Delphes on aMC@NLO

[harikrishnan ramani]

Hi,
I have gone thru earlier tickets about the error
ERROR: Unsupported block type.-----------------------------------] when trying to shower stdhep events created by amc_nlo +herwig6/pythia6
and was of the understanding that its solved. However i have delphes 3.1.2 and still seem to face the same issue. Could you please tell me how to solve this?

[harikrishnan ramani]

this problem persists even for small (100 event) files
Replying to hramani:

Hi,
I have gone thru earlier tickets about the error
ERROR: Unsupported block type.-----------------------------------] when trying to shower stdhep events created by amc_nlo +herwig6/pythia6
and was of the understanding that its solved. However i have delphes 3.1.2 and still seem to face the same issue. Could you please tell me how to solve this?

[Michele Selvaggi]

Hi,

in principle this problem should have been solved indeed. Which version of MadGraph_aMC@NLO are you using?
Can you please try again with the most recent one?

Michele

[harikrishnan ramani]

herwig stdhep file

[harikrishnan ramani]

I tried both 2.1.1 which is the most recent stable version as well as 2.1.2(beta) with the same issues!
I have attached the stdhep file produced by herwig6.

[Suyong Choi]

I have exactly the same problem. I'm trying it out Delphes 3.1.2 and Madgraph_aMC@NLO 2.1.2(beta) on 64-bit Fedora Core 20. Same problem with Madgraph_aMC@NLO 2.1.1. After running a debugger to see where the problem is, there is a case where fBlockType in DelphesSTDHEPReader::ReadBlock() is 0, but this case is not caught by if statements, hence error is thrown.

[Pavel Demin]

Hi,

I'll try to look at this problem. Only, I'll need a detailed list of commands needed to install Madgraph_aMC@NLO + Herwig and to produce a small .hep file (I've never tried to run Madgraph_aMC@NLO + Herwig).

Normally, I tested the MG5 (not NLO) + Pythia chain and it worked.

Regards,

Pavel

[harikrishnan ramani]

Hi,
I think it worked with MG5(tree level) +Pythia6 because tree level processes do not need negative weights.
Thankfully Madgraph_aMC@NLO + Herwig works out of the box. ​https://launchpad.net/mg5amcnlo you could download the latest stable version on the right. launch madgraph as usual.
To produce a simple process at NLO:
generate p p > Z [QCD]
output drellyan
launch drellyan
adding [QCD] to any process you normally produce in madgraph makes it NLO.
In the run card make sure that the chosen shower is Herwig6. this will produce a herwig.hep file which is actually a stdhep file that gives an error when read by Delphes.

Sincerely
Harikrishnan

[Pavel Demin]

Hi Harikrishnan,

Thank you for the instructions!

The problem is in the Delphes' STDHEP reader.

It can be solved by removing the following line from the 'DelphesSTDHEPReader::ReadSTDHEP4()' method in 'classes/DelphesSTDHEPReader.cc':

  xdr_string(fInputXDR, &fBuffer, 100);

I've committed a fixed version of the STDHEP reader to the Delphes repository
and this fixed version will be available in the next release.

You can download a snapshot of the Delphes v3 code repository from

​https://github.com/delphes/delphes/archive/master.zip

Regards,

Pavel