}[Isothermal theorem]\label{thm:iso}1525\Sli[\cite{LHN05}] The fixation probability of a single mutant of fitness $r$, placed uniformly at random,1526equals the Moran value $(1-r^{-1})/(1-r^{-N})$ for every $r$ if and only if $W$ is doubly stochastic.1527\end{theorem}1528 1529\noindent A doubly stochastic matrix is a stochastic map that preserves the uniform distribution, so the1530isothermal graphs are the endomorphisms of the probability space $(V,\text{uniform})$ in $\FinStoch$. For them1531the number of mutants is itself a birth--death chain with ratio $r$ (\Cref{exer:iso}), and structure is1532invisible to selection: on the complete graph and on the cycle the computed fixation probabilities agree1533with the Moran value to ten digits. \chk{L4} Other graphs change the balance between selection and drift.1534\begin{center}1535\begin{tabular}{@{}lccc@{}}1536\toprule1537$N=6$ & well-mixed & star & directed line\\1538\midrule1539advantageous mutant, $r=1.5$ & $0.3654$ & $0.4432$ & $0.1667$\\1540disadvantageous mutant, $r=0.7$ & $0.0571$ & $0.0333$ & $0.1667$\\1541\bottomrule1542\end{tabular}1543\end{center}1544The star amplifies selection in both directions, and on the directed line only a mutant at the root can1545take over, so fixation is $1/N$ whatever the fitness. \chk{L5} The values are exact, from the linear system1546for the absorption probabilities on all $2^6$ subsets.1547 1548\section*{Notes}1549 1550Local-interaction models of this kind go back to Nowak and May \cite{NoMa92}; the source's examples are in the1551tradition of Skyrms's and Alexander's work on the evolution of social structure \cite{Sky04,Alex07}. Evolutionary1552graph theory, the isothermal theorem, amplifiers and suppressors are from \cite{LHN05}. The source ends with a1553simulation of the centipede game on a lattice, in which the socially optimal outcome occurs $13.3$ per cent of1554the time, close to the figure of about $15$ per cent in the experiments of McKelvey and Palfrey1555\src{\S4.2.3}; we have not reproduced it, since it depends on the distribution of initial conditions, which the1556functorial statements above do not see.1557 1558\section*{Exercises}1559 1560\begin{exercise}\label{exer:cover}1561Show that reduction modulo $n$ is a covering $C_{kn}\to C_n$ for $n\ge3$ and not for $n=2$.1562\end{exercise}1563 1564\begin{exercise}\label{exer:twist}1565Verify that the configurations on $C_9$ fixed by the twisted symmetry of \Cref{prop:ring} number $27$, that1566each uses every strategy exactly three times, and that $RPS\,RPS\,RPS$ is a fixed point of $U$ that is not1567among them.1568\end{exercise}1569 1570\begin{exercise}\label{exer:ring2}1571Repeat the analysis of \Cref{prop:thresholds} for the simple ring, and recover the source's two examples.1572\end{exercise}1573 1574\begin{exercise}\label{exer:iso}1575Show that for doubly stochastic $W$ the ratio of the probabilities that the number of mutants increases and1576decreases is $r$ in every state. Why is the star not isothermal?1577\end{exercise}1578%% ---- end ch/07-local.tex ----1579\part{Assessment}1580%% ---- begin ch/08-coda.tex ----1581\chapter{Assessment}\label{ch:coda}1582 1583\section{What the language did}1584 1585Five things, in decreasing order of how much they changed our understanding of the source.1586 1587\emph{It isolated the invariance that separates the dynamics.} The replicator equation is natural for all1588payoff-preserving maps (\Cref{thm:natural}); the Brown--von Neumann--Nash and Smith dynamics are natural only for1589symmetries (\Cref{prop:names}). The survival of dominated strategies, which the source presents as a surprising1590theorem about a class of dynamics, becomes the visible cost of giving up naturality for inclusions and clones:1591a dynamic that sees names instead of individuals keeps reintroducing a bad name.1592 1593\emph{It showed why stability must be a property of sets.} The source moves from evolutionarily stable strategies1594to evolutionarily stable sets as a refinement motivated by examples. \Cref{thm:esset} gives a reason that is1595independent of examples: stability of a point is destroyed by duplicating a strategy, stability of a set is1596preserved, and no notion that depends on how strategies are named can be the right one.1597 1598\emph{It explained simulated patterns.} The rotating clusters on the ring are forced by a twisted symmetry1599and a covering (\Cref{prop:ring}, \Cref{thm:covering}); the growth of cooperation on the circulant is three1600linear inequalities (\Cref{prop:thresholds}). Both were found by asking what the maps of the model are.1601 1602\emph{It separated the game from the learner.} Treating a population as an open system1603(\Cref{prop:softmax}, \Cref{rem:wiring}) makes ``same game, different dynamics'' and ``same dynamics,1604different structure'' into statements about a diagram and its fillings, and it is the form in which the1605subject is used in machine learning, as the companion paper \cite{EGML} develops.1606 1607\emph{It located a non-commuting square.} The source's claim that the replicator dynamics is not the limit of the1608birth--death process is a statement about two orders of limits (\Cref{rem:limits}).1609 1610\section{What it did not do}1611 1612It proved no new theorem of evolutionary game theory. The observations tagged as such are short, and some of1613them are surely known to people who work with these models, whether or not they are written down in this1614form. The fixation formulas, the mean-field limit and the one-third law gain nothing from being called1615morphisms in $\FinStoch$. Chaos in the replicator equation, which the source illustrates with a four-strategy1616example \src{\S3.2.5}, is untouched: we have no categorical handle on it beyond the remark that a strange1617attractor is a chain component with a non-trivial Conley index. Bomze's classification of the phase1618portraits for three strategies is likewise only cited. The centipede simulation is not reproduced. And the1619source's philosophical conclusion, that these models give how-possibly explanations whose robustness must be1620argued case by case \src{\S5}, is one that the present treatment supports in a specific way and does not