Some questions about anonymous functions

Hello everyone!

I have some troubles with understanding of field of using anonymous functions.

The main idea contains something like this:

I have a system of differential equations, so every ode solver gets this system from the local function @odeFunc(t,x), but I need to solve this system...something like stepwise, i.e. at first I solve the first equation of the system, do something with results. Then I add second equation and solve the resulting system. And so on.

So the first question: can I somehow choose needed equations from beforehand given system?

My assumption that I can create an function handle array

f={@(x) f1;
   @(x) f2;
   ...}

and then somehow extract needed functions from this array and form a matrix from them needed for the solver on each step.

The second question is about how to transer this array or it's parts from the main function to another? The simplest case is how to transfer different right parts of the ode system to function @odeFunc(t,x)?

   f={@(x) f1;
   @(x) f2;
   ...}
   [T,X]=ode45(@odeFunc, tspan, x0);
   
   function RPF=odeFunc(t,x)
   %variable rigth part of ode system

[Steven Lord]

I have a system of differential equations, so every ode solver gets this system from the local function @odeFunc(t,x), but I need to solve this system...something like stepwise, i.e. at first I solve the first equation of the system, do something with results. Then I add second equation and solve the resulting system. And so on.

Can you show us the actual differential equations you're trying to solve, and indicate which of those equations you're trying to solve at each step?

[Ivan Khomich]

Ok, I have an example of system:

F1=[u];%First step(here u - is the amplitude of control)

F2=[x(2);...
    -x(2)+u];%Second step

F=[x(2);...
   -x(2)+x(3);...
   -x(3)+u];%Third step

The main point of why I want to do this because of for each step I have a different package of .m files. But they are pretty similar one to another and only difference between their work consists in different equations.

So if I could compose this three systems in something like array and transfer to another .m-files I would decrease number of similar .m-files in three times and so on decrease number of operations I need to change the system.

[darova]

Can you write a custom function (not anonymous)?

function res = F(t,x,case1)
switch
    case 1
        res=[u];%First step(here u - is the amplitude of control)
    case 2
        res=[x(2);...
            -x(2)+u];%Second step
    case 3
        res=[x(2);...
            -x(2)+x(3);...
            -x(3)+u];%Third step
end
end

[Ivan Khomich]

Thank you for your answer. It is an ok variant to solve the problem of switching odes, but the question was: is there a possibility to transfer function from one .m-file to another. Or maybe some ground why it is pointless and only worsen the code.

[J. Alex Lee]

Do you think you need duplication of source code across m-files? I don't think this is productive...

But you can always refer to other functions from functions...if you have the core functions somewhere "above" in your hierarchy of m-file groups, something like

function res = localFn(t,x,step)
res = MasterResFn(t,x,step)
end

and have the MasterResFn be something similar to what darova suggests. Would this accomplish what you want?