Richard Bellman

Richard Bellman (1920–1984) was an American mathematician who developed dynamic programming and made major contributions to control theory, optimization, and applied mathematics.

Dynamic programming: An optimization method that solves a complex problem by combining solutions to smaller, related subproblems. Bellman developed this framework and named it in the 1950s.
Stochastic control: The optimization of systems whose dynamics or observations involve randomness. Bellman extended dynamic programming to decisions under uncertainty.
Richard Courant: A German-American mathematician known for work in analysis, mathematical physics, and applied mathematics. Bellman studied at New York University’s Courant Institute, where Courant shaped applied mathematics.
World War II operations research: The application of quantitative analysis to military planning and resource decisions during World War II. Wartime analysis exposed Bellman to the practical need for systematic optimization.
Curse of dimensionality: The rapid growth in computational cost as the number of variables or dimensions increases. Bellman coined the phrase to describe a major obstacle to applying dynamic programming.
Bellman equation: A recursive equation expressing the value of a decision problem through its immediate reward and future value. It gives mathematical form to Bellman’s principle of optimality.
Reinforcement learning: A machine-learning approach in which an agent learns decisions through interaction and reward. Its value functions and learning algorithms draw directly on Bellman’s equations.
Solomon Lefschetz: A mathematician known for major contributions to algebraic topology and for leading Princeton’s mathematics department. Bellman completed his doctorate at Princeton under Lefschetz’s supervision.
RAND Corporation: An American nonprofit research organization established to support analysis of military and public policy problems. Bellman’s early postwar research at RAND provided a setting for developing dynamic programming.
Approximate dynamic programming: Methods that approximate value functions or policies when exact dynamic programming is computationally infeasible. It addresses the computational limits Bellman identified for high-dimensional problems.

Linked from 4 pages