k-Wave

1. afana1972
   

   Member
   Joined: Mar '13
   Posts: 4

   Hi,
   I'm using the fwhm function to calculate the full width at half maximum of some records. the problem is the function works till it reach some odd values where the wave has a strange distribution (e.g. 1778,1398,1047,1165,1190,805,423,255,212,209,200,190, 193,194,192,192). so the function is crashed with the following massage
   ((("Subscript indices must either be real positive integers or logicals.
   Error in fwhm (line 87)
   m = (f(index-1) - f(index))/(x(index-1) - x(index));)))
   What type of modifications should I carry out to avoid such problem?
   regards
   Ashraf

   Posted 11 years ago #

2. Bradley Treeby
   

   Developer
   Joined: Mar '10
   Posts: 1,643

   Hi Ashraf,

   The function cannot find the full-width at half maximum if the input distribution doesn't decrease to half the maximum on both sides of the peak (for example, if it is monotonically decreasing). For the example you give, I get the error message left half maximum not found. Do you get the same message? One possibility to stop the function throwing an error when the input distribution is not compatible would be to set fwhm_val to zero and then exit the function using return instead of calling error.

   Brad

   Posted 11 years ago #

3. afana1972
   

   Member
   Joined: Mar '13
   Posts: 4

   Thanks Brad for the help.
   I resolved the problem by reorganizing the initial data.
   However, when I calculated the fwhm for a matrix this value is changed according to the size of the matrix, or let's say with the number of variables in the matrix. For example, if I am calcultaing the fwhm for m1 and m2 (m1 represent the data of wave 1 and m2 respresnt the data of wave 2) and t1 and t2 represents the time to m1 and m2 respectivel, the values of fwhm was changed completely if i applied the fwhm model m1 alone or if i applied it to both (m1 and m2, as a matrix and not a vector), which is not logical if i am using the same values all the time.
   m1 = (145,141,167,1012,2044,1874,1029,463,277,239,229,224,222,229,222,212)
   m2 = (141,149,466,1762,2186,1521,733,376,274,252,244,234,237,241,230,227)
   t1 and t2 is time intervals in 2010 ps.
   thanks again

   Posted 11 years ago #

4. Bradley Treeby
   

   Developer
   Joined: Mar '10
   Posts: 1,643

   The current version of fwhm is not vectorised, so won't give the correct output for a matrix input for f. You could look into the source code and modify it for this case, or call the function in a for loop.

   Hope that helps,

   Brad.

   Posted 11 years ago #

5. afana1972
   

   Member
   Joined: Mar '13
   Posts: 4

   Thanks Brady,
   I'll try to call the function in a for loop to solve this issue. I saw also that K-wave allows for depicting voxelPlot, which is very interesting for me. In my work, my idea is to perform a voxel 3D space with K-means clustering to look for difference rock structure. so, what advices you could give me, and what programs i may use for this mission.
   regards,
   Ashraf

   Posted 11 years ago #

6. afana1972
   

   Member
   Joined: Mar '13
   Posts: 4

   Dear Brady,
   I have some problems with fwhm, mainly in the following two cases:
   First, in some cases my data shows a maximum compound of two neighbour points (e.g. 142 149 153 176 252 252 200 160 147 142 140 149 151 149 151 145), and as you can see the maximum is two values of 252. In this case, the function us failed and the fowllowing massage is received:
   "Error using <
   Matrix dimensions must agree."

   Second, if the maximum value in my variable is less than 300 (e.g. 144 147 161 192 195 177 153 148 147 143 144 141 146 152 152 148), again the fwhm fails and the following massage is received:
   "Error using fwhm (line 64)
   left half maximum not found"

   So, what advice you may give me and how can I resolve this problem?
   Thanks again and waiting to hear from you
   Regards,

   Ashraf

   Posted 11 years ago #

7. Bradley Treeby
   

   Developer
   Joined: Mar '10
   Posts: 1,643

   The function fwhm will not work for both the discrete data sets you mention as they do not drop to half the maximum value. You should see the message 'left half maximum not found' which is correct.

   Brad.

   Posted 11 years ago #

RSS feed for this topic

Reply

You must log in to post.