[Met_help] [rt.rap.ucar.edu #43337] History for grid_stat errors
RAL HelpDesk {for John Halley Gotway}
met_help at ucar.edu
Tue Feb 8 16:53:22 MST 2011
----------------------------------------------------------------
Initial Request
----------------------------------------------------------------
MET Help,
I'm running grid_stat v3.0 and I get the following errors:
grid_stat(18710): unaligned access to 0x60000fffffda6f5c, ip=0x400000000016fb61
grid_stat(18710): unaligned access to 0x60000fffffda6f5c, ip=0x400000000016fc10
grid_stat(18710): unaligned access to 0x60000fffffda6f5c, ip=0x400000000016fc31
grid_stat(18710): unaligned access to 0x60000fffffda6f5c, ip=0x400000000016fdb1
grid_stat(18710): unaligned access to 0x60000fffffda6f5c, ip=0x400000000016fdf1
Gridstat runs fine even with these errors, and as far as I can tell, the output values are
reasonable. Do you know what these errors are from? Should I be concerned about them? If
you need test files, let me know.
Thanks,
Derek Stratman
----------------------------------------------------------------
Complete Ticket History
----------------------------------------------------------------
Subject: Re: [rt.rap.ucar.edu #43337] grid_stat errors
From: John Halley Gotway
Time: Mon Jan 10 09:49:39 2011
Derek,
No, I've never seen these error messages before. I did a quick search
online for information about them, and it sounds like they may arise
when the program allocates memory in atypical ways or casts
variables from one type to another. Sounds like they aren't a big
deal and that the program often runs to completion without any
problems. So I do think it's reasonable to ignore them for now.
One thing you could try would be to recompile MET - from the top-level
MET directory, do a "make clean" followed by a "make". Perhaps that'd
help.
I would of course like to replicate this warning message and fix it,
but I'm not hopeful that I'll be able to. These types of system-
dependent things are often very difficult to replicate. But we
might as well give it a shot. Assuming recompiling doesn't help,
could you please send me the following information:
(1) The files you use for running Grid-Stat: forecast file,
observation file, and configuration file.
(2) The command line you use to run Grid-Stat.
(3) A description of your machine including operating system. Try
typing "uname -a".
You can post those data files to our anonymous ftp site by following
these instructions:
http://www.dtcenter.org/met/users/support/met_help.php#ftp
Thanks,
John
On 01/07/2011 10:24 AM, RAL HelpDesk {for Derek.Stratman} wrote:
>
> Fri Jan 07 10:24:11 2011: Request 43337 was acted upon.
> Transaction: Ticket created by Derek.Stratman at noaa.gov
> Queue: met_help
> Subject: grid_stat errors
> Owner: Nobody
> Requestors: Derek.Stratman at noaa.gov
> Status: new
> Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=43337 >
>
>
> MET Help,
>
> I'm running grid_stat v3.0 and I get the following errors:
>
> grid_stat(18710): unaligned access to 0x60000fffffda6f5c,
ip=0x400000000016fb61
> grid_stat(18710): unaligned access to 0x60000fffffda6f5c,
ip=0x400000000016fc10
> grid_stat(18710): unaligned access to 0x60000fffffda6f5c,
ip=0x400000000016fc31
> grid_stat(18710): unaligned access to 0x60000fffffda6f5c,
ip=0x400000000016fdb1
> grid_stat(18710): unaligned access to 0x60000fffffda6f5c,
ip=0x400000000016fdf1
>
> Gridstat runs fine even with these errors, and as far as I can tell,
the output values are
> reasonable. Do you know what these errors are from? Should I be
concerned about them? If
> you need test files, let me know.
>
> Thanks,
> Derek Stratman
>
>
------------------------------------------------
Subject: Re: [rt.rap.ucar.edu #43337] grid_stat errors
From: Derek.Stratman
Time: Mon Jan 10 10:13:42 2011
John,
The files should be in the folder now. I run GridStat with:
grid_stat forecast.grb observation.nc GridStatConfig_process -outdir
${outdirectory}
Here's my machine's description:
Linux gallifrey.protect.nssl 2.6.18-194.26.1.el5 #1 SMP Fri Oct 29
14:25:52 EDT 2010 ia64
ia64 ia64 GNU/Linux
Thanks,
Derek
----- Original Message -----
From: RAL HelpDesk {for John Halley Gotway} <met_help at ucar.edu>
Date: Monday, January 10, 2011 10:49 am
Subject: Re: [rt.rap.ucar.edu #43337] grid_stat errors
> Derek,
>
> No, I've never seen these error messages before. I did a quick
> search online for information about them, and it sounds like they
> may arise when the program allocates memory in atypical ways or
casts
> variables from one type to another. Sounds like they aren't a big
> deal and that the program often runs to completion without any
> problems. So I do think it's reasonable to ignore them for now.
>
> One thing you could try would be to recompile MET - from the top-
> level MET directory, do a "make clean" followed by a "make".
> Perhaps that'd help.
>
> I would of course like to replicate this warning message and fix
> it, but I'm not hopeful that I'll be able to. These types of
> system-dependent things are often very difficult to replicate. But
we
> might as well give it a shot. Assuming recompiling doesn't help,
> could you please send me the following information:
>
> (1) The files you use for running Grid-Stat: forecast file,
> observation file, and configuration file.
> (2) The command line you use to run Grid-Stat.
> (3) A description of your machine including operating system. Try
> typing "uname -a".
>
> You can post those data files to our anonymous ftp site by
> following these instructions:
> http://www.dtcenter.org/met/users/support/met_help.php#ftp
>
> Thanks,
> John
>
> On 01/07/2011 10:24 AM, RAL HelpDesk {for Derek.Stratman} wrote:
> >
> > Fri Jan 07 10:24:11 2011: Request 43337 was acted upon.
> > Transaction: Ticket created by Derek.Stratman at noaa.gov
> > Queue: met_help
> > Subject: grid_stat errors
> > Owner: Nobody
> > Requestors: Derek.Stratman at noaa.gov
> > Status: new
> > Ticket <URL:
> https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=43337 >
> >
> >
> > MET Help,
> >
> > I'm running grid_stat v3.0 and I get the following errors:
> >
> > grid_stat(18710): unaligned access to 0x60000fffffda6f5c,
> ip=0x400000000016fb61> grid_stat(18710): unaligned access to
> 0x60000fffffda6f5c, ip=0x400000000016fc10
> > grid_stat(18710): unaligned access to 0x60000fffffda6f5c,
> ip=0x400000000016fc31> grid_stat(18710): unaligned access to
> 0x60000fffffda6f5c, ip=0x400000000016fdb1
> > grid_stat(18710): unaligned access to 0x60000fffffda6f5c,
> ip=0x400000000016fdf1>
> > Gridstat runs fine even with these errors, and as far as I can
> tell, the output values are
> > reasonable. Do you know what these errors are from? Should I be
> concerned about them? If
> > you need test files, let me know.
> >
> > Thanks,
> > Derek Stratman
> >
> >
>
>
------------------------------------------------
Subject: grid_stat errors
From: John Halley Gotway
Time: Mon Jan 10 12:03:59 2011
Derek,
I ran it here, and as I suspected, I didn't see the messages you saw.
I've attached the output stat file Grid-Stat created. I do see a
problem in the Grid-Stat output however - I see "inf" and
"-inf" values in the normal confidence limits. I ran it through the
debugger and think that we've got a variable overflow problem causing
this. Are you seeing "inf" and "-inf" in your Grid-Stat
output as well?
Perhaps that's the source of the messages you're seeing... or perhaps
not. Regardless, it's a bug we should try to fix.
This is caused by computing statistics using 65536 matched pairs.
When I reduce the number of matched pairs used by performing data
masking on the observation (only looking at points where REFC > 0),
I do get reasonable results for the confidence limits. The number of
matched pairs is reduced down to 12,177.
I'll need to work on this and see if I can figure out a patch that'll
allow for large numbers of matched pairs (i.e. 65536).
Thanks,
John
On 01/10/2011 10:13 AM, RAL HelpDesk {for Derek.Stratman} wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=43337 >
>
> John,
>
> The files should be in the folder now. I run GridStat with:
> grid_stat forecast.grb observation.nc GridStatConfig_process -outdir
${outdirectory}
>
> Here's my machine's description:
> Linux gallifrey.protect.nssl 2.6.18-194.26.1.el5 #1 SMP Fri Oct 29
14:25:52 EDT 2010 ia64
> ia64 ia64 GNU/Linux
>
> Thanks,
> Derek
>
> ----- Original Message -----
> From: RAL HelpDesk {for John Halley Gotway} <met_help at ucar.edu>
> Date: Monday, January 10, 2011 10:49 am
> Subject: Re: [rt.rap.ucar.edu #43337] grid_stat errors
>> Derek,
>>
>> No, I've never seen these error messages before. I did a quick
>> search online for information about them, and it sounds like they
>> may arise when the program allocates memory in atypical ways or
casts
>> variables from one type to another. Sounds like they aren't a big
>> deal and that the program often runs to completion without any
>> problems. So I do think it's reasonable to ignore them for now.
