Miroslav Balda

@ David:
I thank you for thorough testing of my function LMFnlsq. You have found that 5 functions of the highest difficulty out of total 27 failed. It is true only partially. They failed only for the most bad initial estimates of the solution. The good solution has been obtained with the more close estimate for all above examples but Bennett5. In this case only one digit of sum of squares was valid, however, when sum of squares dropped by 8 orders! All those results have been obtained with embedded function for finite difference Jacobian matrix estimates, what strongly diminishes precission of results. Only analytical Jacobian matrix ensures top result. I am affraid that you also used finite difference approximates.

The NIST collection of testing problems is only a recommendation for software authors. In contrary to your opinion, I am satisfied by the results I have obtained. Especially, after testing a less difficult problem Hahn1, which converged for both estimates of the solution, while the function lsqcurvefit from the Optimization Toolbox (OT) did not find any solution. Would you evaluate it also by one star? Not me, because OT is much better. However, my function is smaler and takes fewer iterations to solve the problem. This is the reason, why I regard your evaluation as a bad joke.
One star means: "withdraw your function from FEX!". I will not obey your evaluation, because there are hundreds and may be thousands of people who downloaded and successfully applied my function for solving their problems. Some of them gave the function the top evaluation, even that I never did not declare it as an "industry standard".

This part of the submission is intended for Comments and Ratings. More over, I pleased anybody to write me by e-mail on errors and improvements, see the section Other requirements. This is the reason why I'll not communicate here anymore.

To Judith. What is wrong with your computer is difficult to find from your message. If you used the script LMFsolvetest, which is a part of the zipped file, you would get the solution. The stability has been reached in this case by a careful selection of optional parameters. It is written in the description, that the code is not perfectly stable and that it is better to use the function LMFnlsq (See the comment dated 06 Dec 2007).
If you try the old version (LMFsolveOLD), you would get also the answer. The solution of the problems could be solved much easier, if you read the last sentence in the section "Other requirements".

Man thank you so much, it was amazing!

@ David:
I thank you for thorough testing of my function LMFnlsq. You have found that 5 functions of the highest difficulty out of total 27 failed. It is true only partially. They failed only for the most bad initial estimates of the solution. The good solution has been obtained with the more close estimate for all above examples but Bennett5. In this case only one digit of sum of squares was valid, however, when sum of squares dropped by 8 orders! All those results have been obtained with embedded function for finite difference Jacobian matrix estimates, what strongly diminishes precission of results. Only analytical Jacobian matrix ensures top result. I am affraid that you also used finite difference approximates.

The NIST collection of testing problems is only a recommendation for software authors. In contrary to your opinion, I am satisfied by the results I have obtained. Especially, after testing a less difficult problem Hahn1, which converged for both estimates of the solution, while the function lsqcurvefit from the Optimization Toolbox (OT) did not find any solution. Would you evaluate it also by one star? Not me, because OT is much better. However, my function is smaler and takes fewer iterations to solve the problem. This is the reason, why I regard your evaluation as a bad joke.
One star means: "withdraw your function from FEX!". I will not obey your evaluation, because there are hundreds and may be thousands of people who downloaded and successfully applied my function for solving their problems. Some of them gave the function the top evaluation, even that I never did not declare it as an "industry standard".