Approximate Dynamic Programming for Portfolio Selection Problem

Multi-Shot Approximation of Discounted Cost MDPs

Value-gradient iteration for convex stochastic control

Approximate Dynamic Programming assignment solution for a maze environment at ADPRL at TU Munich

Real-Time Ambulance Dispatching and Relocation

Numerical methods implementations from computational methods class

Implements a simulation-based dynamic programming framework for optimal portfolio allocation over multiple periods. Combines analytical solutions with approximate methods under CRRA and log utility, using Monte Carlo simulation, convex optimization, and value function approximation to compute optimal policies across time and wealth levels.

Implementation of Approximate Smooth Kernel Value Iteration