I need to pursue the developement of a variable within a function, but I don't know how.

Yeah hi,

I am within debugging a Programm of mine and need to pursue a variable within a function. I am calling the function with an ode45-solver, while I am integrating x over t. So my problem is, that I'm using a variying stiffness, which I want to pursue over time, to identificate it's developement. My stiffness is a function of x(1) and I'm simply not sure anymore wheter it is following that, through me defined function, or not. Do you guys have any advice for me?

My Code:

[t_num,x_num]=ode15s(@DGL,tspan,q0);

function f=DGL(t,x)

global c_m c_a M om M_stat D

C=[c_m+c_a*sin(om*x(1)) -(c_m+c_a*sin(om*x(1)))

-(c_m+c_a*sin(om*x(1))) 2*c_m+c_a*sin(om*x(1))];

S=[zeros(2,2) eye(2,2)

-M\C -M\D];

a=M\[M_stat;-M_stat];

h=[zeros(2,1);a];

f=S*x+h;

end

I want to observe the value of C(1,1) over time t. Simply putting in a plot-command gives me no reliable result.

Answer 1

[Torsten]

[t_num,x_num]=ode15s(@DGL,tspan,q0);
C11_num = c_m+c_a*sin(om*x_num(:,1));

[Walter Roberson]

This is the better approach. If you try to monitor each call to your function then you run into the problem that the function is deliberately called 6 times with slightly different parameters in order to be able to estimate the local slope.

[Johannes Graf]

Yeah, I tried this. But could you answer me a question to the ode15s solver or better to the ode-integrators in general? I defined a function which should provide the necessary slopes of my searched variables. But it's only a feeling that the function is partially ignored and the stiffness developes along a certain condition that's hidden to me. Is that even possible?

[Torsten]

I defined a function which should provide the necessary slopes of my searched variables. But it's only a feeling that the function is partially ignored and the stiffness developes along a certain condition that's hidden to me.

I can't understand what you mean here. Could you explain in more detail ?

[Johannes Graf]

The solver is starting at t=x=0 therfore C has to be C=[1,-1;-1,1]*c_m. From here the function calculates S and with S it calculates the slope. My question is, wether it is possible that the ode-Solvers are holding the stiffness constant or varying it after there own rules over the span of various of those steps?

[Torsten]

The solver is starting at t=x=0 therfore C has to be C=[1,-1;-1,1]*c_m.

C at t=0 is [1 -1;-1 2]*c_m.

My question is, wether it is possible that the ode-Solvers are holding the stiffness constant or varying it after there own rules over the span of various of those steps?

Is S the stiffness constant ? If yes: The function DGL is called irregularily by ode15s in order to compute the time derivatives of the functions you solve for. DGL will be called more often when the gradients are steep and less often when the gradients are flat. The aim is to be efficient and precise at the same time. So the computation of S is done as often as necessary in order to keep the error tolerances for the solution (you can change these tolerances RelTol and AbsTol in the options setting for the integrator).

[Johannes Graf]

THANK YOU! That mere text saved my day!!