>>
>> One thing you could try would be to recompile MET - from the top-
>> level MET directory, do a "make clean" followed by a "make".
>> Perhaps that'd help.
>>
>> I would of course like to replicate this warning message and fix
>> it, but I'm not hopeful that I'll be able to. These types of
>> system-dependent things are often very difficult to replicate. But
we
>> might as well give it a shot. Assuming recompiling doesn't help,
>> could you please send me the following information:
>>
>> (1) The files you use for running Grid-Stat: forecast file,
>> observation file, and configuration file.
>> (2) The command line you use to run Grid-Stat.
>> (3) A description of your machine including operating system. Try
>> typing "uname -a".
>>
>> You can post those data files to our anonymous ftp site by
>> following these instructions:
>> http://www.dtcenter.org/met/users/support/met_help.php#ftp
>>
>> Thanks,
>> John
>>
>> On 01/07/2011 10:24 AM, RAL HelpDesk {for Derek.Stratman} wrote:
>>>
>>> Fri Jan 07 10:24:11 2011: Request 43337 was acted upon.
>>> Transaction: Ticket created by Derek.Stratman at noaa.gov
>>> Queue: met_help
>>> Subject: grid_stat errors
>>> Owner: Nobody
>>> Requestors: Derek.Stratman at noaa.gov
>>> Status: new
>>> Ticket <URL:
>> https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=43337 >
>>>
>>>
>>> MET Help,
>>>
>>> I'm running grid_stat v3.0 and I get the following errors:
>>>
>>> grid_stat(18710): unaligned access to 0x60000fffffda6f5c,
>> ip=0x400000000016fb61> grid_stat(18710): unaligned access to
>> 0x60000fffffda6f5c, ip=0x400000000016fc10
>>> grid_stat(18710): unaligned access to 0x60000fffffda6f5c,
>> ip=0x400000000016fc31> grid_stat(18710): unaligned access to
>> 0x60000fffffda6f5c, ip=0x400000000016fdb1
>>> grid_stat(18710): unaligned access to 0x60000fffffda6f5c,
>> ip=0x400000000016fdf1>
>>> Gridstat runs fine even with these errors, and as far as I can
>> tell, the output values are
>>> reasonable. Do you know what these errors are from? Should I be
>> concerned about them? If
>>> you need test files, let me know.
>>>
>>> Thanks,
>>> Derek Stratman
>>>
>>>
>>
>>
------------------------------------------------
Subject: grid_stat errors
From: John Halley Gotway
Time: Mon Jan 10 12:03:59 2011
VERSION MODEL FCST_LEAD FCST_VALID_BEG FCST_VALID_END OBS_LEAD
OBS_VALID_BEG OBS_VALID_END FCST_VAR FCST_LEV OBS_VAR OBS_LEV
OBTYPE VX_MASK INTERP_MTHD INTERP_PNTS FCST_THRESH OBS_THRESH
COV_THRESH ALPHA LINE_TYPE
V3.0 caps_cn 120000 20100511_030000 20100511_030000 000000
20100511_030000 20100511_030000 REFC L0 REFC SFC
ANALYS FULL UW_MEAN 1 >=10.000 >=10.000 NA
NA FHO 65536 0.14536 0.06729 0.17311
V3.0 caps_cn 120000 20100511_030000 20100511_030000 000000
20100511_030000 20100511_030000 REFC L0 REFC SFC
ANALYS FULL UW_MEAN 1 >=10.000 >=10.000 NA
NA CTC 65536 4410 5116 6935 49075
V3.0 caps_cn 120000 20100511_030000 20100511_030000 000000
20100511_030000 20100511_030000 REFC L0 REFC SFC
ANALYS FULL UW_MEAN 1 >=10.000 >=10.000 NA
0.01000 CTS 65536 0.17311 -inf inf 0.13434 0.20916
0.14536 -inf inf 0.11144 0.18321 0.81612 -inf inf
0.77709 0.85802 0.83967 0.64480 1.10102 0.38872 -inf inf
0.28281 0.50893 0.90559 -inf inf 0.87431 0.93716 0.09441 -inf
inf 0.06284 0.12569 0.53706 -inf inf 0.40775
0.67443 0.26791 -inf inf 0.18880 0.36365 0.18640 0.10873
0.28556 0.29431 0.28135 0.30727 0.18224 0.42147 0.31423
0.19613 0.44425 6.09989 5.73091 6.49263 3.44970 11.69860
V3.0 caps_cn 120000 20100511_030000 20100511_030000 000000
20100511_030000 20100511_030000 REFC L0 REFC SFC
ANALYS FULL UW_MEAN 1 >=20.000 >=20.000 NA
NA FHO 65536 0.12297 0.02374 0.05579
V3.0 caps_cn 120000 20100511_030000 20100511_030000 000000
20100511_030000 20100511_030000 REFC L0 REFC SFC
ANALYS FULL UW_MEAN 1 >=20.000 >=20.000 NA
NA CTC 65536 1556 6503 2100 55377
V3.0 caps_cn 120000 20100511_030000 20100511_030000 000000
20100511_030000 20100511_030000 REFC L0 REFC SFC
ANALYS FULL UW_MEAN 1 >=20.000 >=20.000 NA
0.01000 CTS 65536 0.05579 -inf inf 0.03359 0.08244
0.12297 -inf inf 0.09007 0.15726 0.86873 -inf inf
0.83358 0.90076 2.20432 1.43589 3.66667 0.42560 -inf inf
0.23073 0.65714 0.89491 -inf inf 0.86173 0.92730 0.10509 -inf
inf 0.07270 0.13827 0.80692 -inf inf 0.67603
0.90910 0.15316 -inf inf 0.06818 0.24476 0.11395 0.03610
0.20736 0.32051 0.29885 0.34217 0.12319 0.55095 0.20459
0.06969 0.34349 6.30966 5.75197 6.92143 2.36908 17.00421
V3.0 caps_cn 120000 20100511_030000 20100511_030000 000000
20100511_030000 20100511_030000 REFC L0 REFC SFC
ANALYS FULL UW_MEAN 1 >=30.000 >=30.000 NA
NA FHO 65536 0.08067 0.00880 0.01849
V3.0 caps_cn 120000 20100511_030000 20100511_030000 000000
20100511_030000 20100511_030000 REFC L0 REFC SFC
ANALYS FULL UW_MEAN 1 >=30.000 >=30.000 NA
NA CTC 65536 577 4710 635 59614
V3.0 caps_cn 120000 20100511_030000 20100511_030000 000000
20100511_030000 20100511_030000 REFC L0 REFC SFC
ANALYS FULL UW_MEAN 1 >=30.000 >=30.000 NA
0.01000 CTS 65536 0.01849 -inf inf 0.00611 0.03512
0.08067 -inf inf 0.05344 0.11146 0.91844 -inf inf
0.89006 0.94504 4.36221 2.36801 13.00125 0.47607 -inf inf
0.07692 0.83345 0.92678 -inf inf 0.89923 0.95179 0.07322 -inf
inf 0.04821 0.10078 0.89086 -inf inf 0.77082
0.98215 0.09743 -inf inf 0.01587 0.20692 0.08228 -0.00020
0.19019 0.40285 0.36565 0.44005 -0.00096 0.76872 0.15205
-0.00039 0.31960 11.50084 9.86732 13.40478 0.98667 59.92195
V3.0 caps_cn 120000 20100511_030000 20100511_030000 000000
20100511_030000 20100511_030000 REFC L0 REFC SFC
ANALYS FULL UW_MEAN 1 >=40.000 >=40.000 NA
NA FHO 65536 0.01845 0.00044 0.00824
V3.0 caps_cn 120000 20100511_030000 20100511_030000 000000
20100511_030000 20100511_030000 REFC L0 REFC SFC
ANALYS FULL UW_MEAN 1 >=40.000 >=40.000 NA
NA CTC 65536 29 1180 511 63816
V3.0 caps_cn 120000 20100511_030000 20100511_030000 000000
20100511_030000 20100511_030000 REFC L0 REFC SFC
ANALYS FULL UW_MEAN 1 >=40.000 >=40.000 NA
0.01000 CTS 65536 0.00824 -inf inf 0.00153 0.01832
0.01845 -inf inf 0.00458 0.03207 0.97420 -inf inf
0.95877 0.99084 2.23889 0.69917 15.05000 0.05370 -inf inf
0.00000 0.50000 0.98185 -inf inf 0.96759 0.99540 0.01815 -inf
inf 0.00460 0.03241 0.97601 -inf inf 0.81809
1.00000 0.01686 -inf inf 0.00000 0.13338 0.01113 -0.00924
0.12752 0.03555 0.01003 0.06107 -0.03231 0.48928 0.02202
-0.01866 0.22619 3.06920 1.86623 5.04758 0.00000 64.35378
V3.0 caps_cn 120000 20100511_030000 20100511_030000 000000
