how to using Dakota for comparison with experiment values?

[null8fuenf]

Hello!

We have some experiment data and a software, which can calculate the "things", which happens
in the experiment.
The software has some constants, which are not to 100 % known.
These constants we want to optimize with Dakota.
So that the results from the software better fits the experiment-data.

We have up to 7 parameters, which we want to optimize with Dakota. (But we can reduce the 7
to less).

From the software results we are using 3 values.

In dakota we are using in the moment method nl2sol.

We give to dakota the 3 values from experiment and the 3 values which the software calculates
and asking Dakota to optimize the 7 parameters.

We are writing the 3 values from the software line by line into results.out and the 3 values from
experiment line by line into experiment.dat.

results_file = 'results.out'
calibration_data_file 'experiment.dat'

But the results from Dakota which it is giving for the 7 parameters, are very
disappointed, because it is very very far away from experimentdata.

Is there a better method from Dakota as nl2sol to do such optimating work
to fit experiment data as we want to do?

How can I get from Dakota a better optimimating work for the 7 (or less)
parameters as in the moment?

Thanx.

Regards, Astrid