Discussion in particular about the definition of the SCALUP parameter in POWHEG-BOX, and how to interpret it for shower matching.

Appropriate treatment now agreed between POWHEG and Py8 authors: can start running a test version ~now? (Stefan Prestel, Carlo Oleari)

Two options: default scheme and SCALUP_max scheme in POWHEG ⇒ systematic in scheme / scale sensitivity.

Two views on whether it matters that Pythia by default uses an LO $\alpha_s$ (0.134 at $M_Z$, as opposed to ~0.118 for NLO matrix element). Generally agreed that setting the shower to use the same alpha_s as the matrix element is a solid base configuration

Connection to LHE format enhancement discussion: need LHE files to be able to pass all necessary PDF and $\alpha_s$ information for automatic consistency in shower setup.

Argument that $\mu_R = k p_T$ (in argument of $\alpha_s$ for both ME and PS) should naturally have $k \sim 0.4$ from NLL $A_2$ term ~equivalence to scale modification (cf. Frank Krauss). This or $k = 1$ both viable. Explore both.

Effect of $\times 2$ variations on $\mu_R$: include both schemes, so use $k = \{ 0.2, 0.4, 1.0, 2.0 \}$ settings for comparison.

Appropriate treatment for $\mu_F$? Typical shower interpretation is to use this as typical order of shower starting scale, but for matching the shower starts at the kinematic limit but emissions are vetoed until below the POWHEG scale. Covered completely by SCALUP_max systematic definition.

Scale variation in sync between ME and PS will overestimate uncertainties but is the obvious baseline definition before getting fancier.

Situation for NLO matching generators other than POWHEG? What freedoms available / what prescription to use?

Evaluate effect of different matching procedures for dijet events at NLO. Using Rivet ATLAS & CMS analyses of inclusive/dijet spectra, R32 and jet rates (and ratios), jet shapes and pile-up/UE-robust jet mass etc. Restrict to dijet process at NLO for now.

Include scale variations and scheme uncertainties separately.

Downstream impact on tuning: start with discrete sensitivity study rather than a full scale tuning.

• POWHEG-BOX + Pythia8 (2 schemes x 2 $\alpha_s(M_Z)$)
• POWHEG-BOX + Herwig++ (? schemes x 2 $\alpha_s(M_Z)$)
• POWHEG-BOX + PYTHIA6? (1 scheme x 2 $\alpha_s(M_Z)$)
• Sherpa MC@NLO
• BlackHat + Sherpa ?
• Njet + Sherpa ?
• (a)MC@NLO+Herwig++?
• Herwig++ MatchBox?
• Pythia8 UNLOPS?
• HEJ
• More?

Planning session blackboard:

— Klaus Rabbertz 2013/06/17 15:12

We have provided a step by step guide for generation of fixed order events with NJet+Sherpa via the NJet website Event Generation with NJet+Sherpa Get in touch with Valery(yundin@nbi.dk) or myself if it needs any clarification/doesn't work etc.

— Simon Badger 2013/06/11 17:50

Slides on POWHEG+Pythia8 More slides on POWHEG+Pythia and Herwig etc

• systematic variation of underlying event simulation: eigentunes? other ideas?
• PDF uncertainties: how to do them properly in the shower? is this important?