20100511_030000 20100511_030000 REFC L0 REFC SFC
ANALYS FULL UW_MEAN 1 NA NA NA
0.01000 CNT 65536 4.59305 4.47798 4.70812 3.42113 5.75440
11.43629 11.35546 11.51820 9.74058 12.80047 3.49120 3.40803 3.57436
2.72460 4.32237 8.26522 8.20681 8.32442 6.91938 9.51561 0.32839
0.31938 0.33734 0.20830 0.45355 NA NA 0 0 0
1.10185 0.98407 1.21963 -0.07622 2.30322 11.70562 11.62289
11.78946 9.92888 13.22815 1.31561 0.97882 1.74065 5.35653 4.20163
6.39328 138.23362 98.71536 177.26786 137.01955 98.43209 174.71676
11.75728 9.93556 13.31420 -11.80880 -15.08231 0.00000 0.00000
0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000
0.00000 13.99989 6.09045 20.04312
V3.0 caps_cn 120000 20100511_030000 20100511_030000 000000
20100511_030000 20100511_030000 REFC L0 REFC SFC
ANALYS FULL UW_MEAN 1 NA NA NA
NA SL1L2 65536 4.59305 3.49120 47.07521 151.88270 80.50135
V3.0 caps_cn 120000 20100511_030000 20100511_030000 000000
20100511_030000 20100511_030000 REFC L0 REFC SFC
ANALYS REFC_gt0.0 UW_MEAN 1 >=10.000 >=10.000 NA
NA FHO 12177 0.38671 0.36216 0.93167
V3.0 caps_cn 120000 20100511_030000 20100511_030000 000000
20100511_030000 20100511_030000 REFC L0 REFC SFC
ANALYS REFC_gt0.0 UW_MEAN 1 >=10.000 >=10.000 NA
NA CTC 12177 4410 299 6935 533
V3.0 caps_cn 120000 20100511_030000 20100511_030000 000000
20100511_030000 20100511_030000 REFC L0 REFC SFC
ANALYS REFC_gt0.0 UW_MEAN 1 >=10.000 >=10.000 NA
0.01000 CTS 12177 0.93167 0.92555 0.93733 0.86885 0.98361
0.38671 0.37541 0.39814 0.27049 0.50820 0.40593 0.39452 0.41744
0.29504 0.52459 0.41507 0.28694 0.53847 0.38872 0.37740 0.40015
0.27584 0.50928 0.64062 0.62935 0.65175 0.14286 1.00000 0.35938
0.34825 0.37065 0.00000 0.85714 0.06350 0.05804 0.06943 0.00000
0.17393 0.37874 0.36748 0.39012 0.27347 0.49575 0.00313 -0.04923
0.05395 0.02934 -0.01560 0.07428 -0.43985 0.45381 0.00625
-0.10355 0.10238 1.13357 0.93497 1.37435 0.15464 7.34375
V3.0 caps_cn 120000 20100511_030000 20100511_030000 000000
20100511_030000 20100511_030000 REFC L0 REFC SFC
ANALYS REFC_gt0.0 UW_MEAN 1 >=20.000 >=20.000 NA
NA FHO 12177 0.34976 0.12778 0.30024
V3.0 caps_cn 120000 20100511_030000 20100511_030000 000000
20100511_030000 20100511_030000 REFC L0 REFC SFC
ANALYS REFC_gt0.0 UW_MEAN 1 >=20.000 >=20.000 NA
NA CTC 12177 1556 2703 2100 5818
V3.0 caps_cn 120000 20100511_030000 20100511_030000 000000
20100511_030000 20100511_030000 REFC L0 REFC SFC
ANALYS REFC_gt0.0 UW_MEAN 1 >=20.000 >=20.000 NA
0.01000 CTS 12177 0.30024 0.28965 0.31104 0.19672 0.40984
0.34976 0.33871 0.36097 0.24586 0.46721 0.60557 0.59411 0.61692
0.48361 0.72955 1.16493 0.72726 1.91668 0.42560 0.41410 0.43718
0.21051 0.65794 0.68278 0.67182 0.69355 0.54537 0.81321 0.31722
0.30645 0.32818 0.18679 0.45463 0.63466 0.62335 0.64582 0.44679
0.81823 0.24469 0.23480 0.25486 0.10935 0.39289 0.05458 -0.05606
0.21264 0.10839 0.08322 0.13355 -0.13054 0.38107 0.10351
-0.11878 0.35070 1.59484 1.43592 1.77136 0.55206 5.28762
V3.0 caps_cn 120000 20100511_030000 20100511_030000 000000
20100511_030000 20100511_030000 REFC L0 REFC SFC
ANALYS REFC_gt0.0 UW_MEAN 1 >=30.000 >=30.000 NA
NA FHO 12177 0.23298 0.04738 0.09953
V3.0 caps_cn 120000 20100511_030000 20100511_030000 000000
20100511_030000 20100511_030000 REFC L0 REFC SFC
ANALYS REFC_gt0.0 UW_MEAN 1 >=30.000 >=30.000 NA
NA CTC 12177 577 2260 635 8705
V3.0 caps_cn 120000 20100511_030000 20100511_030000 000000
20100511_030000 20100511_030000 REFC L0 REFC SFC
ANALYS REFC_gt0.0 UW_MEAN 1 >=30.000 >=30.000 NA
0.01000 CTS 12177 0.09953 0.09276 0.10674 0.04098 0.18037
0.23298 0.22326 0.24299 0.14754 0.32791 0.76226 0.75218 0.77205
0.65574 0.85246 2.34076 1.19991 5.83333 0.47607 0.46443 0.48774
0.11106 0.87512 0.79389 0.78429 0.80317 0.69723 0.87964 0.20611
0.19683 0.21571 0.12036 0.30277 0.79662 0.78706 0.80585 0.55172
0.96155 0.16619 0.15768 0.17506 0.02941 0.35144 0.09237 -0.02395
0.27239 0.26996 0.23112 0.30881 -0.11436 0.65893 0.16912
-0.04908 0.42816 3.49995 2.98204 4.10782 0.41788 22.65536
V3.0 caps_cn 120000 20100511_030000 20100511_030000 000000
20100511_030000 20100511_030000 REFC L0 REFC SFC
ANALYS REFC_gt0.0 UW_MEAN 1 >=40.000 >=40.000 NA
NA FHO 12177 0.05149 0.00238 0.04435
V3.0 caps_cn 120000 20100511_030000 20100511_030000 000000
20100511_030000 20100511_030000 REFC L0 REFC SFC
ANALYS REFC_gt0.0 UW_MEAN 1 >=40.000 >=40.000 NA
NA CTC 12177 29 598 511 11039
V3.0 caps_cn 120000 20100511_030000 20100511_030000 000000
20100511_030000 20100511_030000 REFC L0 REFC SFC
ANALYS REFC_gt0.0 UW_MEAN 1 >=40.000 >=40.000 NA
0.01000 CTS 12177 0.04435 0.03978 0.04940 0.00820 0.10656
0.05149 0.04657 0.05690 0.00820 0.09840 0.90893 0.90199 0.91542
0.83607 0.97541 1.16111 0.12500 8.00000 0.05370 0.04868 0.05921
0.00000 0.50000 0.94861 0.94321 0.95353 0.89655 0.99152 0.05139
0.04647 0.05679 0.00847 0.10345 0.95375 0.94859 0.95841 0.50000
1.00000 0.02548 0.02206 0.02943 0.00000 0.22222 0.00108 -0.04040
0.20037 0.00232 -0.02367 0.02830 -0.10169 0.48306 0.00215
-0.08420 0.33385 1.04762 0.63322 1.73322 0.00000 28.04375
V3.0 caps_cn 120000 20100511_030000 20100511_030000 000000
20100511_030000 20100511_030000 REFC L0 REFC SFC
ANALYS REFC_gt0.0 UW_MEAN 1 NA NA NA
0.01000 CNT 12177 12.52624 12.14237 12.91010 8.34394 16.25470
16.44488 16.17756 16.72057 14.10156 18.42731 18.78944 18.58038
18.99851 16.90726 20.79734 8.95643 8.81084 9.10658 6.63185 11.38427
0.12219 0.09913 0.14512 -0.10707 0.37096 NA NA 0 0
0 -6.26321 -6.67727 -5.84914 -10.72635 -2.69674 17.73860
17.45025 18.03598 14.91300 20.79172 0.66666 0.43591 0.85628 16.63527
14.51040 18.81384 353.85990 267.99010 458.29753 314.63216 220.57460
428.75203 18.81116 16.37040 21.40788 -22.95125 -26.49725 -19.32103
-18.09597 -20.19476 -15.91572 -12.93366 -15.45607 -6.71084 9.65668
-4.69587 15.41383 18.76647 13.30001 25.21999
V3.0 caps_cn 120000 20100511_030000 20100511_030000 000000
20100511_030000 20100511_030000 REFC L0 REFC SFC
ANALYS REFC_gt0.0 UW_MEAN 1 NA NA NA
NA SL1L2 12177 12.52624 18.78944 253.35642 427.31855
433.25419
------------------------------------------------
Subject: grid_stat errors
From: John Halley Gotway
Time: Mon Jan 10 12:27:56 2011
Derek,
OK, I was able to fix the overflow problem on my machine with just a
minor change to one file. Please try using the attached file
"compute_ci.cc" to replace the file
"METv3.0/lib/vx_met_util/compute_ci.cc". Then, from the top-level MET
directory, do a "make clean" followed by a "make" to rebuild MET.
