Opened 11 years ago
Closed 11 years ago
#185 closed Bug (fixed)
lhco format output seems to be invalid in madgraph framework
| Reported by: | mattelaer | Owned by: | |
|---|---|---|---|
| Priority: | minor | Milestone: | |
| Component: | Delphes code | Version: | Delphes 3 |
| Keywords: | Cc: |
Description
Hi guys,
Maikel report on the MG5 track a problem with the plotting of events created by Delphes:
https://answers.launchpad.net/madgraph5/+question/230244
I've generated p p > e+ e- j
and reproduced the problem. By looking at the log of MA4, you will see:
----------------------------------------------
events information
----------------------------------------------
Events in file : 0
Integrated weight (pb) : 294.89999999999998
Max wgt : -1.0000000000000000
Average wgt : NaN
Passing the cuts (plotted)
--------------------------
Events : 0
Integrated weight (pb) : 0.0000000000000000
Max wgt : -1.0000000000000000
Average wgt : Infinity
So I think that some part of the lhco information are not pass anymore correctly to the lhco output. Could you look at that?
Cheers,
Olivier
PS: I always have problem with Delphes email notification, so if you ask me a question and that I don't react, just send me a direct email. (I hope that adding madgraph in the title of the email, will trigger the correct filter of my email but we will see)
Change History (8)
comment:1 by , 11 years ago
comment:2 by , 11 years ago
Hi Pavel,
Your lhco file is just empty:
(end of the file:)
## In case FROG event display allowed #NEvents_Frog 100 ## Number of events to process #NEvents -1 // -1 means 'all' # ## input PDG tables #PdgTableFilename DELPHESDIR/data/particle.tbl // table with particle pid,mass,charge,... #</mgdelphescard> #</header> #</LesHouchesEvents> ## Integrated weight (pb) : 294.975
You can see the end of the header. and then no event!
Cheers,
Olivier
comment:3 by , 11 years ago
Just to point that one people more complains about this:
https://answers.launchpad.net/madgraph5/+question/231928
(And this is other question raised on this point)
As a reminder, the problem is that Delphes didn't write any output events.
Even for the Standard Model (see the example above)
How is this possible?
What should I do? Advice them to use Delphes2? PGS? Did you fix this and forget to report it?
Cheers,
Olivier
comment:4 by , 11 years ago
Hi Olivier,
I'm afraid that we don't have enough information about this problem.
Could you, please send a list of commands to reproduce this problem?
Is it possible that the problem is in the MG5 scripts?
I don't see any problem with the root2lhco converter.
I run it as
./root2lhco test.root test.lhco
Thanks,
Pavel
comment:5 by , 11 years ago
I've tried to run MG5 to reproduce the problem.
Here are the commands:
wget https://launchpad.net/madgraph5/trunk/1.5.0/+download/MadGraph5_v1.5.11.tar.gz tar -zxf MadGraph5_v1.5.11.tar.gz cd MadGraph5_v1_5_11/ ./bin/mg5 install Delphes install pythia-pgs install MadAnalysis generate p p > z > e+ e- launch
then MG5 indeed hangs at
Creating Plots for Delphes level
Using 'ps' I've found what executable is actually running at this point and tried to run in manually:
/bin/bash /scratch/MG5/MadGraph5_v1_5_11/PROC_sm_0/bin/internal/plot /scratch/MG5/MadGraph5_v1_5_11/MadAnalysis /scratch/MG5/MadGraph5_v1_5_11/td
it hangs after showing the following message:
plot_events found topdrawer found epstosmth found plot_card found cp: `../../../Cards/plot_card.dat' and `./ma_card.dat' are the same file end plot_events PAUSE:
Looks like this plot/MadAnalysis/td combination stops at some point and waits for some user input.
PROC_sm_0/Events/run_01/tag_1_delphes_events.lhco file looks OK:
## Integrated weight (pb) : 717.23 0 1 0 1 0 -2.389 2.441 30.81 0.00 0.0 0.0 0.00 0.0 0.0 2 1 -0.925 -1.023 40.45 0.00 1.0 0.0 0.00 0.0 0.0 3 6 0.000 -0.480 1.06 0.00 0.0 0.0 0.00 0.0 0.0 0 2 0 1 1 -0.972 1.831 48.56 0.00 1.0 0.0 0.00 0.0 0.0 2 1 -1.200 -2.011 47.22 0.00 -1.0 0.0 0.00 0.0 0.0 3 4 0.887 -0.193 28.12 6.96 3.0 0.0 0.64 0.0 0.0 4 6 0.000 -1.570 3.25 0.00 0.0 0.0 0.00 0.0 0.0 0 3 0 1 1 -0.190 -2.362 30.79 0.00 -1.0 0.0 0.00 0.0 0.0 2 1 1.813 0.767 28.00 0.00 1.0 0.0 0.00 0.0 0.0 3 6 0.000 1.438 2.63 0.00 0.0 0.0 0.00 0.0 0.0
I would say that it's up to the MG5 developers to tell us what is the problem.
comment:6 by , 11 years ago
Finally, it look like a bug (or even 2 bugs) in MadAnalysis.
