floating point error ---ara-crop

[Tim Booth]

I have used ara-crop succesfully in the past on ccd micrographs.  Using the same procedure to process data from film I am receiving the following error....

 

cat .ara-crop.crash_report.0
2014-05-21 11:29:19,538 WARNING Failed to load EMAN2 module
2014-05-21 11:29:19,573 INFO Program: arachnid.util.crop
2014-05-21 11:29:19,573 INFO Version: 0.1.9
2014-05-21 11:29:19,573 INFO PID: 32449
2014-05-21 11:29:19,573 INFO Created: 1400686157
2014-05-21 11:29:19,573 INFO Multi-threading with OpenMP - enabled
2014-05-21 11:29:19,574 INFO Processing 651 movie micrographs
2014-05-21 11:29:19,575 INFO Skipping 0 files - restarting from the beginning - --force option specified
2014-05-21 11:29:19,577 INFO Processing 651 files
2014-05-21 11:29:19,577 INFO Particle diameter (in pixels): 290
2014-05-21 11:29:19,577 INFO Mask diameter (in pixels): 348
2014-05-21 11:29:19,577 INFO Window size: 406
2014-05-21 11:29:19,577 INFO High pass filter: 0.002463
2014-05-21 11:29:19,577 INFO Enhancement to windows:
2014-05-21 11:29:19,577 INFO - Ramp filter
2014-05-21 11:29:19,577 INFO - Dedust: 5.000000
2014-05-21 11:29:19,578 INFO - Histogram matching
2014-05-21 11:29:19,578 INFO - Normalize (xmipp style)
2014-05-21 11:29:19,578 INFO Extracting windows from micrograph stacks of movie frames
2014-05-21 11:29:19,599 INFO Using selection file local/screen/select.dat - found 448 micrographs of 651
2014-05-21 11:29:19,600 WARNING Found cached noise file: cluster/win/noise_win_000000.dat - delete if you want to regenerate
2014-05-21 11:29:25,924 ERROR client-processing - error
Traceback (most recent call last):
  File "/usr8/repository/miniconda/lib/python2.7/site-packages/arachnid/core/parallel/mpi_utility.py", line 385, in mpi_reduce
    for index, res in process_tasks.process_mp(process, vals, **extra):
  File "/usr8/repository/miniconda/lib/python2.7/site-packages/arachnid/core/parallel/process_tasks.py", line 62, in process_mp
    yield i, process(val, **extra)
  File "/usr8/repository/miniconda/lib/python2.7/site-packages/arachnid/util/crop.py", line 253, in process
    for j, mic in enumerate(iter_micrographs(filename, indexes, **extra)):
  File "/usr8/repository/miniconda/lib/python2.7/site-packages/arachnid/util/crop.py", line 346, in iter_micrographs
    mic = ndimage_filter.gaussian_highpass(mic, sigma/float(window), True)
  File "/usr8/repository/miniconda/lib/python2.7/site-packages/arachnid/core/image/ndimage_filter.py", line 466, in gaussian_highpass
    img = filter_image(img, gaussian_highpass_kernel(img.shape, high_cutoff, img.dtype), pad)
  File "/usr8/repository/miniconda/lib/python2.7/site-packages/arachnid/core/image/ndimage_filter.py", line 438, in gaussian_highpass_kernel
    kernel[:, :] = 1.0-numpy.exp(-irad*omega)
FloatingPointError: underflow encountered in exp
2014-05-21 11:29:26,001 ERROR ***Unexpected error occurred: FloatingPointError: underflow encountered in exp


See .ara-crop.crash_report for more details
2014-05-21 11:29:26,001 ERROR Unexpected error occurred
Traceback (most recent call last):
  File "/usr8/repository/miniconda/lib/python2.7/site-packages/arachnid/core/app/program.py", line 314, in launch_program
    main_template.main(args, main_module, **param)
  File "/usr8/repository/miniconda/lib/python2.7/site-packages/arachnid/core/app/file_processor.py", line 228, in main
    for index, filename in mpi_utility.mpi_reduce(process, files, init_process=init_process, **extra):
  File "/usr8/repository/miniconda/lib/python2.7/site-packages/arachnid/core/parallel/mpi_utility.py", line 385, in mpi_reduce
    for index, res in process_tasks.process_mp(process, vals, **extra):
  File "/usr8/repository/miniconda/lib/python2.7/site-packages/arachnid/core/parallel/process_tasks.py", line 62, in process_mp
    yield i, process(val, **extra)
  File "/usr8/repository/miniconda/lib/python2.7/site-packages/arachnid/util/crop.py", line 253, in process
    for j, mic in enumerate(iter_micrographs(filename, indexes, **extra)):
  File "/usr8/repository/miniconda/lib/python2.7/site-packages/arachnid/util/crop.py", line 346, in iter_micrographs
    mic = ndimage_filter.gaussian_highpass(mic, sigma/float(window), True)
  File "/usr8/repository/miniconda/lib/python2.7/site-packages/arachnid/core/image/ndimage_filter.py", line 466, in gaussian_highpass
    img = filter_image(img, gaussian_highpass_kernel(img.shape, high_cutoff, img.dtype), pad)
  File "/usr8/repository/miniconda/lib/python2.7/site-packages/arachnid/core/image/ndimage_filter.py", line 438, in gaussian_highpass_kernel
    kernel[:, :] = 1.0-numpy.exp(-irad*omega)
FloatingPointError: underflow encountered in exp

[Robert Langlois]

Hi Tim,

I see the problem, unfortunately I cannot fix it right away (I don't have a computer right now). One workaround would be to set the --sigma or sigma: (not sure if this name is correct) parameter in the crop.cfg to 0. That should turn off the high pass filter (it should have no significant impact on the later processing) this until I can fix this bug.

Regards,

Rob

--
This is for help using the software, please do not submit bugs or feature requests (post them here: https://code.google.com/p/arachnid/issues/list) .
---
You received this message because you are subscribed to the Google Groups "arachnid-general" group.
To unsubscribe from this group and stop receiving emails from it, send an email to arachnid-gener...@googlegroups.com.
To post to this group, send email to arachnid...@googlegroups.com.
Visit this group at http://groups.google.com/group/arachnid-general.
For more options, visit https://groups.google.com/d/optout.

[Robert Langlois]

I just released a new version of the code that should fix this problem.

To unsubscribe from this group and stop receiving emails from it, send an email to arachnid-general+unsubscribe@googlegroups.com.
To post to this group, send email to arachnid-general@googlegroups.com.