Then try running this case again and make sure that the normal
confidence limits in the output
"grid_stat_REF_caps_cn_120000L_20100511_030000V_cts.txt" file contain
reasonable numbers rather than
"inf". Also, please see check to see if those original error messages
you saw are still there - or have they gone away?
Thanks,
John
On 01/10/2011 12:03 PM, RAL HelpDesk {for John Halley Gotway} wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=43337 >
>
> Derek,
>
> I ran it here, and as I suspected, I didn't see the messages you
saw. I've attached the output stat file Grid-Stat created. I do see
a problem in the Grid-Stat output however - I see "inf" and
> "-inf" values in the normal confidence limits. I ran it through the
debugger and think that we've got a variable overflow problem causing
this. Are you seeing "inf" and "-inf" in your Grid-Stat
> output as well?
>
> Perhaps that's the source of the messages you're seeing... or
perhaps not. Regardless, it's a bug we should try to fix.
>
> This is caused by computing statistics using 65536 matched pairs.
When I reduce the number of matched pairs used by performing data
masking on the observation (only looking at points where REFC > 0),
> I do get reasonable results for the confidence limits. The number
of matched pairs is reduced down to 12,177.
>
> I'll need to work on this and see if I can figure out a patch
that'll allow for large numbers of matched pairs (i.e. 65536).
>
> Thanks,
> John
>
>
> On 01/10/2011 10:13 AM, RAL HelpDesk {for Derek.Stratman} wrote:
>>
>> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=43337 >
>>
>> John,
>>
>> The files should be in the folder now. I run GridStat with:
>> grid_stat forecast.grb observation.nc GridStatConfig_process
-outdir ${outdirectory}
>>
>> Here's my machine's description:
>> Linux gallifrey.protect.nssl 2.6.18-194.26.1.el5 #1 SMP Fri Oct 29
14:25:52 EDT 2010 ia64
>> ia64 ia64 GNU/Linux
>>
>> Thanks,
>> Derek
>>
>> ----- Original Message -----
>> From: RAL HelpDesk {for John Halley Gotway} <met_help at ucar.edu>
>> Date: Monday, January 10, 2011 10:49 am
>> Subject: Re: [rt.rap.ucar.edu #43337] grid_stat errors
>>> Derek,
>>>
>>> No, I've never seen these error messages before. I did a quick
>>> search online for information about them, and it sounds like they
>>> may arise when the program allocates memory in atypical ways or
casts
>>> variables from one type to another. Sounds like they aren't a big
>>> deal and that the program often runs to completion without any
>>> problems. So I do think it's reasonable to ignore them for now.
>>>
>>> One thing you could try would be to recompile MET - from the top-
>>> level MET directory, do a "make clean" followed by a "make".
>>> Perhaps that'd help.
>>>
>>> I would of course like to replicate this warning message and fix
>>> it, but I'm not hopeful that I'll be able to. These types of
>>> system-dependent things are often very difficult to replicate.
But we
>>> might as well give it a shot. Assuming recompiling doesn't help,
>>> could you please send me the following information:
>>>
>>> (1) The files you use for running Grid-Stat: forecast file,
>>> observation file, and configuration file.
>>> (2) The command line you use to run Grid-Stat.
>>> (3) A description of your machine including operating system. Try
>>> typing "uname -a".
>>>
>>> You can post those data files to our anonymous ftp site by
>>> following these instructions:
>>> http://www.dtcenter.org/met/users/support/met_help.php#ftp
>>>
>>> Thanks,
>>> John
>>>
>>> On 01/07/2011 10:24 AM, RAL HelpDesk {for Derek.Stratman} wrote:
>>>>
>>>> Fri Jan 07 10:24:11 2011: Request 43337 was acted upon.
>>>> Transaction: Ticket created by Derek.Stratman at noaa.gov
>>>> Queue: met_help
>>>> Subject: grid_stat errors
>>>> Owner: Nobody
>>>> Requestors: Derek.Stratman at noaa.gov
>>>> Status: new
>>>> Ticket <URL:
>>> https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=43337 >
>>>>
>>>>
>>>> MET Help,
>>>>
>>>> I'm running grid_stat v3.0 and I get the following errors:
>>>>
>>>> grid_stat(18710): unaligned access to 0x60000fffffda6f5c,
>>> ip=0x400000000016fb61> grid_stat(18710): unaligned access to
>>> 0x60000fffffda6f5c, ip=0x400000000016fc10
>>>> grid_stat(18710): unaligned access to 0x60000fffffda6f5c,
>>> ip=0x400000000016fc31> grid_stat(18710): unaligned access to
>>> 0x60000fffffda6f5c, ip=0x400000000016fdb1
>>>> grid_stat(18710): unaligned access to 0x60000fffffda6f5c,
>>> ip=0x400000000016fdf1>
>>>> Gridstat runs fine even with these errors, and as far as I can
>>> tell, the output values are
>>>> reasonable. Do you know what these errors are from? Should I be
>>> concerned about them? If
>>>> you need test files, let me know.
>>>>
>>>> Thanks,
>>>> Derek Stratman
>>>>
>>>>
>>>
>>>
>
------------------------------------------------
Subject: Re: [rt.rap.ucar.edu #43337] grid_stat errors
From: Derek.Stratman
Time: Mon Jan 10 12:47:09 2011
John,
I may be getting "inf" and "-inf" in the output because I set
boot_rep_prop = 0.01 in the
configuration file in the hopes of saving computation time since I
don't need MET's
confidence intervals (i.e., I'm calculating my own error bars due to
spatial and temporal
biases). So basically, I only need the actual scores (GSS, CSI, etc.)
to be correct, which I think
they are based on what I'm seeing. Do you agree that that's the
reason for the "inf" and "-
inf" in the output?
Thanks,
Derek
----- Original Message -----
From: RAL HelpDesk {for John Halley Gotway} <met_help at ucar.edu>
Date: Monday, January 10, 2011 1:27 pm
Subject: Re: [rt.rap.ucar.edu #43337] grid_stat errors
> Derek,
>
> OK, I was able to fix the overflow problem on my machine with just
> a minor change to one file. Please try using the attached file
> "compute_ci.cc" to replace the file
> "METv3.0/lib/vx_met_util/compute_ci.cc". Then, from the top-level
> MET directory, do a "make clean" followed by a "make" to rebuild
MET.
>
> Then try running this case again and make sure that the normal
> confidence limits in the output
> "grid_stat_REF_caps_cn_120000L_20100511_030000V_cts.txt" file
> contain reasonable numbers rather than
> "inf". Also, please see check to see if those original error
> messages you saw are still there - or have they gone away?
>
> Thanks,
> John
>
> On 01/10/2011 12:03 PM, RAL HelpDesk {for John Halley Gotway} wrote:
> >
> > <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=43337 >
> >
> > Derek,
> >
> > I ran it here, and as I suspected, I didn't see the messages you
> saw. I've attached the output stat file Grid-Stat created. I do
> see a problem in the Grid-Stat output however - I see "inf" and
> > "-inf" values in the normal confidence limits. I ran it through
> the debugger and think that we've got a variable overflow problem
> causing this. Are you seeing "inf" and "-inf" in your Grid-Stat
> > output as well?
> >
> > Perhaps that's the source of the messages you're seeing... or
> perhaps not. Regardless, it's a bug we should try to fix.
> >
> > This is caused by computing statistics using 65536 matched pairs.
> When I reduce the number of matched pairs used by performing data
> masking on the observation (only looking at points where REFC > 0),
> > I do get reasonable results for the confidence limits. The
> number of matched pairs is reduced down to 12,177.
> >
> > I'll need to work on this and see if I can figure out a patch
> that'll allow for large numbers of matched pairs (i.e. 65536).
> >
> > Thanks,
> > John
> >
> >
> > On 01/10/2011 10:13 AM, RAL HelpDesk {for Derek.Stratman} wrote:
> >>
> >> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=43337 >
> >>
> >> John,
> >>
> >> The files should be in the folder now. I run GridStat with:
> >> grid_stat forecast.grb observation.nc GridStatConfig_process -
> outdir ${outdirectory}
> >>
> >> Here's my machine's description:
> >> Linux gallifrey.protect.nssl 2.6.18-194.26.1.el5 #1 SMP Fri Oct
> 29 14:25:52 EDT 2010 ia64
> >> ia64 ia64 GNU/Linux
> >>
> >> Thanks,
> >> Derek
> >>
> >> ----- Original Message -----
> >> From: RAL HelpDesk {for John Halley Gotway} <met_help at ucar.edu>
> >> Date: Monday, January 10, 2011 10:49 am
> >> Subject: Re: [rt.rap.ucar.edu #43337] grid_stat errors
> >>> Derek,
> >>>
> >>> No, I've never seen these error messages before. I did a quick
> >>> search online for information about them, and it sounds like
> they
> >>> may arise when the program allocates memory in atypical ways or
> casts>>> variables from one type to another. Sounds like they
> aren't a big
> >>> deal and that the program often runs to completion without any
> >>> problems. So I do think it's reasonable to ignore them for now.
