CharlesMarg
summary
Describes a new Stan 2.17 feature: a Stan 2.17 nonlinear algebraic solver that finds steady states by root finding instead of long simulation and returns derivatives via an implicit-function Jacobian. Demonstrates embedding numerical ODE solves with integrate_ode_rk45 inside the algebraic function, exposes tuning knobs (relative tolerance, function tolerance, max steps), notes a prior limitation on initial guesses now fixed as parameter-dependent initial guesses in 2.18, and shows models that run and fit data.
AI-written summary · about
chapters (10)
- 0:00Nonlinear algebraic solver
- 0:58Drug diffusion
- 4:20Steady State
- 6:01Goal
- 6:46Algorithm
- 8:51Jacobian
- 10:38Pharmacometrics
- 11:42Algebraic Solver
- 13:42Stand Model
- 13:56Bill Gillespie
related talks
28mStan: why does it exist? when is it useful? why do I use it?
32mData-Driven Inverse Modeling with Incomplete Observations
34mFast Algorithms for Multivariate ODEs arising in Sampling and Learning
29mPartial Evaluation of Automatic Differentiation for Differential-Algebraic Equati...
01h 01mRobustness, Stochastics, Uncertainty 2
16mAlgorithmic Differentiation: C++ & Extremum Estimation
33mFinding Multiple Solutions of ODEs with Neural Networks
27mTalk by Ali Kavis (UT Austin)