Opened 11 years ago
Closed 11 years ago
#545 closed task (fixed)
Revive MLM matching
| Reported by: | Juergen Reuter | Owned by: | ALL |
|---|---|---|---|
| Priority: | P0 | Milestone: | v2.2.0 |
| Component: | core | Version: | 2.1.1 |
| Severity: | blocker | Keywords: | |
| Cc: | mail@… |
Description
This should include the corresponding entries in the header of the LHEF format.
Change History (28)
comment:1 Changed 11 years ago by
| Priority: | P3 → P4 |
|---|
comment:2 Changed 11 years ago by
| Type: | defect → task |
|---|
comment:3 Changed 11 years ago by
For the MLM matching with PYTHIA, there where additional tags for the MLM matching settings to be written into the header of the LHE file. This should not be forgotten.
comment:4 Changed 11 years ago by
| Owner: | changed from kilian to Juergen Reuter |
|---|---|
| Status: | new → assigned |
There is no other stupid guy doing this !
comment:5 Changed 11 years ago by
| Priority: | P4 → P0 |
|---|---|
| Severity: | normal → major |
comment:6 Changed 11 years ago by
Uff, it seems that in 2.2.0 at some point one gets stuck ... :( Investigating
comment:7 Changed 11 years ago by
Uff, it seems that in 2.2.0 at some point one gets stuck ... :( Investigating
comment:8 Changed 11 years ago by
(Sorry for the double entry). On my laptop, using the EEMatching3P.sin SINDARIN file, and checkpoint = 25, the program stalls after generating 13,725 events. WTF!? Any idea, how I can check where the program is stuck?
comment:9 Changed 11 years ago by
It gets stuck in the middle of an event:
<event> 14 1 1.0000000000E+00 9.1000000000E+01 -1.0000000000E+00 1.1780000000E-01 11 -1 0 0 0 0 0.0000000000E+00 0.0000000000E+00 4.5499999997E+01 4.5500000000E+01 5.1099700000E-04 0.0000000000E+00 9.0000000000E+00 -11 -1 0 0 0 0 0.0000000000E+00 0.0000000000E+00 -4.5499999997E+01 4.5500000000E+01 5.1099700000E-04 0.0000000000E+00 9.0000000000E+00 -2 1 1 2 0 501 -5.8227482585E+00 3.8356974321E+00 -7.5780912457E+00 1.0297788064E+01 0.0000000000E+00 0.0000000000E+00 9.0000000000E+00 21 1 1 2 501 502 -8.4359925301E+00 8.3248265979E+00 -1.0714087351E+01 1.5976870020E+01 0.0000000000E+00 0.0000000000E+00 9.0000000000E+00 21 1 1 2 502 503 -4.8752057726E+00 3.5224408401E+00 -6.0769401400E+00 8.5501124123E+00 0.0000000000E+00 0.0000000000E+00 9.0000000000E+00
comment:10 Changed 11 years ago by
Next info: on the DESY cluster it stucks at _exactly_ the same position! gdb tells me that it breaks in function unlink (). This should delete a file!? (is it also for scratch files).
comment:11 Changed 11 years ago by
Next info: it was not the putting of the code inside the Noweb file and the polishing. I checked with the last commit before, it also stucks after 13,700 events.
comment:12 Changed 11 years ago by
I checked how often the subroutine apply_shower_particle_set is called, depending on the seed this happens a few thousand times, then everything is reset, apply_shower_particle_set is called infintely many times, but no events are written out.
comment:13 Changed 11 years ago by
| Cc: | mail@… added |
|---|
So what happens is that at a certain point, all events are vetoed by the mlm_matching subroutine for the reason: "not enough PS jets". So it seems that Pythia generates too less PS jets that survive the yt cut? Hm, SS, any ideas?
comment:14 follow-up: 16 Changed 11 years ago by
EEMatching4P.sin is stuck in the same way as EEMatching3P.sin is, but EEMatching5P.sin runs through (with the seed denoted). Big question now: did this ever happen in 2.1.1. as well and just got unnoticed? What is the origin of the infinite veto cascade!?
comment:15 Changed 11 years ago by
One more info: using EEMatching3P.sin with seed = 1, after call 13999, the 2->2 process is _never_ called again, and this is the very moment at which the 2->3 process starts having _all_ events rejected. WTF!?
comment:16 Changed 11 years ago by
Replying to jr_reuter:
EEMatching4P.sin is stuck in the same way as EEMatching3P.sin is, but EEMatching5P.sin runs through (with the seed denoted). Big question now: did this ever happen in 2.1.1. as well and just got unnoticed? What is the origin of the infinite veto cascade!?
