alt_setup weights in event records

Asked by Ulrike Schnoor on 2019-09-18

Is there a way to write the alternative event weights obtained by alt_setup also to the lcio output? Is it already written to stdhep and if so, how can it be retrieved? So far I have only found them in ascii (.evt).
Thanks!

Question information

Language:
English Edit question
Status:
Solved
For:
WHIZARD Edit question
Assignee:
Juergen Reuter Edit question
Last query:
2019-09-20
Last reply:
2019-10-30
Wolfgang Kilian (whkilian) said : #1

We just discussed this in the Whizard team via phone.

The alternate weights are read and write in the internal format (.evx), of course. The output in the ASCII format is available as the simplest option to make it visible.

It should be straightforward to supply extra weight values to LCIO (or HepMC) formats, since those have generic containers. IIRC, it has not been implemented yet for LCIO. We may cross-check with the LCIO API whether there is a preferred way to do this.

I don't think that event formats such as STDHEP which have fixed records, can support such extra information.

Juergen Reuter (j.r.reuter) said : #2

In addition, it would be good to have a SINDARIN to see what you wanted to do (the process could be simplified as much as possible for this). In LCIO, per default only one float can be set as weight. It is trivial, to add additional weights but I'd rather do this adjoint with the LC software group (Frank G. et al.) in order to have a well-defined interface. The Tokyo meeting could serve as a platform for this. The rescanning is vetoed at the moment, and simple enabling does not work because of some bug (one of which is easy to fix). Please let us know what you exactly want.

Ulrike Schnoor (ulrike-sch) said : #3

Hi Jürgen and Wolfgang,
thanks for your replies. What I want to do is write out 9 weights needed in order to reweight for the 3 TGCs ka, la, g1z. You can find the sindarin with the rescanning here: https://cernbox.cern.ch/index.php/s/bMochnhSPzLwMj7
Now, I have a workaround where I store the stdhep and the evt and then I get the weights from the evt and apply them later to the fully simulated file. However, it would be nicer if the weights could be transfered to the full simulation file - which would be easiest if it is in lcio. Let's see if there is some general interest.
Cheers,
Ulrike

Juergen Reuter (j.r.reuter) said : #4

Hi Ulrike,
first of all, great that you have a workaround (by the way, your email just came when I started to look again into your other problem with the missing displaced vertices in the showered and hadronized event files). So we don't have hard pressure to find a solution, I will play around with this a little, and your task would be to prepare a discussion item on weight formats in LCIO for the Tokyo meeting (given that you accept your membership).
Cheers,
    JRR

Ulrike Schnoor (ulrike-sch) said : #5

Hi Jürgen,
thank you, I will do that.
Cheers,
Ulrike

Juergen Reuter (j.r.reuter) said : #6

I committed a fix. First of all, LCIO can now also be used for rescans, and alternate cross sections and weights are allowed. They are individually labelled/named by their index.

Ulrike Schnoor (ulrike-sch) said : #7

Thanks a lot, Jürgen! I will test it.

Juergen Reuter (j.r.reuter) said : #8

2.8.2 has been released which contains this feature.

Juergen Reuter (j.r.reuter) said : #9

2.8.2 has been released.