High performance ordinary differential equation (ODE) and differential-algebraic equation (DAE) solvers, including neural ordinary differential equations (neural ODEs) and scientific machine learning (SciML)
523
stars
199
forks
source link
DFBDF performs poorly when integrating `sin(t)` #2130
Open
topolarity opened 5 months ago
Describe the bug 🐞
DFBDF appears to solve the system
dx ~ y, y ~ sin(t)
very poorly.Minimal Reproducible Example 👇
The solution with
abstol = 1e-3, reltol = 1e-3
is also noticeably bad:A few things jump out to me here:
u[2]
seems to be solved very precisely but for the previous timestep (for the first few time points at least)u
is about[2e-5, 5e-5]
which is ~one and a half orders of magnitude worse than the tolerancesEnvironment:
using Pkg; Pkg.status()