search

ocesou / hooke

Automatically exported from code.google.com/p/hooke
GNU Lesser General Public License v3.0
0 stars 0 forks source link

Peaks and convfilt doesn't agree on the number of peaks #3

Open GoogleCodeExporter opened 8 years ago

GoogleCodeExporter commented 8 years ago 
What steps will reproduce the problem?
1. Generate a playlist
2. Launch the convfilt
3. Try to compare the number of peaks recognize by convfilt for a curve
with the number of peaks recognize by peaks

What is the expected output? What do you see instead?
Convfilt and peaks should give us the same number of peaks.

What version of the product are you using? On what operating system?
Hooke 0.8.3 Gnu/linux kubuntu 8.04

Please provide any additional information below.

Original issue reported on code.google.com by fabrizio...@gmail.com on 22 Apr 2009 at 2:28

GoogleCodeExporter commented 8 years ago 
Ouch, this bug is still around!

Could you try to reproduce this on the 100-curve I27 dataset that you can 
download
from the home page, and report data? It would help to work both on the same set.

Original comment by devicera...@gmail.com on 22 Apr 2009 at 4:57

GoogleCodeExporter commented 8 years ago

Original comment by devicera...@gmail.com on 22 Apr 2009 at 4:58

GoogleCodeExporter commented 8 years ago 
No I can't. But I can send you my curve! We have a really useful utility here:
"attach a file". I will do that as soon as possible.

Original comment by fabrizio...@gmail.com on 23 Apr 2009 at 3:14

GoogleCodeExporter commented 8 years ago 
Here I have attached a curve.  This curve have passed a "convfilt 2 6", but with
"peaks 6" I obtain "0 peaks" (that is right because it have not peaks).

This is my settings in convfilt.conf:
<convfilt minpeaks="5" mindeviation="4" seedouble="10"
convolution="[6.0,-1.0,-1.0,-1.0,-1.0,-1.0,-1.0]" blindwindow="40"/>

Original comment by fabrizio...@gmail.com on 24 Apr 2009 at 2:32

Attachments:

GoogleCodeExporter commented 8 years ago 
I cannot reproduce the bug here. It rightly tells me that doesn't pass the 
convfilt.

I suspect it can be playlist-related -that is, it happens when you cycle along 
a lot
of files. We had this problem before (remember?), could be another instance of 
this
kind of bug.

Original comment by devicera...@gmail.com on 28 Apr 2009 at 11:33

GoogleCodeExporter commented 8 years ago 
Ok. Then what should I do?  Should I send you my complete playlist?
Tell me.

Original comment by fabrizio...@gmail.com on 28 Apr 2009 at 12:26

GoogleCodeExporter commented 8 years ago 
Ummm, most probably yes.

Original comment by devicera...@gmail.com on 28 Apr 2009 at 4:19

GoogleCodeExporter commented 8 years ago 
How large it is? If it's a lot of files, probably we should find a short enough
playlist that reproduces the problem. 

You could also try to debug this. Just add a call like 

do_peaks('6') 

in the code of convfilt, and check for the output of the convfilt, if do_peaks()
returns a value and convfilt returns another.

Original comment by devicera...@gmail.com on 28 Apr 2009 at 4:23

GoogleCodeExporter commented 8 years ago

Original comment by devicera...@gmail.com on 4 May 2009 at 2:45

GoogleCodeExporter commented 8 years ago 
Ping.

Original comment by fabrizio...@gmail.com on 19 Oct 2009 at 3:58