Hello Everyone, 

Figured I would see if anyone is experienced with this stuff prior to spending two hours crafting a message with screenshots / diagrams / uploading code etc. I am using Matlab (R2018b) Update 2 and  Excel 2016 ( 64 bit ) doing some Wavelet coefficient thresholding analysis. I can provide specifics if anyone is experienced with the aforementioned software and processes.



[ - ]
Reply by DonaldJLucas23January 26, 2019

I am not familiar with the process that you are using so I can't help there, but I had a major red flag go up for using the 64-bit version of Excel.  According to Microsoft, the 64-bit version of Office/Excel is only for very limited special purpose applications and is incompatible with many add-ins.  Microsoft doesn't recommend it for general purpose applications due to compatibility issues.  I don't know what problem they have with the 64-bit version, but I suspect that you should be using the 32-bit version (even on a 64-bit processor) unless you are familiar with Microsoft's reasons for offering a 64-bit version of Excel that is incompatible with most other applications.  In other words, you should only be using it if you know exactly what you are doing and why you are using the 64-bit version of Excel.  As far as I know, this also holds true for other Microsoft Office apps.  I use Microsoft Office 365 that always keeps it up to date with the most current available version and a 64-bit version was not offered.

I hope that this is helpful and my apologies if this situation has changed in the last 12 months since I last checked and I am simply not aware of the change yet.

[ - ]
Reply by NegativeKurtosisJanuary 26, 2019


Thanks for the reply / info. I'm not currently experiencing issues with either Excel or Matlab. I've programmed them to communicate effectively for my needs.I switched to the 64 bit version of Excel for heavy computational demands. I also have a couple 64 bit version Excel add ins that work flawlessly. Thanks again for the info; very easy to not be up - to - date with these issues.


[ - ]
Reply by MichaelRWJanuary 26, 2019

Matlab and Excel are very common tools.  I would think a lot of people on this site have used them or their equivalents.

Post your questions...

[ - ]
Reply by DonaldJLucas23January 26, 2019
I use Excel daily - my concern was only with the 64-bit version.  I think that Richard has successfully worked around the issues with it by finding the appropriate 64-bit plug-ins.  Not all plugins have those available.  Glad to hear that at least some apps are working ok with the 64-bit version, though, as it is long overdue.  Here is the link to the limitations of the 64-bit version as published by Microsoft:  

Choose between the 64-bit or 32-bit version of Office

Again, though, if you have compatible 64-bit plug-ins, it will work ok.  They just aren't as available as you might assume.

[ - ]
Reply by NegativeKurtosisJanuary 26, 2019

Post deleted by author

[ - ]
Reply by NegativeKurtosisJanuary 26, 2019

Not sure what happened to my post. It appears I may have accidentally deleted it trying to add to it. Anyway, here is my question / problem :

Using the Matlab Wavelet Analyzer :

Wavelet Analyzer.JPG

I select the 1 - D Wavelet Coefficient Selection :

1 - D Wavelet Coefficient Selection.JPG

Then I run the Stepwise Coefficient Process:

Wavelet Coefficient Stepwise Process.JPG

I am wondering if the Stepwise Coefficient Selection Process is 'transferable' and can be saved in a .mat file and utilized on new DAQ input values ? The Matlab documentation is pretty clear on how to save a file but I am unclear on whether or not I can access the previously obtained Stepwise Coefficient Selection Values obtained from the prior Stepwise Movie step.

It appears that the function 'wthrmngr' provides arguments that MAY access those previously obtained Stepwise Coefficient Selection Values via the following 

code :

thr = wthrmngr(opt,method,C,L)

From Matlab -

1-D Discrete Wavelet Decomposition Options

In these options, X is the signal, the wavelet coefficients are in the vector C, and the lengths of the coefficient vectors are in L. The argument alpha is the sparsity parameter, and scale defines the multiplicative threshold rescaling.

For additional information regarding the wavelet decomposition, see wavedec. To learn more about alpha and scale, see wdcbm and wden respectively.


1-D denoising using level-dependent thresholds

  • thr = wthrmngr('dw1ddenoLVL','sqtwolog',C,L,scale)
  • thr = wthrmngr('dw1ddenoLVL','rigrsure',C,L,scale)
  • thr = wthrmngr('dw1ddenoLVL','heursure',C,L,scale)
  • thr = wthrmngr('dw1ddenoLVL','minimaxi',C,L,scale)
  • thr = wthrmngr('dw1ddenoLVL','penalhi',C,L,alpha), where 2.5 < alpha < 10
  • thr = wthrmngr('dw1ddenoLVL','penalme',C,L,alpha), where 1.5 < alpha < 2.5
  • thr = wthrmngr('dw1ddenoLVL','penallo',C,L,alpha), where 1 < alpha < 2

However, I am not sure which syntax's to use. Do I understand this correctly ? Can anyone provide clarification if I don't understand this properly ?