How can I reduce the fsolve error?

Hi All

I'm solving a set of 9 second order equations with fsolve. It seems that it has a constant error on one of the variables even from the initial condition of the equations but then solving by variating the constants of the equations, the solution summary indicates a perfect solution with 1e-13 tolerance.

What can I do to resolve the initial error?

[Matt J]

Best thing is if you post your code (so that we can run it too) and show us what you mean.

[farzad]

Sorry I can't share the code

[Matt J]

Well, your post is really unclear. We can't tell what you mean without some sort of illustration. Maybe provide an alternative to your problem that produces the same issue?

[dpb]

Well, it's futile to expect anybody to answer such a detailed question that is entirely dependent upon the specifics of the function, its inputs and, quite possibly, the details of how the function is implemented. The crystal ball is often able to help, but this is far beyond even its power.

Only two questions

1. If you substitute the final solution into the function as initial condition does the symptom reappear, and
2. If you substitute the final solution into the function standalone, do you get zero (to near machine precision) as the result?

Altho from a practical standpoint, if it finds the solution, what difference does it make?