Thank you very much for your kind reply. Indeed I feel what I tried to
explain must have been a bit confusing!

A bit more information in detail:

0.2813+-0.0984 A  +  0.0192+-0.0184 B    ---->   0.1052+-0.0509 X

is a biochemical reaction I obtain from a parameter identification
(using several Matlab optimizers). I use the Fisher Information matrix
to "a posteriori" estimate the confidence limits for all parameters as
shown in the reaction.

As of now, I have never been interested in the confidence limits, so I
could easily normalize a reaction such as

0.2813 A  +  0.0192 B    ---->   0.1052 X

with respect to X to obtain

2.67 A   + 0.18 B --> X .

Now I am very much interested in the confidence limits (or
intervals) of
all parameters.
The identification procedure yields:

0.2813+-0.0984 A  +  0.0192+-0.0184 B    ---->   0.1052+-0.0509 X

Is there a possibility to calculate the coefficients of A and B such
that the biochemical reaction can be rewritten as?

xx A  +  yy B    ---->    X

without having to change the parameter estimation procedure just by
using the output as I obtain it so far. I realize that it is
problematic, that the coefficient of X is now deterministic equal to
one, where the uncertainties expressed by the confidence limits should
now be included in the coefficients of the other reaction components A
and B.

The normalization is necessary to compare the identification results
with a reference reaction scheme.

Can you think of a proper solution to this problem?

Any help on that greatly appreciated. I hope these explanations better
explain my problem.

Thanks in advance

Juergen U. Fritz ( [EMAIL PROTECTED] )
.
.
=================================================================
Instructions for joining and leaving this list, remarks about the
problem of INAPPROPRIATE MESSAGES, and archives are available at:
.                  http://jse.stat.ncsu.edu/                    .
=================================================================

Reply via email to