I get a warning with ODE45 and the code runs but slowly and message constantly appears while running: arning: The value of local variables may have been changed to match the globals...

Please help! First time to post a question here.

Warning: The value of local variables may have been changed to match the globals.  Future versions of
MATLAB will require that you declare a variable to be global before you use that variable.

I have tried using all global and passing all through function. Some work both ways and some don't.

%%%SCRIPT
global  TD y0  xi T 

inputsize=1559;
m=1;
k=1;
h=0.02;
numberTDs=5;

%Import data and zero time and acceleration matrices
data=importdata('EarthquakeData.xlsx');
[ii,jj]=size(data);
acc=zeros(1,inputsize);
%T=zeros(1,inputsize);
T=0:h:(inputsize-1)*h;

n=1;
for i=1:ii
    for j=1:jj
            acc(1,n)=data(i,j);
            n=n+1;
            if n>inputsize
                break 
            end
    end
end
y0=[0,0];
[t,Y]=ode45('goodfunction1',T,y0,[],acc);

%%%FUNCTION
function ydot=goodfunction1(t,y,T,acc)
global   TD xi y0 T 
a=interp1(T,acc,t);
y0=[0,0]; 
ydot=[y(2);-(4*3.14159*xi/TD)*y(2)-(2*3.14159/TD)^2*y(1)-a];
end

Answer 1

[Walter Roberson]

inputsize=1559;
m=1;
k=1;
h=0.02;
numberTDs=5;

%Import data and zero time and acceleration matrices
data=importdata('EarthquakeData.xlsx');
[ii,jj]=size(data);
acc=zeros(1,inputsize);
%T=zeros(1,inputsize);
T=0:h:(inputsize-1)*h;

n=1;
for i=1:ii
    for j=1:jj
            acc(1,n)=data(i,j);
            n=n+1;
            if n>inputsize
                break 
            end
    end
end
y0 = [0,0];
C1 = (4*pi*xi/TD);
C2 = (2*pi/TD)^2;
[t, Y] = ode45( @(t,y) goodfunction1(t, y, acc, C1, C2), T, y0);

%%%FUNCTION
function ydot = goodfunction1(t, y, acc, C1, C2 )
  a = interp1(T, acc, t);
  ydot = [y(2); -C1*y(2) - C2*y(1) - a];
end

However... your use of interp1 to interpolate tells us that your equations are at best piecewise continuous. That is not good enough for any of the ode*() solvers. ode45() will probably detect the inconsistency in equations the first time a makes a non-linear change, and will spend a whole lot of time trying to narrow down the instability. This would be very slow if it works at all.

In any case in which your equation suddenly changes like this, you should be stopping the ode and starting again with another ode45 call using the boundary conditions returned by the previous one. ode45() cannot handle discontinuities in any derivative that it is working with (and probably not in two more beyond that.)

[Matthew Gill]

You are my hero! One thing though, I had to put the T as global. Why is that???

I see what you are saying about the nonlinearity of interpolation. However, that's the way the professors told us to do it and I wouldn't think there would be another way. It gives good results, but I have to do several iterations of this and it takes like 15 min to run. Not sure what else I can do.

[Walter Roberson]

[t, Y] = ode45( @(t,y) goodfunction1(t, y, T, acc, C1, C2), T, y0);

function ydot = goodfunction1(t, y, T, acc, C1, C2 )

[Matthew Gill]

Thank you

[Walter Roberson]

Oops, typo corrected.