> >>>
> >>> One thing you could try would be to recompile MET - from the
> top-
> >>> level MET directory, do a "make clean" followed by a "make".
> >>> Perhaps that'd help.
> >>>
> >>> I would of course like to replicate this warning message and
> fix
> >>> it, but I'm not hopeful that I'll be able to. These types of
> >>> system-dependent things are often very difficult to replicate.
> But we
> >>> might as well give it a shot. Assuming recompiling doesn't
> help,
> >>> could you please send me the following information:
> >>>
> >>> (1) The files you use for running Grid-Stat: forecast file,
> >>> observation file, and configuration file.
> >>> (2) The command line you use to run Grid-Stat.
> >>> (3) A description of your machine including operating system.
> Try
> >>> typing "uname -a".
> >>>
> >>> You can post those data files to our anonymous ftp site by
> >>> following these instructions:
> >>> http://www.dtcenter.org/met/users/support/met_help.php#ftp
> >>>
> >>> Thanks,
> >>> John
> >>>
> >>> On 01/07/2011 10:24 AM, RAL HelpDesk {for Derek.Stratman} wrote:
> >>>>
> >>>> Fri Jan 07 10:24:11 2011: Request 43337 was acted upon.
> >>>> Transaction: Ticket created by Derek.Stratman at noaa.gov
> >>>> Queue: met_help
> >>>> Subject: grid_stat errors
> >>>> Owner: Nobody
> >>>> Requestors: Derek.Stratman at noaa.gov
> >>>> Status: new
> >>>> Ticket <URL:
> >>> https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=43337 >
> >>>>
> >>>>
> >>>> MET Help,
> >>>>
> >>>> I'm running grid_stat v3.0 and I get the following errors:
> >>>>
> >>>> grid_stat(18710): unaligned access to 0x60000fffffda6f5c,
> >>> ip=0x400000000016fb61> grid_stat(18710): unaligned access to
> >>> 0x60000fffffda6f5c, ip=0x400000000016fc10
> >>>> grid_stat(18710): unaligned access to 0x60000fffffda6f5c,
> >>> ip=0x400000000016fc31> grid_stat(18710): unaligned access to
> >>> 0x60000fffffda6f5c, ip=0x400000000016fdb1
> >>>> grid_stat(18710): unaligned access to 0x60000fffffda6f5c,
> >>> ip=0x400000000016fdf1>
> >>>> Gridstat runs fine even with these errors, and as far as I can
> >>> tell, the output values are
> >>>> reasonable. Do you know what these errors are from? Should I
> be
> >>> concerned about them? If
> >>>> you need test files, let me know.
> >>>>
> >>>> Thanks,
> >>>> Derek Stratman
> >>>>
> >>>>
> >>>
> >>>
> >
>
>
------------------------------------------------
Subject: Re: [rt.rap.ucar.edu #43337] grid_stat errors
From: John Halley Gotway
Time: Mon Jan 10 12:54:09 2011
Derek,
No, that's not the reason. And the "inf" values are showing up in the
columns for the normal CI's, not the bootstrap CI's. By the way, if
you want to disable the bootstrap CI's, do so by setting the
number of replicates to 0:
n_boot_rep = 0;
Even when I set the number of replicates to 0, I still see the "inf"
values in the normal columns (_NCL and _NCU). It's due to variable
overflow in the "compute_wilson_ci" subroutine in the
"compute_ci.cc" file. I made a small change there to use a "long
long" rather than an "int" and that solved the problem.
Please update that file with the version I sent you, recompile MET,
and verify that it solve the "inf" problem on your machine as well.
I'd like the make sure it fixes it for you before posting it as
a bugfix on our website.
Thanks,
John
On 01/10/2011 12:47 PM, RAL HelpDesk {for Derek.Stratman} wrote:
> they are based on what I'm seeing. Do you agree that that's the
reason for the "inf" and "-
------------------------------------------------
Subject: Re: [rt.rap.ucar.edu #43337] grid_stat errors
From: Derek.Stratman
Time: Mon Jan 10 13:18:37 2011
John,
I'm in the middle of a two-day job running wavelet_stat, so once
that's done, I'll have MET
recompiled and tested for you. Until then, to clarify with you, the
skill score values (and
associated contingency table statistics) before the bug is fixed
should be the same once the
bug is fixed, correct? I'm asking since I'm only concerned with
those values and not the
confidence interval values for now. Time is limited for me with a
conference fast
approaching, so that's why I'm asking/clarifying.
Thanks,
Derek
----- Original Message -----
From: RAL HelpDesk {for John Halley Gotway} <met_help at ucar.edu>
Date: Monday, January 10, 2011 1:54 pm
Subject: Re: [rt.rap.ucar.edu #43337] grid_stat errors
> Derek,
>
> No, that's not the reason. And the "inf" values are showing up in
> the columns for the normal CI's, not the bootstrap CI's. By the
> way, if you want to disable the bootstrap CI's, do so by setting the
> number of replicates to 0:
> n_boot_rep = 0;
>
> Even when I set the number of replicates to 0, I still see the
> "inf" values in the normal columns (_NCL and _NCU). It's due to
> variable overflow in the "compute_wilson_ci" subroutine in the
> "compute_ci.cc" file. I made a small change there to use a "long
> long" rather than an "int" and that solved the problem.
>
> Please update that file with the version I sent you, recompile MET,
> and verify that it solve the "inf" problem on your machine as well.
> I'd like the make sure it fixes it for you before posting it as
> a bugfix on our website.
>
> Thanks,
> John
>
> On 01/10/2011 12:47 PM, RAL HelpDesk {for Derek.Stratman} wrote:
> > they are based on what I'm seeing. Do you agree that that's the
> reason for the "inf" and "-
>
>
------------------------------------------------
Subject: Re: [rt.rap.ucar.edu #43337] grid_stat errors
From: John Halley Gotway
Time: Mon Jan 10 13:22:27 2011
Derek,
Yes, that's correct. The values of the actual statistics are correct
and remain unchanged. The fix gives you actual normal CI's rather
than -inf to inf.
Thanks,
John
On 01/10/2011 01:18 PM, RAL HelpDesk {for Derek.Stratman} wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=43337 >
>
> John,
>
> I'm in the middle of a two-day job running wavelet_stat, so once
that's done, I'll have MET
> recompiled and tested for you. Until then, to clarify with you, the
skill score values (and
> associated contingency table statistics) before the bug is fixed
should be the same once the
> bug is fixed, correct? I'm asking since I'm only concerned with
those values and not the
> confidence interval values for now. Time is limited for me with a
conference fast
> approaching, so that's why I'm asking/clarifying.
>
> Thanks,
> Derek
>
> ----- Original Message -----
> From: RAL HelpDesk {for John Halley Gotway} <met_help at ucar.edu>
> Date: Monday, January 10, 2011 1:54 pm
> Subject: Re: [rt.rap.ucar.edu #43337] grid_stat errors
>> Derek,
>>
>> No, that's not the reason. And the "inf" values are showing up in
>> the columns for the normal CI's, not the bootstrap CI's. By the
>> way, if you want to disable the bootstrap CI's, do so by setting
the
>> number of replicates to 0:
>> n_boot_rep = 0;
>>
>> Even when I set the number of replicates to 0, I still see the
>> "inf" values in the normal columns (_NCL and _NCU). It's due to
>> variable overflow in the "compute_wilson_ci" subroutine in the
>> "compute_ci.cc" file. I made a small change there to use a "long
>> long" rather than an "int" and that solved the problem.
>>
>> Please update that file with the version I sent you, recompile MET,
>> and verify that it solve the "inf" problem on your machine as well.
>> I'd like the make sure it fixes it for you before posting it as
>> a bugfix on our website.
>>
>> Thanks,
>> John
>>
>> On 01/10/2011 12:47 PM, RAL HelpDesk {for Derek.Stratman} wrote:
>>> they are based on what I'm seeing. Do you agree that that's the
>> reason for the "inf" and "-
>>
>>
------------------------------------------------
Subject: Re: [rt.rap.ucar.edu #43337] grid_stat errors
From: Derek.Stratman
Time: Mon Jan 10 13:31:02 2011
John,
Thanks for the info. I'll have the bug fixed once my current job is
done running. That won't
be until later tomorrow, but once that's done, I'll update MET and let
you know if the fix
solves the CI problem.
Thanks for your help,
Derek
----- Original Message -----
From: RAL HelpDesk {for John Halley Gotway} <met_help at ucar.edu>
Date: Monday, January 10, 2011 2:22 pm
Subject: Re: [rt.rap.ucar.edu #43337] grid_stat errors
> Derek,
>
> Yes, that's correct. The values of the actual statistics are
> correct and remain unchanged. The fix gives you actual normal CI's
> rather than -inf to inf.
