Game Theory Explained A Mathematical Introduction with Optimization (Christopher Griffin) (Z-Library)

【One-Line Pitch】 A rigorous, self-contained mathematical introduction to game theory that pairs classical concepts—from probability and utility theory to game trees and Nash equilibria—with optimization techniques, ideal for upper-level undergraduates or self-learners comfortable with calculus and matrices who want a proof-based foundation rather than a pop-science overview. 【Book Arc】 - **Opening (~0%–10%)**: Sets the stage with a preface explaining the book's design for a one-semester course, then dives into probability theory through the lens of "games against the house" (e.g., roulette, blackjack). This stage builds the probabilistic toolkit—sample spaces, events, conditional probability—needed for later game analysis. - **Early (~10%–23%)**: Introduces elementary utility theory, formalizing how players value uncertain outcomes via lotteries and the expected utility theorem. This bridges probability to decision-making, establishing the preference axioms (like von Neumann–Morgenstern) that underpin rational choice in games. - **Early (~23%–32%)**: Moves to game trees and extensive-form games, defining directed graphs, paths, and trees as the structural backbone. It covers perfect-information games (e.g., rock-paper-scissors with sequential moves) and then generalizes to incomplete information via information sets, showing how strategies are mappings from decision nodes or information sets to actions. - **Middle (~32%–48%)**: Deepens extensive-form analysis with chance moves, strategy spaces, and payoff functions, including worked examples like the Battle of the Bismark Sea and simplified poker. It also sketches an inductive proof of equilibrium existence in finite perfect-information games, then transitions to normal-form (strategic) games, where matrices become the primary representation and equilibrium definitions are restated. - **Middle (~48% onward)**: The book pivots to optimization as a core tool, with the preface highlighting self-contained coverage of Karush–Kuhn–Tucker conditions and quadratic programming for finding Nash equilibria in bimatrix games. This stage connects game theory to mathematical optimization, a theme the author calls underemphasized elsewhere, and sets up cooperative games (Chapter 10) and evolutionary game theory (Appendix C) as advanced extensions. 【Key Takeaways】 - **Probability is the language of games against chance** (Opening): Master sample spaces, events, and conditional probability first—they are prerequisites for modeling uncertainty in games like roulette or blackjack. The book uses real casino examples to make abstract definitions concrete. - **Conditional probability sharpens in-game decisions** (Early): When you condition on observed information (e.g., cards already dealt), you can compute updated odds, like estimating a ≤35% chance of drawing a helpful card in blackjack. This is the practical bridge from theory to strategy. - **Utility theory converts preferences into numbers** (Early): Lotteries and compound lotteries are reduced to simple ones via probability weighting, and the expected utility theorem proves that rational preferences can be represented by a utility function. This justifies treating payoffs as comparable numerical values. - **Game trees model sequential play explicitly** (Early): Extensive-form games use directed trees where nodes are decision points, edges are moves, and information sets capture what players know. A pure strategy is a complete plan—one action per decision node or information set—not just a single move. - **Information sets generalize perfect to imperfect information** (Early): When every information set is a singleton, the game is equivalent to one with complete information; otherwise, players act without full knowledge, as in poker. This distinction is crucial for modeling real strategic uncertainty. - **Equilibrium is a no-regret condition** (Middle): A strategy profile is an equilibrium if no player can improve their payoff by unilaterally changing strategy. This definition recurs across extensive and normal forms, and the book proves existence for finite perfect-information games via induction on tree height. - **Normal form compresses games into matrices** (Middle): Strategic-form games represent players, strategy sets, and payoff functions as a matrix, enabling algebraic analysis. This is the gateway to optimization-based solution methods. - **Optimization is the book's distinctive lens** (Middle): The author emphasizes self-contained coverage of Karush–Kuhn–Tucker conditions and quadratic programming for solving bimatrix games, making advanced solution techniques accessible without prerequisites. This is a rare, underappreciated connection between game theory and optimization. 【Reading Tips】 - **Skim the probability review if you're comfortable** (Opening): Chapters 1–2 cover probability and utility theory; if you already know conditional probability and expected value, skim examples but don't skip the utility axioms—they underpin later payoff assumptions. - **Deep-read the game tree chapter** (Early): This is the conceptual core for extensive-form games. Work through the rock-paper-scissors and Battle of the Bismark Sea examples by hand, drawing trees and labeling information sets, to internalize how strategies are defined. - **Treat the equilibrium proof as a template** (Middle): The induction proof for equilibrium existence in perfect-information games is worth studying line-by-line—it's a model for how game theory arguments are structured, and it clarifies why finite games always have solutions. - **Use the appendices for math refreshers** (Throughout): Appendix A (matrix arithmetic) and B (calculus) are designed to fill gaps; consult them before the normal-form and optimization sections if you're rusty, rather than struggling through notation. - **Focus on the optimization connection** (Middle–Late): The quadratic programming method for Nash equilibria is a highlight; even if you don't implement it, understand the setup—it shows how game theory and optimization are intertwined, a theme the author champions. 【Coverage Limits】 This guide covers the book's opening through the transition to normal-form games and optimization (roughly the first half). Excerpts do not cover later chapters on cooperative games (Chapter 10) or evolutionary game theory (Appendix C) in detail, nor the full optimization algorithms beyond their introduction.