Well, JRR, EEMatching5P.sin gets stuck the same way. I forgot to switch on the parton shower, so nothing at all happened. That is the reason why this one ran through. So, apparently, _all_ cases are failing, also for all kinds of seeds!
comment:17 Changed 11 years ago by
I checked the same for 2.1.1.: there you have events for the 2->2 process until the very end.
comment:18 Changed 11 years ago by
After having thought about it, I guess the logic is the other way round: because there is no valid event found any longer the 2->3 process, the 2->2 is never called again.... I'm in need of help, utterly!
comment:19 Changed 11 years ago by
I still cannot formulate my suspicion, but maybe it has to do with the fact how vetoed events are treated. In 2.1, the only way an event could get vetoed was by the parton shower (namely by the MLM and dummy CKKW matching routines). This was transferred into the event%is_vetoed flag. In the simulations module there was a condition in simulation_generate_event which is only for the case that a luminosity value was given exceeding n_events, and the whole thing was wrapped into a MAX_TRIES_FOR_SINGLE_EVENT loop (still not being reimplemented). The generate loop was only exited if the event was valid and not vetoed. Now, the event_transform returns with a probability that is either 1 if the event is valid and not vetoed and 0 otherwise. If the probability is smaller than a random number x the event is rejected. So either it is always accepted or always rejected, same situation basically as in WHIZARD 2.1. Seems that this doesn't bite nor hurt. HELP!!!
comment:20 Changed 11 years ago by
| Owner: | changed from Juergen Reuter to ALL |
|---|---|
| Severity: | major → blocker |
| Status: | assigned → new |
GIVING UP ... SOMEONE HAS TO TAKE OVER.... :(((
comment:21 Changed 11 years ago by
The matching with the Whizard shower runs through for EEMatching[345]W.sin and produces the desired, showered LHE files. Only EEMatching3W.sin gives an Error
after approx 74000 events:
Interactions:
Interaction number 1
Shower parton:
1 [ 11] 0.000E+00 0.000E+00 4.550E+01 4.55000E+01
2.61118E-07 2.61118E-07
0.00000E+00 ******** I T 1 CI 0 0
Shower parton:
2 [ -11] 0.000E+00 0.000E+00 -4.550E+01 4.55000E+01
2.61118E-07 2.61118E-07
0.00000E+00 ******** I T 1 CI 0 0
Shower parton:
3 [ 2] 7 ********* 8.712E+00 -2.969E+01 3.44889E+01
1.94451E+02 1.94451E+02
0.00000E+00 ******** ******** ******** ******** ******** bF s C:
12 13T 1 CI 1 0
Shower parton:
4 [ -2] 9 1.906E+01 3.588E+00 2.919E+01 3.75020E+01
1.78523E+02 1.78523E+02
0.00000E+00 ******** ******** ******** ******** ******** bF s C:
10 11T 1 CI 0 2
Shower parton:
5 [ 21] 7 ********* ********* 5.091E-01 1.90091E+01
4.22620E+01 4.22620E+01
0.00000E+00 ******** ******** ******** ******** ******** bF s C:
14 15T 1 CI 2 1
Partons:
Shower parton:
1 [ 11] 0.000E+00 0.000E+00 4.550E+01 4.55000E+01
2.61118E-07 2.61118E-07
0.00000E+00 ******** I T 1 CI 0 0
Shower parton:
2 [ -11] 0.000E+00 0.000E+00 -4.550E+01 4.55000E+01
