SIGFPE exception during M_SIMAMP process
305 views
Skip to first unread message
Georg Fischer
Oct 27, 2011, 12:27:45 PM10/27/11
to MAINSAR
Dear all,
during step_master_timing i get the following error:
(this is the same as Huaining Yang reported on 11th April in his post
"problem while running step_master_timing")
DEBUG : tmp result phi,lambda,h --> x,y,z: 0.563596, 1.71628, 3994 --
> -782898 5.34317e+06 3.38995e+06
DEBUG : tmp result phi,lambda,h --> x,y,z: 0.563596, 1.7163, 4007 --
> -782977 5.34317e+06 3.38996e+06
PROGRESS: SAM: Writing radar coded DEM and the incidence angle (theta)
to files, buffer#: 1
PROGRESS: SAM: Start interpolation....
Caught SIGFPE: floating point exception, zero division, etc.
Caught SIGFPE: floating point exception, zero division, etc.
Caught SIGFPE: floating point exception, zero division, etc.
The error occurs during the M_SIMAMP process of the Doris software (I
didn't find any hints in the Doris manual or the corresponding yahoo
group).
The radarcoding of the DEM seems to finish correctly, but when the
interpolation of the radarcoded DEM to the grid of the master SLC
starts an neverending SIGFPE exception is thrown.
An important note is, that this error only occurs if oversampling is
applied. If oversampling is skipped, the whole step_master_timing
process finishes correctly.
Since i want to use oversampling, i appreciate any advice how to solve
this!
regards,
Georg Fischer
M.Sc Student
Dept. of Geography
Ludwig-Maximilians-University Munich
during step_master_timing i get the following error:
(this is the same as Huaining Yang reported on 11th April in his post
"problem while running step_master_timing")
DEBUG : tmp result phi,lambda,h --> x,y,z: 0.563596, 1.71628, 3994 --
> -782898 5.34317e+06 3.38995e+06
DEBUG : tmp result phi,lambda,h --> x,y,z: 0.563596, 1.7163, 4007 --
> -782977 5.34317e+06 3.38996e+06
PROGRESS: SAM: Writing radar coded DEM and the incidence angle (theta)
to files, buffer#: 1
PROGRESS: SAM: Start interpolation....
Caught SIGFPE: floating point exception, zero division, etc.
Caught SIGFPE: floating point exception, zero division, etc.
Caught SIGFPE: floating point exception, zero division, etc.
The error occurs during the M_SIMAMP process of the Doris software (I
didn't find any hints in the Doris manual or the corresponding yahoo
group).
The radarcoding of the DEM seems to finish correctly, but when the
interpolation of the radarcoded DEM to the grid of the master SLC
starts an neverending SIGFPE exception is thrown.
An important note is, that this error only occurs if oversampling is
applied. If oversampling is skipped, the whole step_master_timing
process finishes correctly.
Since i want to use oversampling, i appreciate any advice how to solve
this!
regards,
Georg Fischer
M.Sc Student
Dept. of Geography
Ludwig-Maximilians-University Munich
Lu
Oct 27, 2011, 10:13:33 PM10/27/11
to MAINSAR
It seems that this error was caused by the triangulation-based
interpolation which is implemented using the open-source Triangle
library in Doris. Maybe you can debug into the library to find out the
reason. But I am not sure why this error only occurs when the DEM is
oversampled.
Cheer,
Lu
On 10月28日, 上午12时27分, Georg Fischer <georg.fischer...@googlemail.com>
wrote:
interpolation which is implemented using the open-source Triangle
library in Doris. Maybe you can debug into the library to find out the
reason. But I am not sure why this error only occurs when the DEM is
oversampled.
Cheer,
Lu
On 10月28日, 上午12时27分, Georg Fischer <georg.fischer...@googlemail.com>
wrote:
Georg Fischer
Oct 28, 2011, 9:50:06 AM10/28/11
to MAINSAR
I compared the debug logfiles step_master_timing.log for the procedure
with oversampled SLCs and without oversampling. Everything looks fine
until the error message, as mentioned before.
Well i postet already the log when oversampling was applied to the
master SLC:
DEBUG : tmp result phi,lambda,h --> x,y,z: 0.541183, 0.526289, -4 --
> 4.73113e+06 2.74852e+06 3.2666e+06
DEBUG : tmp result phi,lambda,h --> x,y,z: 0.541183, 0.526304, -4 --
> 4.73109e+06 2.74859e+06 3.2666e+06
SLC:
DEBUG : tmp result phi,lambda,h --> x,y,z: 0.541183, 0.526289, -4 --
> 4.73113e+06 2.74852e+06 3.2666e+06
DEBUG : tmp result phi,lambda,h --> x,y,z: 0.541183, 0.526304, -4 --
> 4.73109e+06 2.74859e+06 3.2666e+06
INFO : Master range spacing: 18.6255
INFO : Range-azimuth spacing ratio: 4.59998
INFO : Interpolation in: 1 buffers of 1000 lines and 0 extra buffer
of 0 lines.
DEBUG : Opening output file: master_sam.raw
DEBUG : Opened output file: master_sam.raw
PROGRESS: SAM: Interpolation buffer# [l0:lN, p0:pN]: 1 [7725: 8724,
3939: 4138]
..and then it finishes succesful.
