Save variables while using ode solver

10 views (last 30 days)
[t,y]=ode45(@fun,tspan,yo);
function dy=fun(t,y)
dy=zeros(3,1);
constant1 = k1*y(3)
dy(1)= constant1;
dy(2)=f1(y(3),y(2))
dy(3)=f2(y(1),y(2))
end
I want to save the variable, constant1, each time the ode solver calls the function fun.
Is there a way to get the iteration count?
If yes, I can use
save_constant1(iter) = constant1.
Any suggestions?
  2 Comments
Steven Lord
Steven Lord on 24 Oct 2019
What do you want to happen if the ODE solver evaluates your function for a particular time step but then rejects that step and evaluates your function at an earlier time (a smaller time step from the previous time?)
Deepa Maheshvare
Deepa Maheshvare on 25 Oct 2019
Thanks a lot for the response. I suppose the step that is rejected will also be counted in the total number of iterations that the solver takes to solve the differential equations. So, I'd like to save the value of variable, constant1, for all iterations. Or, I could also create two variables, one that stores for all iterations and the other that skips the values computed at the steps that are rejected.

Sign in to comment.

Accepted Answer

Dinesh Yadav
Dinesh Yadav on 29 Oct 2019
What you can try is at every iteration save the variables into a .mat file
save('odef.mat', 'constant1', '-append')
This will append the new values and keep your old values also.
Hope it helps.
  2 Comments
alesmaz
alesmaz on 1 Nov 2019
Hi, I've tried your method in my code, where p is a vector of 4 parameters that I need to save at each ODE iteration in order to calculate their confidence intervals. I've put your string before the 'end' but my code overwrites always p without keeping the previous one.
Is there a way to save p at each ODE iteration without losing the one obtained in the previous iteration? Thank you.
function f=bernardode(p,t)
t=temposp;
options=odeset('AbsTol',1e-6,'RelTol',1e-6);
[T,Z]=ode45(@bernard2,t,z0,options);
function dz = bernard2(t,z)
dzdt=zeros(4,1);
dzdt(1)=-(p(3)*(1-qmin/z(1)))*(((Io/(sigma*L*z(2)))*(1-exp(-sigma*L*z(2))))/(p(4)+(Io/(sigma*L*z(2)))*(1-exp(-sigma*L*z(2)))))*z(1);
dzdt(2)=(p(3)*(1-qmin/z(1)))*(((Io/(sigma*L*z(2)))*(1-exp(-sigma*L*z(2))))/(p(4)+(Io/(sigma*L*z(2)))*(1-exp(-sigma*L*z(2)))))*z(2);
dzdt(3)=p(2)*z(4)+(p(3)*(1-qmin/z(1)))*(((Io/(sigma*L*z(2)))*(1-exp(-sigma*L*z(2))))/(p(4)+(Io/(sigma*L*z(2)))*(1-exp(-sigma*L*z(2)))))*(1-p(1)-z(3));
dzdt(4)=(p(1)-z(4))*(p(3)*(1-qmin/z(1)))*(((Io/(sigma*L*z(2)))*(1-exp(-sigma*L*z(2))))/(p(4)+(Io/(sigma*L*z(2)))*(1-exp(-sigma*L*z(2)))))-p(2)*z(4);
dz=dzdt;
end
f=Z;
end

Sign in to comment.

More Answers (0)

Products

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!