>
> Thanks,
> John
>
> On 01/10/2011 01:18 PM, RAL HelpDesk {for Derek.Stratman} wrote:
> >
> > <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=43337 >
> >
> > John,
> >
> > I'm in the middle of a two-day job running wavelet_stat, so once
> that's done, I'll have MET
> > recompiled and tested for you. Until then, to clarify with you,
> the skill score values (and
> > associated contingency table statistics) before the bug is fixed
> should be the same once the
> > bug is fixed, correct? I'm asking since I'm only concerned with
> those values and not the
> > confidence interval values for now. Time is limited for me with
> a conference fast
> > approaching, so that's why I'm asking/clarifying.
> >
> > Thanks,
> > Derek
> >
> > ----- Original Message -----
> > From: RAL HelpDesk {for John Halley Gotway} <met_help at ucar.edu>
> > Date: Monday, January 10, 2011 1:54 pm
> > Subject: Re: [rt.rap.ucar.edu #43337] grid_stat errors
> >> Derek,
> >>
> >> No, that's not the reason. And the "inf" values are showing up
> in
> >> the columns for the normal CI's, not the bootstrap CI's. By the
> >> way, if you want to disable the bootstrap CI's, do so by setting
> the>> number of replicates to 0:
> >> n_boot_rep = 0;
> >>
> >> Even when I set the number of replicates to 0, I still see the
> >> "inf" values in the normal columns (_NCL and _NCU). It's due to
> >> variable overflow in the "compute_wilson_ci" subroutine in the
> >> "compute_ci.cc" file. I made a small change there to use a
> "long
> >> long" rather than an "int" and that solved the problem.
> >>
> >> Please update that file with the version I sent you, recompile
> MET,
> >> and verify that it solve the "inf" problem on your machine as
> well.
> >> I'd like the make sure it fixes it for you before posting it as
> >> a bugfix on our website.
> >>
> >> Thanks,
> >> John
> >>
> >> On 01/10/2011 12:47 PM, RAL HelpDesk {for Derek.Stratman} wrote:
> >>> they are based on what I'm seeing. Do you agree that that's
> the
> >> reason for the "inf" and "-
> >>
> >>
>
>
------------------------------------------------
Subject: Re: [rt.rap.ucar.edu #43337] grid_stat errors
From: Derek.Stratman
Time: Wed Jan 19 14:08:03 2011
John,
Sorry for the long delay. Someone else is working on fixing the bug
since MET isn't ran from
my personal work computer, but they found the following errors when
working on it.
-Derek
*** Making vx_met_util Library ***
/usr/bin/g++ apply_mask.cc -Wall -Wshadow -static -c
-DMET_BASE=\"/usr/local/METv3.0\" -I.. -I/usr/include
-I/usr/local/netcdf3/include
/usr/bin/g++ compute_ci.cc -Wall -Wshadow -static -c -I..
-I/usr/include
compute_ci.cc: In function ‘void compute_cts_stats_ci_bca(const
gsl_rng*, const NumArray&, const NumArray&, int, CTSInfo*&, int, int,
int, const char*)’:
compute_ci.cc:327: error: ‘make_temp_file_name’ was not declared in
this
scope
compute_ci.cc:570: error: ‘remove_temp_file’ was not declared in this
scope
compute_ci.cc:581: error: ‘remove_temp_file’ was not declared in this
scope
compute_ci.cc: In function ‘void compute_mcts_stats_ci_bca(const
gsl_rng*, const NumArray&, const NumArray&, int, MCTSInfo&, int, int,
const char*)’:
compute_ci.cc:667: error: ‘make_temp_file_name’ was not declared in
this
scope
compute_ci.cc:786: error: ‘remove_temp_file’ was not declared in this
scope
compute_ci.cc:795: error: ‘remove_temp_file’ was not declared in this
scope
compute_ci.cc: In function ‘void compute_cnt_stats_ci_bca(const
gsl_rng*, const NumArray&, const NumArray&, int, int, int, CNTInfo&,
int, int, const char*)’:
compute_ci.cc:868: error: ‘make_temp_file_name’ was not declared in
this
scope
compute_ci.cc:1145: error: ‘remove_temp_file’ was not declared in this
scope
compute_ci.cc:1154: error: ‘remove_temp_file’ was not declared in this
scope
compute_ci.cc: In function ‘void compute_cts_stats_ci_perc(const
gsl_rng*, const NumArray&, const NumArray&, int, double, CTSInfo*&,
int,
int, int, const char*)’:
compute_ci.cc:1241: error: ‘make_temp_file_name’ was not declared in
this scope
compute_ci.cc:1449: error: ‘remove_temp_file’ was not declared in this
scope
compute_ci.cc:1458: error: ‘remove_temp_file’ was not declared in this
scope
compute_ci.cc: In function ‘void compute_mcts_stats_ci_perc(const
gsl_rng*, const NumArray&, const NumArray&, int, double, MCTSInfo&,
int,
int, const char*)’:
compute_ci.cc:1542: error: ‘make_temp_file_name’ was not declared in
this scope
compute_ci.cc:1639: error: ‘remove_temp_file’ was not declared in this
scope
compute_ci.cc:1647: error: ‘remove_temp_file’ was not declared in this
scope
compute_ci.cc: In function ‘void compute_cnt_stats_ci_perc(const
gsl_rng*, const NumArray&, const NumArray&, int, int, int, double,
CNTInfo&, int, int, const char*)’:
compute_ci.cc:1719: error: ‘make_temp_file_name’ was not declared in
this scope
compute_ci.cc:1962: error: ‘remove_temp_file’ was not declared in this
scope
compute_ci.cc:1970: error: ‘remove_temp_file’ was not declared in this
scope
compute_ci.cc: In function ‘void compute_nbrcts_stats_ci_bca(const
gsl_rng*, const NumArray&, const NumArray&, int, NBRCTSInfo*&, int,
int,
const char*)’:
compute_ci.cc:2050: error: ‘make_temp_file_name’ was not declared in
this scope
compute_ci.cc:2293: error: ‘remove_temp_file’ was not declared in this
scope
compute_ci.cc:2304: error: ‘remove_temp_file’ was not declared in this
scope
compute_ci.cc: In function ‘void compute_nbrcnt_stats_ci_bca(const
gsl_rng*, const NumArray&, const NumArray&, int, NBRCNTInfo&, int,
const
char*)’:
compute_ci.cc:2373: error: ‘make_temp_file_name’ was not declared in
this scope
compute_ci.cc:2466: error: ‘remove_temp_file’ was not declared in this
scope
compute_ci.cc:2475: error: ‘remove_temp_file’ was not declared in this
scope
compute_ci.cc: In function ‘void compute_nbrcts_stats_ci_perc(const
gsl_rng*, const NumArray&, const NumArray&, int, double, NBRCTSInfo*&,
int, int, const char*)’:
----- Original Message -----
From: RAL HelpDesk {for John Halley Gotway} <met_help at ucar.edu>
Date: Monday, January 10, 2011 2:22 pm
Subject: Re: [rt.rap.ucar.edu #43337] grid_stat errors
> Derek,
>
> Yes, that's correct. The values of the actual statistics are
> correct and remain unchanged. The fix gives you actual normal CI's
> rather than -inf to inf.
>
> Thanks,
> John
>
> On 01/10/2011 01:18 PM, RAL HelpDesk {for Derek.Stratman} wrote:
> >
> > <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=43337 >
> >
> > John,
> >
> > I'm in the middle of a two-day job running wavelet_stat, so once
> that's done, I'll have MET
> > recompiled and tested for you. Until then, to clarify with you,
> the skill score values (and
> > associated contingency table statistics) before the bug is fixed
> should be the same once the
> > bug is fixed, correct? I'm asking since I'm only concerned with
> those values and not the
> > confidence interval values for now. Time is limited for me with
> a conference fast
> > approaching, so that's why I'm asking/clarifying.
> >
> > Thanks,
> > Derek
> >
> > ----- Original Message -----
> > From: RAL HelpDesk {for John Halley Gotway} <met_help at ucar.edu>
> > Date: Monday, January 10, 2011 1:54 pm
> > Subject: Re: [rt.rap.ucar.edu #43337] grid_stat errors
> >> Derek,
> >>
> >> No, that's not the reason. And the "inf" values are showing up
> in
> >> the columns for the normal CI's, not the bootstrap CI's. By the
> >> way, if you want to disable the bootstrap CI's, do so by setting
> the>> number of replicates to 0:
> >> n_boot_rep = 0;
> >>
> >> Even when I set the number of replicates to 0, I still see the
> >> "inf" values in the normal columns (_NCL and _NCU). It's due to
> >> variable overflow in the "compute_wilson_ci" subroutine in the
> >> "compute_ci.cc" file. I made a small change there to use a
> "long
> >> long" rather than an "int" and that solved the problem.
> >>
> >> Please update that file with the version I sent you, recompile
> MET,
> >> and verify that it solve the "inf" problem on your machine as
> well.
