Resources
A compact collection of reading notes, slides, and replication code that I use in computational economics and neural methods for dynamic models.
Reading Notes: AI Methods for Dynamic Economic Models
Paper-by-paper notes with derivations and replication attempts, kept as I work through the literature that my own solvers build on. Written in Chinese; hosted publicly on Feishu.
Neural solution methods for dynamic and heterogeneous-agent models
DeepHAM
Estimating Nonlinear Heterogeneous-Agent Models with Neural Networks
Deep Equilibrium Nets
Deep Learning for Solving Dynamic Economic Models (Maliar et al.)
使用深度学习方法求解 HAM 类模型
Solving High-Dimensional Dynamic Programming with Continuous Neural Networks
Continuous time, FBSDE, and HANK
Mean field games
Presentations: AI Methods for Solving HAMs
Code
Replication: Deep Learning for Solving Dynamic Economic Models
Krusell-Smith model replication materials.
Replication: Deep Equilibrium NETs
Benchmark implementation for DEQN-style methods.
Replication: Estimating Nonlinear Heterogeneous Agent Models with Neural Networks
RANK + ZLB replication materials.
High-Dimensional Dynamic Programming with PyTorch
Deep learning implementation for high-dimensional dynamic programming problems.
Aiyagari Model with Transitions
Code notes for transition dynamics in an Aiyagari environment.
CUDA Parallel Aiyagari Solver
CUDA implementation and performance comparison against Matlab and Fortran baselines.