Here is my test LHCO file:
## Integrated weight (pb) : 717.23 # typ eta phi pt jmas ntrk btag had/em dum1 dum2 0 1 0 1 0 -2.389 2.441 30.81 0.00 0.0 0.0 0.00 0.0 0.0 2 1 -0.925 -1.023 40.45 0.00 1.0 0.0 0.00 0.0 0.0 3 6 0.000 -0.480 1.06 0.00 0.0 0.0 0.00 0.0 0.0 0 2 0 1 1 -0.972 1.831 48.56 0.00 1.0 0.0 0.00 0.0 0.0 2 1 -1.200 -2.011 47.22 0.00 -1.0 0.0 0.00 0.0 0.0 3 4 0.887 -0.193 28.12 6.96 3.0 0.0 0.64 0.0 0.0 4 6 0.000 -1.570 3.25 0.00 0.0 0.0 0.00 0.0 0.0 0 3 0 1 1 -0.190 -2.362 30.79 0.00 -1.0 0.0 0.00 0.0 0.0 2 1 1.813 0.767 28.00 0.00 1.0 0.0 0.00 0.0 0.0 3 6 0.000 1.438 2.63 0.00 0.0 0.0 0.00 0.0 0.0 0 4 0 1 1 -1.185 2.883 44.47 0.00 -1.0 0.0 0.00 0.0 0.0 2 1 -0.422 -0.603 40.57 0.00 1.0 0.0 0.00 0.0 0.0 3 6 0.000 1.370 4.97 0.00 0.0 0.0 0.00 0.0 0.0
It contains 4 events and 2 comment lines.
If I run MadAnalysis/plot_events with this file, it seems to work but only finds 3 events out of 4:
input name of event file:
-------------------------
test.lhco
Found LHC Olympics (PGS4) Event File
output file: plots.top
Classes will be read from file: ./ma_card.dat
Found the following classes in input classes file:
class # 0 is mET 12 -12 14 -14 16 -16 122
class # 1 is jet 1 -1 2 -2 3 -3 4 -4 21
class # 2 is b 5 -5
Now start parsing events a first time
created new class a for particle 22
created new class e+ for particle -11
created new class e- for particle 11
Setting up graphs
28 plots set up
Setting up acceptance cuts
Cuts set up
SEE plots.log FOR MORE INFO
Now parsing events to fill histograms
----------------------------------------------
events information
----------------------------------------------
Events in file : 3
Integrated weight (pb) : 717.230000000000
Max wgt : 1.00000000000000
Average wgt : 1.00000000000000
Passing the cuts (plotted)
--------------------------
Events : 3
Integrated weight (pb) : 717.230000000000
Max wgt : 1.00000000000000
Average wgt : 1.00000000000000
Now I just remove the second comment line:
## Integrated weight (pb) : 717.23 0 1 0 1 0 -2.389 2.441 30.81 0.00 0.0 0.0 0.00 0.0 0.0 2 1 -0.925 -1.023 40.45 0.00 1.0 0.0 0.00 0.0 0.0 3 6 0.000 -0.480 1.06 0.00 0.0 0.0 0.00 0.0 0.0 0 2 0 1 1 -0.972 1.831 48.56 0.00 1.0 0.0 0.00 0.0 0.0 2 1 -1.200 -2.011 47.22 0.00 -1.0 0.0 0.00 0.0 0.0 3 4 0.887 -0.193 28.12 6.96 3.0 0.0 0.64 0.0 0.0 4 6 0.000 -1.570 3.25 0.00 0.0 0.0 0.00 0.0 0.0 0 3 0 1 1 -0.190 -2.362 30.79 0.00 -1.0 0.0 0.00 0.0 0.0 2 1 1.813 0.767 28.00 0.00 1.0 0.0 0.00 0.0 0.0 3 6 0.000 1.438 2.63 0.00 0.0 0.0 0.00 0.0 0.0 0 4 0 1 1 -1.185 2.883 44.47 0.00 -1.0 0.0 0.00 0.0 0.0 2 1 -0.422 -0.603 40.57 0.00 1.0 0.0 0.00 0.0 0.0 3 6 0.000 1.370 4.97 0.00 0.0 0.0 0.00 0.0 0.0
If I run MadAnalysis/plot_events with this new file, it does not find any event:
input name of event file:
-------------------------
test.lhco
Found LHC Olympics (PGS4) Event File
output file: plots.top
Classes will be read from file: ./ma_card.dat
Found the following classes in input classes file:
class # 0 is mET 12 -12 14 -14 16 -16 122
class # 1 is jet 1 -1 2 -2 3 -3 4 -4 21
class # 2 is b 5 -5
Now start parsing events a first time
Setting up graphs
3 plots set up
Setting up acceptance cuts
Cuts set up
SEE plots.log FOR MORE INFO
Now parsing events to fill histograms
----------------------------------------------
events information
----------------------------------------------
Events in file : 0
Integrated weight (pb) : 717.230000000000
Max wgt : -1.00000000000000
Average wgt : NaN
Passing the cuts (plotted)
--------------------------
Events : 0
Integrated weight (pb) : 0.00000000000000
Max wgt : -1.00000000000000
Average wgt : NaN
comment:7 by , 11 years ago
I've finally found the source of the problem. It's a MadAnalysis bug.
The bug is at line 2907 in MadAnalysis/plot_events.f
do while(index(buff," # typ ") .eq. 0)
MadAnalysis requires " # typ " in the beginning of the file with at least 2 spaces before the '#' sign and exactly 2 spaces between '#' and 'typ'.
See MG5 bug #1198655 for more details:
comment:8 by , 11 years ago
| Resolution: | → fixed |
|---|---|
| Status: | new → closed |
Workaround for this MadAnalysis bug is in trunk now and will be available in Delphes 3.0.10.
Hi Olivier,
I'm afraid that we don't have enough information to fix this problem.
We need to know what exactly makes MadAnalysis4 unhappy.
Could you attach an example of .lhco file?
Cheers,
Pavel