ODE45 say ' must return a column vector' with experiment data.

When I use ODE 45 with velocity1 [9000x1] vector , it doesn't work well.

So I have a question.

I used 'interp1' for chagning [9000x1] to [1x9000] vector.

But when i run this, it says "@(TIME,X)G(TIME,X) must return a column vector".

So when I attach interp1 into ode45 like "ode45(@(time,x) interp1(t, g(time, x),time), time, 0.003);

xa2 has only "y=0.003".

Could you help me how to solve this problem? Thank you so much.

v = interp1(time, velocity1, t1);
g = @(time, x) -abs(v)*a(4)*x(1)*abs(x(1))^(n-1)-a(5)*v*abs(x(1))^n+a(6)*v;
[time, xa2] = ode45(@(time,x) g(time, x), time, 0.003);

[madhan ravi]

in the text file you attached there are 4 columns what do they represent? what are the values of time , velocity and t1 in interp1?

[CHOI HYUNDUK]

That represnts [Time, acceleration, displacement, force].

But i remade time '(0:0.0006:5.3994)' like this to start from 0.

and t1 is the time trasposed.

and velocity1 was got by kalman filter using acc and displacement data.

So if you check the workspace , everything is just 9000x1 or 1x9000vector.

Thank you so much!

[madhan ravi]

in the text file there are 4 columns please mention what datas they represent

[Torsten]

@CHOI HYUNDUK: I wonder why you can't just take the code I supplied.

The variable "time" in your call to interp1 is at the wrong position ; it has to be the last, not the first input parameter to "interp1".

[CHOI HYUNDUK]

@Torsten I am sorry, Torsten. I tried again seeing you answer!

Answer 1

[Jan]

"I used 'interp1' for chagning [9000x1] to [1x9000] vector." This is not the right job for an interpolation, but this is simply a transpose().

The posted code does not run due to the missing variables. Please post running code, such that we can check a suggested solution without the need to guess, what inputs you have.

ODE45 is not designed to process a non-differentiable function. The linearily interpolated data are not differentiable, so you run the integration outside the numerical specifications. This replies a final value, but you might get a huge accumulated rounding error.

Defining the function to be integrated as anonymous functions impedes the debugging. If you use a standard function instead and provide the parameters only by an anonymous function (see Answers: Anonymous for params), you can simply set a breakpoint in the function and check the cause of the error message. The error message claims "that the function does not reply a column vector" - but if your initial value is a scalar, the function to be integrated must reply a scalar also. Therefore it is not clear, why the output depends on "-abs(v)" at all.