You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The integral collocation method implementation of this paper is numerically unstable.
We use the Chebyshev transform class provided by Boost to compute the fixed point operator (Lagrange interpolation at Chebyshev nodes).
The implementation of the ODE solver is located at include/ode_solvers/integral_collocation.hpp.
To Reproduce
Steps to reproduce the behavior:
Simulate the ode x''(t) = -x(t) with x(0) = 0 and x'(0) = 1.
The numerical solver returns a sine function with increasing amplitude.
I believe this has to be with the parameterization of the algorithm.
Expected behavior
The solution is x(t) = sin(t) which is bounded above by 1.
The text was updated successfully, but these errors were encountered:
papachristoumarios
changed the title
Integrall Collocation method is numerically unstable
Integral Collocation method is numerically unstable
Oct 21, 2020
Describe the bug
The integral collocation method implementation of this paper is numerically unstable.
We use the Chebyshev transform class provided by Boost to compute the fixed point operator (Lagrange interpolation at Chebyshev nodes).
The implementation of the ODE solver is located at
include/ode_solvers/integral_collocation.hpp
.To Reproduce
Steps to reproduce the behavior:
Simulate the ode
x''(t) = -x(t)
withx(0) = 0
andx'(0) = 1
.The numerical solver returns a sine function with increasing amplitude.
I believe this has to be with the parameterization of the algorithm.
Expected behavior
The solution is
x(t) = sin(t)
which is bounded above by 1.The text was updated successfully, but these errors were encountered: