Enforcing values in ODE solver

Question

Pulkit 2024-2-27

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/2087771-enforcing-values-in-ode-solver

评论： Pulkit 2024-2-29

I am trying to solve a stiff ode with ode15s. There is a variable for which there exists a physical constraint due to which it cannot exceed a certain maximum (the allowed maximum also changes with other variables).

I could use events to stop the ode integration and reset the variable but there is a regime where the event persists for some time so, I don't know if exiting the solver each time after just one iteration is efficient.

I also tried using global variables, so that I can always manually control it but then it becomes unphysical when solver rejects a step and goes back.

Is there a way to apply the constraint within the solver? Is there some other solver that could help or would it be better to write one by myself?

Thanks!

1 个评论
显示 -1更早的评论隐藏 -1更早的评论

Torsten 2024-2-28

编辑：Torsten 2024-2-28

I could use events to stop the ode integration and reset the variable but there is a regime where the event persists for some time so, I don't know if exiting the solver each time after just one iteration is efficient.

This is only possible if you define the event as a discontinuous function, and this is not allowed. An allowed event function has a zero crossing without remaining in the state.

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Walter Roberson 2024-2-27

1
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/2087771-enforcing-values-in-ode-solver#answer_1417656

Use events, and remember that events are only detected when a value crosses the trigger, and are not detected when the system starts in that state.

1 个评论
显示 -1更早的评论隐藏 -1更早的评论

Pulkit 2024-2-29

Oh in that case, I don't think i can make use of built in solvers. The variable reaching a maximum is not a one-time event but a whole regime that would persist for some time. Thanks for the reply.

请先登录，再进行评论。