I tried to use short int instead of real as format for the oversample
output, but still the same error.
Another aspect that is confusing me alot is, that i understand that
triangle is used during this interpolation step. But i am able to
succesfully run step_master_timing even if i've removed triangle?!?!
regards,
Georg
with oversampled SLCs and without oversampling. Everything looks fine
until the error message, as mentioned before.
Well i postet already the log when oversampling was applied to the
master SLC:
DEBUG : tmp result phi,lambda,h --> x,y,z: 0.541183, 0.526289, -4 --
> 4.73113e+06 2.74852e+06 3.2666e+06
DEBUG : tmp result phi,lambda,h --> x,y,z: 0.541183, 0.526304, -4 --
> 4.73109e+06 2.74859e+06 3.2666e+06
PROGRESS: SAM: Writing radar coded DEM and the incidence angle (theta)
to files, buffer#: 1
PROGRESS: SAM: Start interpolation....
Caught SIGFPE: floating point exception, zero division, etc.
Caught SIGFPE: floating point exception, zero division, etc.
Caught SIGFPE: floating point exception, zero division, etc.
And the same section when i did not apply oversampling to the master
to files, buffer#: 1
PROGRESS: SAM: Start interpolation....
Caught SIGFPE: floating point exception, zero division, etc.
Caught SIGFPE: floating point exception, zero division, etc.
Caught SIGFPE: floating point exception, zero division, etc.
SLC:
DEBUG : tmp result phi,lambda,h --> x,y,z: 0.541183, 0.526289, -4 --
> 4.73113e+06 2.74852e+06 3.2666e+06
DEBUG : tmp result phi,lambda,h --> x,y,z: 0.541183, 0.526304, -4 --
> 4.73109e+06 2.74859e+06 3.2666e+06
PROGRESS: SAM: Writing radar coded DEM and the incidence angle (theta)
to files, buffer#: 1
PROGRESS: SAM: Start interpolation....
INFO : Master azimuth spacing: 4.04903
to files, buffer#: 1
PROGRESS: SAM: Start interpolation....
INFO : Master range spacing: 18.6255
INFO : Range-azimuth spacing ratio: 4.59998
INFO : Interpolation in: 1 buffers of 1000 lines and 0 extra buffer
of 0 lines.
DEBUG : Opening output file: master_sam.raw
DEBUG : Opened output file: master_sam.raw
PROGRESS: SAM: Interpolation buffer# [l0:lN, p0:pN]: 1 [7725: 8724,
3939: 4138]
..and then it finishes succesful.
I tried to use short int instead of real as format for the oversample
output, but still the same error.
Another aspect that is confusing me alot is, that i understand that
triangle is used during this interpolation step. But i am able to
succesfully run step_master_timing even if i've removed triangle?!?!
regards,
Georg
Lu
Oct 28, 2011, 10:16:43 AM10/28/11
to MAINSAR
Do you have carefully checked the oversampled DEM together with your
input card settings to see if there is any problem? I guess so because
I also tried using both original and oversampled DEM for
step_master_timing, and no error occurred.
On 10月28日, 下午9时50分, Georg Fischer <georg.fischer...@googlemail.com>
wrote:
> > > Ludwig-Maximilians-University Munich- 隐藏被引用文字 -
>
> - 显示引用的文字 -
input card settings to see if there is any problem? I guess so because
I also tried using both original and oversampled DEM for
step_master_timing, and no error occurred.
On 10月28日, 下午9时50分, Georg Fischer <georg.fischer...@googlemail.com>
wrote:
>
> - 显示引用的文字 -
Georg Fischer
Oct 28, 2011, 10:47:11 AM10/28/11
to MAINSAR
Well the DEM is never oversampled before step_master_timing. The DEM
is kind of oversampling during the interpolation step, when the DEM
gets transformed into the radar geometry of the master SLC.
I always ment the oversampling of the master SLC during
step_master_ovs. This makes the differences if the error occurs or
not.
Or did i misunderstood you?
is kind of oversampling during the interpolation step, when the DEM
gets transformed into the radar geometry of the master SLC.
I always ment the oversampling of the master SLC during
step_master_ovs. This makes the differences if the error occurs or
not.
Or did i misunderstood you?
Georg Fischer
Oct 28, 2011, 11:28:12 AM10/28/11
to MAINSAR
Oh boy... Doris beta v4.04 solves this!
Anyway thanks for the help!
On Oct 28, 4:47 pm, Georg Fischer <georg.fischer...@googlemail.com>
wrote:
Anyway thanks for the help!
On Oct 28, 4:47 pm, Georg Fischer <georg.fischer...@googlemail.com>
wrote:
Mike Johnson
Feb 8, 2015, 8:13:13 PM2/8/15
to mai...@googlegroups.com, georg.fi...@googlemail.com
Hi Georg,
I get the same problem as you.Did you just fix it using Doris beta v4.04? I want to know if I can fix it in Doris v4.02.Can you help me? Thank you very much!
在 2011年10月28日星期五 UTC+8下午11:28:12，Georg Fischer写道：
I get the same problem as you.Did you just fix it using Doris beta v4.04? I want to know if I can fix it in Doris v4.02.Can you help me? Thank you very much!
在 2011年10月28日星期五 UTC+8下午11:28:12，Georg Fischer写道：
Reply all
Reply to author
Forward
0 new messages