2.61118E-07 2.61118E-07
0.00000E+00 ******** I T 1 CI 0 0
Shower parton:
3 [ 2] 7 ********* 8.712E+00 -2.969E+01 3.44889E+01
1.94451E+02 1.94451E+02
0.00000E+00 ******** ******** ******** ******** ******** bF s C:
12 13T 1 CI 1 0
Shower parton:
4 [ -2] 9 1.906E+01 3.588E+00 2.919E+01 3.75020E+01
1.78523E+02 1.78523E+02
0.00000E+00 ******** ******** ******** ******** ******** bF s C:
10 11T 1 CI 0 2
Shower parton:
5 [ 21] 7 ********* ********* 5.091E-01 1.90091E+01
4.22620E+01 4.22620E+01
0.00000E+00 ******** ******** ******** ******** ******** bF s C:
14 15T 1 CI 2 1
Shower parton:
7 [ 94] 9 ********* ********* -2.919E+01 5.34980E+01
1.63416E+03 1.63416E+03
0.00000E+00 ******** ******** ******** ******** ******** bF s
C: 3 5T 1 CI 0 0
Shower parton:
9 [ 94] 0.000E+00 ********* 3.553E-15 9.10000E+01
8.28100E+03 8.28100E+03
0.00000E+00 ******** ******** ******** ******** ******** bF
C: 7 4T 1 CI 0 0
-------------------------------------------------------
Shower parton:
10 [ -2] 4 1.886E+01 5.828E+00 2.699E+01 3.41618E+01
4.91923E+01 4.91923E+01
0.00000E+00 ******** ******** ******** ******** ******** bF s C:
16 17F 1 CI 0 6
Shower parton:
11 [ 21] 4 2.012E-01 ********* 2.199E+00 3.34023E+00
1.26406E+00 1.26406E+00
0.00000E+00 ******** ******** ******** ******** ******** bF s C:
18 19F 1 CI 6 2
Shower parton:
12 2 3 0.000E+00 0.000E+00 0.000E+00 1.34539E+01
1.81008E+02 2.00695E+01
0.00000E+00 Ff F 1 CI 4 0
Shower parton:
13 21 3 0.000E+00 0.000E+00 0.000E+00 2.10350E+01
4.42470E+02 4.07239E+01
0.00000E+00 FfsF 1 CI 1 4
Shower parton:
14 21 5 0.000E+00 0.000E+00 0.000E+00 2.83753E+00
8.05159E+00 4.22620E+01
0.00000E+00 Ff F 1 CI 2 5
Shower parton:
15 21 5 0.000E+00 0.000E+00 0.000E+00 1.61716E+01
2.61521E+02 4.22620E+01
0.00000E+00 Ff F 1 CI 5 1
Shower parton:
16 -2 10 0.000E+00 0.000E+00 0.000E+00 2.69181E+01
7.24586E+02 4.91923E+01
0.00000E+00 Ff F 1 CI 0 7
Shower parton:
17 21 10 0.000E+00 0.000E+00 0.000E+00 7.24363E+00
5.24701E+01 4.91923E+01
0.00000E+00 Ff F 1 CI 7 6
Shower parton:
18 21 11 0.000E+00 0.000E+00 0.000E+00 7.91149E-01
6.25916E-01 1.26406E+00
0.00000E+00 Ff F 1 CI 6 8
Shower parton:
19 21 11 0.000E+00 0.000E+00 0.000E+00 2.54908E+00
6.49782E+00 1.26406E+00
0.00000E+00 Ff F 1 CI 8 2
Total Momentum: 0.000E+000.000E+000.000E+00 9.100E+01
ISR finished: T
FSR finished: F
******************************************************************************
*** ERROR: BUG: too many loops in simulate_children_ana (?)
******************************************************************************
| (WHIZARD run continues)
Any Ideas? Could this be a similar configuration as the one Pythia stucks at? In the meantime I will make the error message a bit less arcane..
comment:22 Changed 11 years ago by
As promised the polished version of that same event which causes simulate_children_ana to return after 1000 tries. What puzzles me is that partons 12-19 have exactly zero three momenta. Looks almost as if it hasn't been set? I will try to compile with -finit-real=nan...
