Hi,

I am trying to implement some numerical algorithm for an optimal control problem (still).

It involves double cycle. In the inner one two DE systems are solved sequentially and the result is used to update values of some objective function.

This part uses remeber option thanks to pagan  and is ok.

However, in the outer cycle I use the values of this objective function to obtain solution for yet another DE and update ICs for the inner cycle.

Dear all,

I have seen in the forum some post on this recently but cannot find it and thus ask this question again.

I have an output from a dsolve/numeric as a set of indexed names, like KK[1], KK[2], TT[1], TT[2], etc,

where KK[1]=eval(k(t),dsn) with dsn - output of dsolve/numeric.

I want to save to the file the result of last iteration (say, 15th) which is KK[15] in order not to repat all the calculations every time I open the file. In the file,...

Hi,

I would like to ask, what is the way to use the Direct Search (or other similar package) to solve not a static optimization problem for, say f(x_1,x_2), but the dynamic one, say for f(x_1(t),x_2(t),t)?

I think there must be the way of adapting the algorithm for that.

Any ideas welcome.

Best

Dear all,

Some time ago I asked a question concerning the algorythm for numerical dsolve and related issues here.

I got a nice answer from pagan and implemented the algorythm.

However, since my system is big, the iterations take long time.

Here is the code itself:

sys1:=[TRIC[1],TRIC[2],TRIC[3],TRIC[7],TRIC[8],TRIC[9]]:
> sys2:=[TRIC[4],TRIC[5],TRIC[6],TRIC[10],TRIC[11],TRIC[12]]:
> GRAD:=subs(u=FF[1](t),a=FF[2](t),k=k(t),tau=tau(t),psi[1...

Dear all,

I am stuck with my gradient projection algorythm. I asked several questions on this but nothing helped.

So, the basic problem I have is the following:

I need to make a procedure, which would take as an input some function of time, x(t),

calculate numerical solution for the ODE system

diff(y(t),t)=f(t,x(t),y(t))->y_1(t)

via dsolve/numeric

then update the value of the function x(t)=f(y_1(t)),

giving as a result x(t...