> >> I'd like the make sure it fixes it for you before posting it as
> >> a bugfix on our website.
> >>
> >> Thanks,
> >> John
> >>
> >> On 01/10/2011 12:47 PM, RAL HelpDesk {for Derek.Stratman} wrote:
> >>> they are based on what I'm seeing. Do you agree that that's
> the
> >> reason for the "inf" and "-
> >>
> >>
>
>
------------------------------------------------
Subject: Re: [rt.rap.ucar.edu #43337] grid_stat errors
From: John Halley Gotway
Time: Fri Jan 21 15:56:22 2011
Derek,
OK, I see what's going on. Those changes to that file I sent you
(compute_ci.cc) were built on top of the latest set patches for
METv3.0. In fact, that update I sent you is now included in the
latest set of patches for METv3.0.
So I'd recommend retrieving the latest set of patches from here:
http://www.dtcenter.org/met/users/support/known_issues/METv3.0/index.php
Just follow the instructions at the top of the page to retrieve and
apply the patches. Then recompile MET, being sure to do a 'make
clean' first.
Just let me know if you guys run into any issues or continue to
experience the problem you were having.
Thanks,
John
On 01/19/2011 02:08 PM, RAL HelpDesk {for Derek.Stratman} wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=43337 >
>
> John,
>
> Sorry for the long delay. Someone else is working on fixing the
bug since MET isn't ran from
> my personal work computer, but they found the following errors when
working on it.
>
> -Derek
>
>
> *** Making vx_met_util Library ***
>
> /usr/bin/g++ apply_mask.cc -Wall -Wshadow -static -c
> -DMET_BASE=\"/usr/local/METv3.0\" -I.. -I/usr/include
> -I/usr/local/netcdf3/include
> /usr/bin/g++ compute_ci.cc -Wall -Wshadow -static -c -I..
-I/usr/include
> compute_ci.cc: In function ‘void compute_cts_stats_ci_bca(const
> gsl_rng*, const NumArray&, const NumArray&, int, CTSInfo*&, int,
int,
> int, const char*)’:
> compute_ci.cc:327: error: ‘make_temp_file_name’ was not declared in
this
> scope
> compute_ci.cc:570: error: ‘remove_temp_file’ was not declared in
this scope
> compute_ci.cc:581: error: ‘remove_temp_file’ was not declared in
this scope
> compute_ci.cc: In function ‘void compute_mcts_stats_ci_bca(const
> gsl_rng*, const NumArray&, const NumArray&, int, MCTSInfo&, int,
int,
> const char*)’:
> compute_ci.cc:667: error: ‘make_temp_file_name’ was not declared in
this
> scope
> compute_ci.cc:786: error: ‘remove_temp_file’ was not declared in
this scope
> compute_ci.cc:795: error: ‘remove_temp_file’ was not declared in
this scope
> compute_ci.cc: In function ‘void compute_cnt_stats_ci_bca(const
> gsl_rng*, const NumArray&, const NumArray&, int, int, int, CNTInfo&,
> int, int, const char*)’:
> compute_ci.cc:868: error: ‘make_temp_file_name’ was not declared in
this
> scope
> compute_ci.cc:1145: error: ‘remove_temp_file’ was not declared in
this scope
> compute_ci.cc:1154: error: ‘remove_temp_file’ was not declared in
this scope
> compute_ci.cc: In function ‘void compute_cts_stats_ci_perc(const
> gsl_rng*, const NumArray&, const NumArray&, int, double, CTSInfo*&,
int,
> int, int, const char*)’:
> compute_ci.cc:1241: error: ‘make_temp_file_name’ was not declared in
> this scope
> compute_ci.cc:1449: error: ‘remove_temp_file’ was not declared in
this scope
> compute_ci.cc:1458: error: ‘remove_temp_file’ was not declared in
this scope
> compute_ci.cc: In function ‘void compute_mcts_stats_ci_perc(const
> gsl_rng*, const NumArray&, const NumArray&, int, double, MCTSInfo&,
int,
> int, const char*)’:
> compute_ci.cc:1542: error: ‘make_temp_file_name’ was not declared in
> this scope
> compute_ci.cc:1639: error: ‘remove_temp_file’ was not declared in
this scope
> compute_ci.cc:1647: error: ‘remove_temp_file’ was not declared in
this scope
> compute_ci.cc: In function ‘void compute_cnt_stats_ci_perc(const
> gsl_rng*, const NumArray&, const NumArray&, int, int, int, double,
> CNTInfo&, int, int, const char*)’:
> compute_ci.cc:1719: error: ‘make_temp_file_name’ was not declared in
> this scope
> compute_ci.cc:1962: error: ‘remove_temp_file’ was not declared in
this scope
> compute_ci.cc:1970: error: ‘remove_temp_file’ was not declared in
this scope
> compute_ci.cc: In function ‘void compute_nbrcts_stats_ci_bca(const
> gsl_rng*, const NumArray&, const NumArray&, int, NBRCTSInfo*&, int,
int,
> const char*)’:
> compute_ci.cc:2050: error: ‘make_temp_file_name’ was not declared in
> this scope
> compute_ci.cc:2293: error: ‘remove_temp_file’ was not declared in
this scope
> compute_ci.cc:2304: error: ‘remove_temp_file’ was not declared in
this scope
> compute_ci.cc: In function ‘void compute_nbrcnt_stats_ci_bca(const
> gsl_rng*, const NumArray&, const NumArray&, int, NBRCNTInfo&, int,
const
> char*)’:
> compute_ci.cc:2373: error: ‘make_temp_file_name’ was not declared in
> this scope
> compute_ci.cc:2466: error: ‘remove_temp_file’ was not declared in
this scope
> compute_ci.cc:2475: error: ‘remove_temp_file’ was not declared in
this scope
> compute_ci.cc: In function ‘void compute_nbrcts_stats_ci_perc(const
> gsl_rng*, const NumArray&, const NumArray&, int, double,
NBRCTSInfo*&,
> int, int, const char*)’:
>
> ----- Original Message -----
> From: RAL HelpDesk {for John Halley Gotway} <met_help at ucar.edu>
> Date: Monday, January 10, 2011 2:22 pm
> Subject: Re: [rt.rap.ucar.edu #43337] grid_stat errors
>> Derek,
>>
>> Yes, that's correct. The values of the actual statistics are
>> correct and remain unchanged. The fix gives you actual normal CI's
>> rather than -inf to inf.
>>
>> Thanks,
>> John
>>
>> On 01/10/2011 01:18 PM, RAL HelpDesk {for Derek.Stratman} wrote:
>>>
>>> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=43337 >
>>>
>>> John,
>>>
>>> I'm in the middle of a two-day job running wavelet_stat, so once
>> that's done, I'll have MET
>>> recompiled and tested for you. Until then, to clarify with you,
>> the skill score values (and
>>> associated contingency table statistics) before the bug is fixed
>> should be the same once the
>>> bug is fixed, correct? I'm asking since I'm only concerned with
>> those values and not the
>>> confidence interval values for now. Time is limited for me with
>> a conference fast
>>> approaching, so that's why I'm asking/clarifying.
>>>
>>> Thanks,
>>> Derek
>>>
>>> ----- Original Message -----
>>> From: RAL HelpDesk {for John Halley Gotway} <met_help at ucar.edu>
>>> Date: Monday, January 10, 2011 1:54 pm
>>> Subject: Re: [rt.rap.ucar.edu #43337] grid_stat errors
>>>> Derek,
>>>>
>>>> No, that's not the reason. And the "inf" values are showing up
>> in
>>>> the columns for the normal CI's, not the bootstrap CI's. By the
>>>> way, if you want to disable the bootstrap CI's, do so by setting
>> the>> number of replicates to 0:
>>>> n_boot_rep = 0;
>>>>
>>>> Even when I set the number of replicates to 0, I still see the
>>>> "inf" values in the normal columns (_NCL and _NCU). It's due to
>>>> variable overflow in the "compute_wilson_ci" subroutine in the
>>>> "compute_ci.cc" file. I made a small change there to use a
>> "long
>>>> long" rather than an "int" and that solved the problem.
>>>>
>>>> Please update that file with the version I sent you, recompile
>> MET,
>>>> and verify that it solve the "inf" problem on your machine as
>> well.
>>>> I'd like the make sure it fixes it for you before posting it as
>>>> a bugfix on our website.
>>>>
>>>> Thanks,
>>>> John
>>>>
>>>> On 01/10/2011 12:47 PM, RAL HelpDesk {for Derek.Stratman} wrote:
>>>>> they are based on what I'm seeing. Do you agree that that's
>> the
>>>> reason for the "inf" and "-
>>>>
>>>>
>>
>>
------------------------------------------------
Subject: Re: [rt.rap.ucar.edu #43337] grid_stat errors
From: Derek.Stratman
Time: Mon Feb 07 14:18:48 2011
Hi John,
Sorry for the very long delay. I'm not getting any errors now, so
this thread can be closed. Thanks for your help!
