Resent Invalid and Validate errors de_modfit_85_bundle4_4s_south4s_1_

Has any one else noticed a high number of invalid errors with de_modfit_85_bundle4_4s_south4s_1_ work units?

---

Yes, I have. Seems quite similar to the xxx84xxx0 situation not too long ago. I am aborting the relevant tasks again until the problem is solved. I am keeping an eye out for other anomalies.

---

Plus SETI Classic = 21,082 WUs

Can confirm. Since last week I'm getting many 85s invalidated. Recently I also can see some marked as 'Too many errors'.

Sorry but I'm giving up with this project until somebody sorts out the frequent issues as no one seems to be bothered at this time. Great pity.

---

Seems like de_modfit_84_bundle4_4s_south4s_1_1561047003_ are also creating a lot of invalid results.
Got 100 invalid results yesterday, only 85 and 84.

Right now I have 13 invalid task on this computer, with 98 validated and 22 inconclusive. All the invalid tasks are Separation v1.46 tasks. This appears to be happening with de_modfit_84 and de_modfit_85 files.

Here is the Stderr for one of the files:

<core_client_version>7.14.2</core_client_version>
<![CDATA[
<stderr_txt>
<search_application> milkyway_separation 1.46 Windows x86_64 double </search_application>
Reading preferences ended prematurely
Error loading Lua script 'astronomy_parameters.txt': [string "number_parameters: 4..."]:1: '<name>' expected near '4' 
Switching to Parameter File 'astronomy_parameters.txt'
<number_WUs> 4 </number_WUs>
<number_params_per_WU> 26 </number_params_per_WU>
Using SSE4.1 path
Integral 0 time = 1012.482696 s
Integral 1 time = 72.815637 s
Running likelihood with 31964 stars
Likelihood time = 0.742747 s
Non-finite result:  setting likelihood to -999
<background_integral> 0.000037570759755 </background_integral>
<stream_integral>  99.770597784118664  -0.000000000000002  2.975633165946907  31.903999743820332 </stream_integral>
<background_likelihood> -3.304855300694308 </background_likelihood>
<stream_only_likelihood>  -3.850858455061610  -1.#IND00000000000  -33.432437063013211  -3.345848296216324 </stream_only_likelihood>
<search_likelihood> -999.000000000000000 </search_likelihood>
Using SSE4.1 path
Integral 0 time = 962.681409 s
Integral 1 time = 66.608038 s
Running likelihood with 31964 stars
Likelihood time = 0.714490 s
Non-finite result:  setting likelihood to -999
<background_integral1> 0.000056193474869 </background_integral1>
<stream_integral1>  112.923813599446660  -0.000000000000000  6.056076443722354  34.858924693631693 </stream_integral1>
<background_likelihood1> -3.277210709097926 </background_likelihood1>
<stream_only_likelihood1>  -3.458606239195324  -1.#IND00000000000  -90.323903172119387  -3.362009522740810 </stream_only_likelihood1>
<search_likelihood1> -999.000000000000000 </search_likelihood1>
Using SSE4.1 path
Integral 0 time = 988.222037 s
Integral 1 time = 72.907524 s
Running likelihood with 31964 stars
Likelihood time = 0.777539 s
<background_integral2> 0.000055474761194 </background_integral2>
<stream_integral2>  138.240648517785560  0.000000000000000  3.273614114441112  32.353426341780249 </stream_integral2>
<background_likelihood2> -3.220026502247338 </background_likelihood2>
<stream_only_likelihood2>  -3.603610892548526  -225.946861528629880  -69.172319179365019  -3.469036352446314 </stream_only_likelihood2>
<search_likelihood2> -2.680819329206185 </search_likelihood2>
Using SSE4.1 path
Integral 0 time = 929.860979 s
Integral 1 time = 72.370386 s
Running likelihood with 31964 stars
Likelihood time = 0.809307 s
<background_integral3> 0.000040761402645 </background_integral3>
<stream_integral3>  131.392854802297900  0.000000000000000  5.855449497825772  36.142942293270110 </stream_integral3>
<background_likelihood3> -3.259187575862191 </background_likelihood3>
<stream_only_likelihood3>  -3.716972425906632  -227.654181625117020  -114.306934247414770  -3.267494168057085 </stream_only_likelihood3>
<search_likelihood3> -2.688803829698713 </search_likelihood3>
04:42:33 (13296): called boinc_finish(0)

</stderr_txt>
]]>

I was going to say two things appeared suspicious to me, but after I compared it to some of my valid results, the same parts of the output are in the valid tasks. Still:

Error loading Lua script 'astronomy_parameters.txt': [string "number_parameters: 4..."]:1: '<name>' expected near '4' 

I'm curious why this error is in stderr. Additionally, I thought perhaps the non-finite result could be a problem, but that doesn't appear in every output, valid or invalid.

Here are my current invalidated results with 84 and 85 being significant high:

de_modfit_81 2
de_modfit_82 1
de_modfit_84 97
de_modfit_85 77
de_modfit_86 1

Yeah, same here with the really high Invalids.

Evidently something is wrong on the project side of things. Hopefully an Admin gets in here and sees what's going on and it gets fixed soon as this is not good for us or the project.

Hi everyone,

Invalidated tasks is a known issue with Separation right now. I will be taking down the 2 most problematic runs (de_modfit_84_bundle4_4s_south4s_1 and de_modfit_85_bundle4_4s_south4s_1), as these have converged and are optimized. Stripes 84 and 85 have a fairly large section where a chunk of data has been purposefully omitted, and I believe this may in part lead to the problems that people have been having.

I think that the problems stem from computer precision differences between different platforms and versions of the client. As runs become more and more optimized, we've noticed that these errors show up more and more.

- Tom

Thanks Tom! :-)

Seems like some 84s and 85s are STILL being sent. Aborting those tasks until problem is permanently solved.

I know. Tom posted on 4 September he was taking down the 84 and 85 Stripes. Yet I continued to have them sent on 6 September.
https://milkyway.cs.rpi.edu/milkyway/forum_thread.php?id=4487&postid=69005

---

Stripes 84 and 85 were taken down. These are validations of older tasks, not new tasks. Workunits generally take up to a week or two to finally stop trickling in validations.

Tom

Good to know.

Good to know.

Yes, good to know I will have to continue scrubbing them from the cache when received.

---

Me too.