| FazBrowse GitHub Viewer | Trending | | Home |
| Tools: [Download Repo ZIP] [Original HTTPS Page] |
Records FinancePy's Heston EulerLog MC scheme on an AADC tape and replays all paths in one batch call, producing exact pathwise Greeks w.r.t. all 6 model parameters (S0, v0, kappa, theta, sigma, rho). 200K paths + 6 Greeks in 0.07s vs 9.7s for finite differences (144x). All Greeks verified against FD on same paths (ratio 0.998-1.02). Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
|
Looks good. Give me some time to look at this. |
Sorry, something went wrong.
| Back | FazBrowse Home | New Git URL |
Summary
Adds an example showing how AADC (Automatic Adjoint Differentiation) can compute exact pathwise Greeks for FinancePy's Heston Monte Carlo simulation.
The example records the EulerLog MC scheme on an AADC tape and replays all paths in a single batch call, producing exact gradients w.r.t. all 6 model parameters (S0, v0, kappa, theta, sigma, rho) at negligible extra cost.
Results (200K paths, 50 steps)
All Greeks verified against finite differences on the same paths (ratio 0.998–1.02).
Key technique
Dependencies
🤖 Generated with Claude Code