Derek
----- Original Message -----
From: RAL HelpDesk {for John Halley Gotway} <met_help at ucar.edu>
Date: Friday, January 21, 2011 4:56 pm
Subject: Re: [rt.rap.ucar.edu #43337] grid_stat errors
> Derek,
>
> OK, I see what's going on. Those changes to that file I sent you
> (compute_ci.cc) were built on top of the latest set patches for
> METv3.0. In fact, that update I sent you is now included in the
> latest set of patches for METv3.0.
>
> So I'd recommend retrieving the latest set of patches from here:
>
>
http://www.dtcenter.org/met/users/support/known_issues/METv3.0/index.php
> Just follow the instructions at the top of the page to retrieve and
> apply the patches. Then recompile MET, being sure to do a 'make
> clean' first.
>
> Just let me know if you guys run into any issues or continue to
> experience the problem you were having.
>
> Thanks,
> John
>
> On 01/19/2011 02:08 PM, RAL HelpDesk {for Derek.Stratman} wrote:
> >
> > <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=43337 >
> >
> > John,
> >
> > Sorry for the long delay. Someone else is working on fixing the
> bug since MET isn't ran from
> > my personal work computer, but they found the following errors
> when working on it.
> >
> > -Derek
> >
> >
> > *** Making vx_met_util Library ***
> >
> > /usr/bin/g++ apply_mask.cc -Wall -Wshadow -static -c
> > -DMET_BASE=\"/usr/local/METv3.0\" -I.. -I/usr/include
> > -I/usr/local/netcdf3/include
> > /usr/bin/g++ compute_ci.cc -Wall -Wshadow -static -c -I.. -
> I/usr/include> compute_ci.cc: In function ‘void
> compute_cts_stats_ci_bca(const
> > gsl_rng*, const NumArray&, const NumArray&, int, CTSInfo*&, int,
> int,
> > int, const char*)’:
> > compute_ci.cc:327: error: ‘make_temp_file_name’ was not declared
> in this
> > scope
> > compute_ci.cc:570: error: ‘remove_temp_file’ was not declared in
> this scope
> > compute_ci.cc:581: error: ‘remove_temp_file’ was not declared in
> this scope
> > compute_ci.cc: In function ‘void compute_mcts_stats_ci_bca(const
> > gsl_rng*, const NumArray&, const NumArray&, int, MCTSInfo&, int,
> int,
> > const char*)’:
> > compute_ci.cc:667: error: ‘make_temp_file_name’ was not declared
> in this
> > scope
> > compute_ci.cc:786: error: ‘remove_temp_file’ was not declared in
> this scope
> > compute_ci.cc:795: error: ‘remove_temp_file’ was not declared in
> this scope
> > compute_ci.cc: In function ‘void compute_cnt_stats_ci_bca(const
> > gsl_rng*, const NumArray&, const NumArray&, int, int, int,
> CNTInfo&,
> > int, int, const char*)’:
> > compute_ci.cc:868: error: ‘make_temp_file_name’ was not declared
> in this
> > scope
> > compute_ci.cc:1145: error: ‘remove_temp_file’ was not declared in
> this scope
> > compute_ci.cc:1154: error: ‘remove_temp_file’ was not declared in
> this scope
> > compute_ci.cc: In function ‘void compute_cts_stats_ci_perc(const
> > gsl_rng*, const NumArray&, const NumArray&, int, double,
> CTSInfo*&, int,
> > int, int, const char*)’:
> > compute_ci.cc:1241: error: ‘make_temp_file_name’ was not declared
> in
> > this scope
> > compute_ci.cc:1449: error: ‘remove_temp_file’ was not declared in
> this scope
> > compute_ci.cc:1458: error: ‘remove_temp_file’ was not declared in
> this scope
> > compute_ci.cc: In function ‘void compute_mcts_stats_ci_perc(const
> > gsl_rng*, const NumArray&, const NumArray&, int, double,
> MCTSInfo&, int,
> > int, const char*)’:
> > compute_ci.cc:1542: error: ‘make_temp_file_name’ was not declared
> in
> > this scope
> > compute_ci.cc:1639: error: ‘remove_temp_file’ was not declared in
> this scope
> > compute_ci.cc:1647: error: ‘remove_temp_file’ was not declared in
> this scope
> > compute_ci.cc: In function ‘void compute_cnt_stats_ci_perc(const
> > gsl_rng*, const NumArray&, const NumArray&, int, int, int,
> double,
> > CNTInfo&, int, int, const char*)’:
> > compute_ci.cc:1719: error: ‘make_temp_file_name’ was not declared
> in
> > this scope
> > compute_ci.cc:1962: error: ‘remove_temp_file’ was not declared in
> this scope
> > compute_ci.cc:1970: error: ‘remove_temp_file’ was not declared in
> this scope
> > compute_ci.cc: In function ‘void
> compute_nbrcts_stats_ci_bca(const
> > gsl_rng*, const NumArray&, const NumArray&, int, NBRCTSInfo*&,
> int, int,
> > const char*)’:
> > compute_ci.cc:2050: error: ‘make_temp_file_name’ was not declared
> in
> > this scope
> > compute_ci.cc:2293: error: ‘remove_temp_file’ was not declared in
> this scope
> > compute_ci.cc:2304: error: ‘remove_temp_file’ was not declared in
> this scope
> > compute_ci.cc: In function ‘void
> compute_nbrcnt_stats_ci_bca(const
> > gsl_rng*, const NumArray&, const NumArray&, int, NBRCNTInfo&,
> int, const
> > char*)’:
> > compute_ci.cc:2373: error: ‘make_temp_file_name’ was not declared
> in
> > this scope
> > compute_ci.cc:2466: error: ‘remove_temp_file’ was not declared in
> this scope
> > compute_ci.cc:2475: error: ‘remove_temp_file’ was not declared in
> this scope
> > compute_ci.cc: In function ‘void
> compute_nbrcts_stats_ci_perc(const
> > gsl_rng*, const NumArray&, const NumArray&, int, double,
> NBRCTSInfo*&,
> > int, int, const char*)’:
> >
> > ----- Original Message -----
> > From: RAL HelpDesk {for John Halley Gotway} <met_help at ucar.edu>
> > Date: Monday, January 10, 2011 2:22 pm
> > Subject: Re: [rt.rap.ucar.edu #43337] grid_stat errors
> >> Derek,
> >>
> >> Yes, that's correct. The values of the actual statistics are
> >> correct and remain unchanged. The fix gives you actual normal
> CI's
> >> rather than -inf to inf.
> >>
> >> Thanks,
> >> John
> >>
> >> On 01/10/2011 01:18 PM, RAL HelpDesk {for Derek.Stratman} wrote:
> >>>
> >>> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=43337 >
> >>>
> >>> John,
> >>>
> >>> I'm in the middle of a two-day job running wavelet_stat, so
> once
> >> that's done, I'll have MET
> >>> recompiled and tested for you. Until then, to clarify with
> you,
> >> the skill score values (and
> >>> associated contingency table statistics) before the bug is
> fixed
> >> should be the same once the
> >>> bug is fixed, correct? I'm asking since I'm only concerned
> with
> >> those values and not the
> >>> confidence interval values for now. Time is limited for me
> with
> >> a conference fast
> >>> approaching, so that's why I'm asking/clarifying.
> >>>
> >>> Thanks,
> >>> Derek
> >>>
> >>> ----- Original Message -----
> >>> From: RAL HelpDesk {for John Halley Gotway} <met_help at ucar.edu>
> >>> Date: Monday, January 10, 2011 1:54 pm
> >>> Subject: Re: [rt.rap.ucar.edu #43337] grid_stat errors
> >>>> Derek,
> >>>>
> >>>> No, that's not the reason. And the "inf" values are showing
> up
> >> in
> >>>> the columns for the normal CI's, not the bootstrap CI's. By
> the
> >>>> way, if you want to disable the bootstrap CI's, do so by
> setting
> >> the>> number of replicates to 0:
> >>>> n_boot_rep = 0;
> >>>>
> >>>> Even when I set the number of replicates to 0, I still see the
> >>>> "inf" values in the normal columns (_NCL and _NCU). It's due
> to
> >>>> variable overflow in the "compute_wilson_ci" subroutine in the
> >>>> "compute_ci.cc" file. I made a small change there to use a
> >> "long
> >>>> long" rather than an "int" and that solved the problem.
> >>>>
> >>>> Please update that file with the version I sent you, recompile
> >> MET,
> >>>> and verify that it solve the "inf" problem on your machine as
> >> well.
> >>>> I'd like the make sure it fixes it for you before posting it as
> >>>> a bugfix on our website.
> >>>>
> >>>> Thanks,
> >>>> John
> >>>>
> >>>> On 01/10/2011 12:47 PM, RAL HelpDesk {for Derek.Stratman}
wrote:
> >>>>> they are based on what I'm seeing. Do you agree that that's
> >> the
> >>>> reason for the "inf" and "-
> >>>>
> >>>>
> >>
> >>
>
>
------------------------------------------------
More information about the Met_help
mailing list