Improving the speed of code with passing to a function input a cell array or structure.

I'm trying to pass some amount of numerical data as a function input.

I had that row in my code:

[z1,y1]=ode45(@(z,y) famplifire(sigma_pa,N0,sigma_pe,sigma_se,k,y,z),zspan,startval);

I'm trying to replace four parameters: sigma_pa,N0,sigma_pe,sigma_se. I've tryed to replace them with cell array, but the time of execution of code nevertheless rises. So I had some questions:

1) Using of cell arrays insted of usual numerical data can improve quickness of my code?

2) I had another idea to use a struct, but I don't know how. If you can, give me some advices (or simple examples) please. Or I can make another question with my code.

Answer 1

[Matt J]

We can't see what famplifire [sic.] is doing with your variables so there isn't any specific comments that can be made. However, the rule of thumb is basically that if you must loop through a small number of large arrays/matrices, it can be advantageous to hold them in cells. Otherwise, it is probably better to concatenate them into a single array.

Structures are essentially the same as cell arrays. They are just indexed differently.

[George Bashkatov]

So, famplifire solve two ODE'S. Here it is:

function dydz = famplifire1(C,k,y,z)
k=C{2}*C{1}/(C{1}+C{3}); 
dy1 = y(1).*(-C{1}.*C{2}+(C{1}+C{3}).*(k.*y(1)./(y(1)+y(2)+1))); 
dy2 = y(2).*C{4}.*(k.*y(1)./(y(1)+y(2)+1)); 
 dydz = [dy1; dy2];

I've changed all constants to C{i}, i=1..4.

[Matt J]

If the C{i} are all scalars, it would be better if C were numeric rather than a cell array

[George Bashkatov]

Thank you a lot