Monte Carlo Error Analysis with Motofit

Share icon

I have been trying to calculate the uncertainty in the parameters obtained from fitting a titration curve with a custom fit function. As suggested in previous posts, the motofit package can be used for this purpose.


http://www.igorexchange.com/node/2330

However, I have not been able to obtain meaningful results with the Moto_montecarlo function in the the latest release of the motofit v4 package (http://sourceforge.net/projects/motofit/files/motofit/ ). The matrix generated contains the same values for all the rows. These values are the same a the intial guess values. The command that I used is

Moto_Montecarlo ("myfitfunc", param_w, yy_w, xx_w, err_w, "holdstring", iter)

I also tried using a more recent version of the installer ( http://sourceforge.net/p/motofit/svn/775/tree/trunk/installers/), but Igor cannot find the definition for "sockitwavetostring."

I have two questions:



  • Has anyone tried using the Monte Carlo analysis in motofit with a user generated generated function?


  • Is there another version of motofit that I can use to run the Monte Carlo analysis? Ideally, I would like to use a version that runs multiple threads to speed up the process


    • Thanks
Isaac, I have tried doing this (I wrote the package).

Could you post a packed experiment file that demonstrates what you are trying to do?
Can you download and install the latest version and let me know if you still have problems?

https://sourceforge.net/projects/motofit/files/latest/download

I have just downloaded this and have checked that the code compiles. I also used this version to do a Monte Carlo analysis.

I remember fixing some issues along this line, when I was created code that could fit in parallel.

March 20, 2013 at 01:18 pm - Permalink

Andy,

I'll pack an experiment with a simple example of what I am trying to do. In the mean time, I tried installing the latest version of motofit and I got three errors after executing the motofitInstaller.exe and restarting Igor. It seems I am missing a few XOP libraries.

First, I get a popup window saying "XOP dynamic link failed." Then, I get a compilation error for  sockitwavetostring . I commented the lines with the previous command. Finally, I got a compilation error for "easyhttp" which I also commented.

The previous errors disappear after running platypusInstaller.exe but I get two additional compilation errors.

1) A couple of commas were missing in the following line of Pla.reductionPOD.ipf
 print "(Polarized Reduction) These are the datasets in angle "ii": Ffname = ", Ffname


2) a missing wave reference in Pla_Xrayreduction.ipf
wfprintf saveID, "%g \t %g \t %g \t %g \n", qq, rr, dr, dq


Thanks for the help

March 20, 2013 at 05:21 pm - Permalink

Shucks, looks like I'm going to be bug hunting.
When you get the message "xop dynamic link failed" does Igor give anymore information? I want to know which xop is failing to load.

It's probably better to email me directly.

March 20, 2013 at 06:56 pm - Permalink

I've just fixed the SOCKIT xop, but not easyHttp. You should be able to install Motofit from the release page, and it should work. Bloomin' Visual Studio. I never have these problems on OS X.

Just Motofit, you don't need to install platypusInstaller.exe

March 20, 2013 at 08:10 pm - Permalink