Prev: matrix import
Next: Distributed detection
From: Amanda on 25 Apr 2010 17:03
Dear All,

I am using the lsqnonlin algorithm to fit ODE parameters (4 parameters & 2 ODEs) to my biochemical network and get for some versions of the model the message:

"Warning: Matrix is close to singular or badly scaled.
Results may be inaccurate. RCOND = 1.828317e-16."
The objective function 'fun' calls the ode15s solver.

Which matrix is meant here?

Cheers, Amanda
From: Torsten Hennig on 25 Apr 2010 22:44
> Dear All,
>
> I am using the lsqnonlin algorithm to fit ODE
> parameters (4 parameters & 2 ODEs) to my biochemical
> network and get for some versions of the model the
> message:
>
> "Warning: Matrix is close to singular or badly
> scaled.
> Results may be inaccurate. RCOND =
> . RCOND = 1.828317e-16."
> The objective function 'fun' calls the ode15s solver.
>
> Which matrix is meant here?
>
> Cheers, Amanda

Either the Jacobian for ode15s (df_i/dy_j) or the
Jacobian used by lsqnonlin.
I'd check the values of the intermediate parameters
supplied by lsqnonlin to be used to integrate the
ode-system - maybe they attain unreasonable values.

Best wishes
Torsten.
 | 
Pages: 1
Prev: matrix import
Next: Distributed detection