Interactions:
Interaction number 1
Shower parton <nr> <type> <parent> <mom(0:4)>
1 11 4.55000E+01 0.00000E+00 0.00000E+00 4.55000E+01
<p4square> <t> <scale> <c1> <c2>
2.61118E-07 2.61118E-07 0.00000E+00 0 0
Parton belongs to 1
Shower parton <nr> <type> <parent> <mom(0:4)>
2 -11 4.55000E+01 0.00000E+00 0.00000E+00 -4.55000E+01
<p4square> <t> <scale> <c1> <c2>
2.61118E-07 2.61118E-07 0.00000E+00 0 0
Parton belongs to 1
Shower parton <nr> <type> <parent> <mom(0:4)>
3 2 7 3.44889E+01 -6.11496E+00 8.71190E+00 -2.96941E+01
<p4square> <t> <scale> <c1> <c2>
1.94451E+02 1.94451E+02 0.00000E+00 1 0
Parton is FSR, is simulated, has children: 12 13, belongs to 1
Shower parton <nr> <type> <parent> <mom(0:4)>
4 -2 9 3.75020E+01 1.90588E+01 3.58781E+00 2.91850E+01
<p4square> <t> <scale> <c1> <c2>
1.78523E+02 1.78523E+02 0.00000E+00 0 2
Parton is FSR, is simulated, has children: 10 11, belongs to 1
Shower parton <nr> <type> <parent> <mom(0:4)>
5 21 7 1.90091E+01 -1.29438E+01 -1.22997E+01 5.09096E-01
<p4square> <t> <scale> <c1> <c2>
4.22620E+01 4.22620E+01 0.00000E+00 2 1
Parton is FSR, is simulated, has children: 14 15, belongs to 1
Partons:
Shower parton <nr> <type> <parent> <mom(0:4)>
1 11 4.55000E+01 0.00000E+00 0.00000E+00 4.55000E+01
<p4square> <t> <scale> <c1> <c2>
2.61118E-07 2.61118E-07 0.00000E+00 0 0
Parton belongs to 1
Shower parton <nr> <type> <parent> <mom(0:4)>
2 -11 4.55000E+01 0.00000E+00 0.00000E+00 -4.55000E+01
<p4square> <t> <scale> <c1> <c2>
2.61118E-07 2.61118E-07 0.00000E+00 0 0
Parton belongs to 1
Shower parton <nr> <type> <parent> <mom(0:4)>
3 2 7 3.44889E+01 -6.11496E+00 8.71190E+00 -2.96941E+01
<p4square> <t> <scale> <c1> <c2>
1.94451E+02 1.94451E+02 0.00000E+00 1 0
Parton is FSR, is simulated, has children: 12 13, belongs to 1
Shower parton <nr> <type> <parent> <mom(0:4)>
4 -2 9 3.75020E+01 1.90588E+01 3.58781E+00 2.91850E+01
<p4square> <t> <scale> <c1> <c2>
1.78523E+02 1.78523E+02 0.00000E+00 0 2
Parton is FSR, is simulated, has children: 10 11, belongs to 1
Shower parton <nr> <type> <parent> <mom(0:4)>
5 21 7 1.90091E+01 -1.29438E+01 -1.22997E+01 5.09096E-01
<p4square> <t> <scale> <c1> <c2>
4.22620E+01 4.22620E+01 0.00000E+00 2 1
Parton is FSR, is simulated, has children: 14 15, belongs to 1
Shower parton <nr> <type> <parent> <mom(0:4)>
7 94 9 5.34980E+01 -1.90588E+01 -3.58781E+00 -2.91850E+01
<p4square> <t> <scale> <c1> <c2>
1.63416E+03 1.63416E+03 0.00000E+00 0 0
Parton is FSR, is simulated, has children: 3 5, belongs to 1
Shower parton <nr> <type> <parent> <mom(0:4)>
9 94 9.10000E+01 0.00000E+00 -8.88178E-16 3.55271E-15
<p4square> <t> <scale> <c1> <c2>
8.28100E+03 8.28100E+03 0.00000E+00 0 0
Parton is FSR, has children: 7 4, belongs to 1
-------------------------------------------------------
Shower parton <nr> <type> <parent> <mom(0:4)>
10 -2 4 3.41618E+01 1.88576E+01 5.82793E+00 2.69863E+01
<p4square> <t> <scale> <c1> <c2>
4.91923E+01 4.91923E+01 0.00000E+00 0 6
Parton is FSR, is simulated, has children: 16 17, doesnt belong to 1
Shower parton <nr> <type> <parent> <mom(0:4)>
11 21 4 3.34023E+00 2.01219E-01 -2.24012E+00 2.19873E+00
<p4square> <t> <scale> <c1> <c2>
1.26406E+00 1.26406E+00 0.00000E+00 6 2
Parton is FSR, is simulated, has children: 18 19, doesnt belong to 1
Shower parton <nr> <type> <parent> <mom(0:4)>
12 2 3 1.34539E+01 0.00000E+00 0.00000E+00 0.00000E+00
<p4square> <t> <scale> <c1> <c2>
1.81008E+02 2.00695E+01 0.00000E+00 4 0
Parton is FSR, is final, doesnt belong to 1
Shower parton <nr> <type> <parent> <mom(0:4)>
13 21 3 2.10350E+01 0.00000E+00 0.00000E+00 0.00000E+00
<p4square> <t> <scale> <c1> <c2>
4.42470E+02 4.07239E+01 0.00000E+00 1 4
Parton is FSR, is final, is simulated, doesnt belong to 1
Shower parton <nr> <type> <parent> <mom(0:4)>
14 21 5 2.83753E+00 0.00000E+00 0.00000E+00 0.00000E+00
<p4square> <t> <scale> <c1> <c2>
8.05159E+00 4.22620E+01 0.00000E+00 2 5
