How does MTEX handle big chinks of blank data?

[Emanuel Benatti]

Hi everybody

the question is what is in the title. Let's say I have a measured pole figures, all of which are missing all the data of the center (from the center to maybe 15 degrees of the polar angle. What does MTEX do with these big chunks of blank data when doing de PDF to ODF inversion? It treates them as if they were zero? If not, then what?

Best regards

Emanuel

Please be as detailed as possible, explain what you want to do, what kind of data you are using. Attaching simplified code and data is always useful. Click the the button { } to inline code and finally remove these lines before posting :)

[Ralf Hielscher]

Hi Emanuel,

missing data as treated as missing data. There is simply no information and thus there is no ODF fitted.

Ralf.

[Emanuel Benatti]

Hello Ralf, and thanks for the quick response.

Maybe I should be more specific about the problem I'm having. In the attachment (Mg_experimental_PF.pdf) you can see an extruded Mg powder. Because of the experimental setup I can't measure the central part of any pole figure. Now, despite that, in the 0002 PF two fibers can be seen clearly, and when I recalculate the pole figures from the ODF with (Mg_recalculated_zerorange_PF.pdf) and without the zero range method (Mg_recalculated_PF.pdf) the fiber is completed in a more or less correct way, but you can also see some smaller components up and down this fiber. I have tried the followong things to eliminate this component:

• Change the resolution in the calcODF routine
• Change the bandwidth in the calcODF routine
• Remove the ghost correction
• Use and not use the zero range method
• Remove different pole figures from the calculation

With no success on any case (I got the best results with the zero range method). Is there a way to work around this or to solve it?

Thanks again

Emanuel

PD: the material is magnesium. Its structure is HCP (6/mmm) with cell paramters a = b = 3.2 A and c = 5.2 A

[Ralf Hielscher]

Could you please send the raw data?

Thanks,

Ralf.

[Emanuel Benatti]

Sure Ralf, sorry I forgot. I've also attached the script file I've been using to process the data.

Thanks again!

El miércoles, 6 de julio de 2016, 16:13:35 (UTC+2), Emanuel Benatti escribió:

[Ralf Hielscher]

Hi Emanuel,

I had a look at you data and played around a bit with the reconstruction but I'm afraid there is no magic combination of parameters that would result in a nicer ODF. This is where the reconstruction algorithm needs to improved. However, this is quite complicated. It is on my list. But it won't happen in close future.

Sorry,

Ralf.

[Emanuel Benatti]

Hi, thanks for taking the time for checking my data. It's good to know that is not only me who have this problem. I also wanted to tell you that I found a workaround for this particular problem (at least works more or less well with fiber textures). What I basically did was to extrapolate the data using the griddatan function of MATLAB. It's not a perfect solution, but the output is improved somehow. Also, I think the routine can be useful in other situations so I attach the code in case someone finds it useful.

Regards

Emanuel

El miércoles, 6 de julio de 2016, 16:13:35 (UTC+2), Emanuel Benatti escribió: