error solving differential equation

Hello everyone !!

I am trying to solve with simulink a simple differential equation such as : dT/dt = A T + C, where A is a squared matrix with zero elements except for A(i,i), A(i,i+1) and A(i-1,i).

Whatever I put as initial conditions or vector for C, the result is a vector with values of order 10^^300!!!

I am using an integrator, isn'it compatible with matrix?

I am quite new to Simulink and any help from you would be really nice, thank you.

yorelchr

[Shubham Gupta]

I can't really say why your output becoming too large since I don't know what your model looks like and I have very little information of the differential equation you have shared, but I can suggest what are te possible mistakes that you are making.

Let's start with the basic, say you have [A] matrix of dimension, [m x m], then your (T) and (C) vectors should be [m x 1]. Now, when you will make simulink model with your constants [A] and (C) you should get [m x 1] as output so, you will have to use demux to separate all the output.

Even after doing this, double check you differential equation. I hope this helps.

Note : Input initial condition for the differential equation by double clicking integrator block.

[Christine]

Thank you for your reply Shubham Gupta.

I think that my problem is a problem of stiff matrix. I have tried with ode15s solver but results are not what they should be.

[Shubham Gupta]

If you can share the model and/or differential equation it would be helpful to understand the actual problem. One possible way for integrator to give large values is if the differential equation is always increasing or worse always increasing exponentially.

For example if you integrate exp(x) from t=0 to t=100. Then, you know that you will get the values in higher order. So, I am reluctant to say this but it might be possible that you made some kind of mistake in the equation or model and the differential equation became always increasing.

Also, Changing the solver shouldn't change your basic trend for the result, imo. I can't comment any further unless I see the model and/or equation.