Parton is FSR, is final, doesnt belong to 1
Shower parton <nr> <type> <parent> <mom(0:4)>
15 21 5 1.61716E+01 0.00000E+00 0.00000E+00 0.00000E+00
<p4square> <t> <scale> <c1> <c2>
2.61521E+02 4.22620E+01 0.00000E+00 5 1
Parton is FSR, is final, doesnt belong to 1
Shower parton <nr> <type> <parent> <mom(0:4)>
16 -2 10 2.69181E+01 0.00000E+00 0.00000E+00 0.00000E+00
<p4square> <t> <scale> <c1> <c2>
7.24586E+02 4.91923E+01 0.00000E+00 0 7
Parton is FSR, is final, doesnt belong to 1
Shower parton <nr> <type> <parent> <mom(0:4)>
17 21 10 7.24363E+00 0.00000E+00 0.00000E+00 0.00000E+00
<p4square> <t> <scale> <c1> <c2>
5.24701E+01 4.91923E+01 0.00000E+00 7 6
Parton is FSR, is final, doesnt belong to 1
Shower parton <nr> <type> <parent> <mom(0:4)>
18 21 11 7.91149E-01 0.00000E+00 0.00000E+00 0.00000E+00
<p4square> <t> <scale> <c1> <c2>
6.25916E-01 1.26406E+00 0.00000E+00 6 8
Parton is FSR, is final, doesnt belong to 1
Shower parton <nr> <type> <parent> <mom(0:4)>
19 21 11 2.54908E+00 0.00000E+00 0.00000E+00 0.00000E+00
<p4square> <t> <scale> <c1> <c2>
6.49782E+00 1.26406E+00 0.00000E+00 8 2
Parton is FSR, is final, doesnt belong to 1
Total Momentum [0:4]: 9.10000E+01 0.00000E+00 0.00000E+00 0.00000E+00
ISR finished: T
FSR finished: F
******************************************************************************
*** ERROR: BUG: too many loops in simulate_children_ana (?)
******************************************************************************
comment:23 Changed 11 years ago by
Program received signal SIGFPE: Floating-point exception - erroneous arithmetic operation incoming at exactely this event.. backtracing the exception now :/
comment:24 follow-up: 25 Changed 11 years ago by
The "too many loops in simulate_children_ana"-issue has nothing to do with MLM matching. It occurs if in the WHIZARD PS for a splitting the splitting of the childrens cannot be found, probably due to momentum conservation. It shouldn't occur and in my tests never did...
I can look into the MLM-problems but it'll be at least a weak until I have enough time.
comment:25 Changed 11 years ago by
Replying to sschmidt:
The "too many loops in simulate_children_ana"-issue has nothing to do with MLM matching. It occurs if in the WHIZARD PS for a splitting the splitting of the childrens cannot be found, probably due to momentum conservation. It shouldn't occur and in my tests never did...
I can look into the MLM-problems but it'll be at least a weak until I have enough time.
Indeed. Maybe it just didn't show up earlier due to lucky seeds. It happens in 2 out of 10 seeds. I have increased the numerical safety net you have for cstep and run it now with 50 seeds to make sure that the Whizard shower is completely fine now.
comment:26 Changed 11 years ago by
Event_transforms should have an option to veto. (WK promised that)
JRR and BCN will rewrite the SINDARIN files.
comment:27 Changed 11 years ago by
As of r5623, all Pythia examples for e+ e- are working, the WHIZARD shower examples in principle, too, but there is the infinite loop in shower_simulate_children_ana that occurs in 5-6% of the cases. Still not understood. Drell-Yan examples under investigation.
comment:28 Changed 11 years ago by
| Resolution: | → fixed |
|---|---|
| Status: | new → closed |
We are roughly back to the status of 2.1. Improvements are the setup for writing and reading scratch LHE event for transferral to and from Pythia. Some hiccups of the form
`Failed to read LHEF event information. Will assume end of file has been reached.` are not happening now because of this improvement. All matching are running and only very rarely run into the problems now split off into #632. Also now, there are some poor man's tests for FSR and ISR matching (formerly fsr_shower and isr_shower). Closing this one.
Requires reactivation of the Parton Shower (#44)