Dan Clements – Jan 2006 Glasgow ATLAS physics meeting

1. Progress With Jet Studies Dan Clements Dan Clements – Jan 2006 Glasgow ATLAS physics meeting

2. Summary Errors on Inclusive Jet Cross-Section (study with NLOJET) • Disagreement between JETRAD and NLOJET has been resolved. • Work currently being written up for ATLAS note. Pseudo-Data Generation For ATLAS • Data sets for the Inclusive Jet Cross-Section have been created and smeared for statistics and Jet Energy Scale, available at: http://ppewww.ph.gla.ac.uk/~clements/pseudo/pseudoindex.html Grids For NLO Jet Cross-Section Calculations • Code currently under development and testing. http://ppewww.ph.gla.ac.uk/~clements/grids/gridindex.html Dan Clements – Jan 2006 Glasgow ATLAS physics meeting

3. Grids For NLO Jet Cross-Section Calculations (1) Background: At a hadron-collider the cross-section for a hard 2-hadron scattering process is given by: Parton momenta PDFs Sum over parton types Parton Level Cross-Section, (the bit theorists worry about) Hadron momenta Dan Clements – Jan 2006 Glasgow ATLAS physics meeting

4. Grids For NLO Jet Cross-Section Calculations (2) Problem: • Generating NLO Cross-Sections takes a long time (~days). • For PDF fitting need to create lots of cross-section predictions (for different PDFs) to compare with experimental data. Solution: • Attempt to split the cross-section calculation into two pieces, so that the parton-level cross-section is calculated independently of the PDFs. Create a grid of weights in (x1,x2,Q2) Dan Clements – Jan 2006 Glasgow ATLAS physics meeting

5. Grids For NLO Jet Cross-Section Calculations (3) Event with weight wi, x1, x2, Q2 NLOJET event generator Fill Grid with weight wi, at point (x1,x2,Q2) Grid of weights in (x1,x2,Q2) Jet Cross-Section PDFs defined at (x1,x2,Q2) Multiply and add over (x1,x2,Q2) Step 1: Fill the Grid Step 2: Multiply grid by PDFs to generate Cross-Section Dan Clements – Jan 2006 Glasgow ATLAS physics meeting

6. Grids For NLO Jet Cross-Section Calculations (4) • Step 1 is slow, but step 2 is fast allowing for a variety of PDFs to be used to generate cross-sections quickly. • Grid is finite and introduces errors into the calculation. Current Work: • Using and developing code from Tancredi Carli to calculate grids for the Inclusive Jet Cross-Section. • Finding an optimum grid size for cross-section calculations (i.e. accuracy vs computer memory). • Modifying code to allow for efficient step1 production, error analysis and implementation of different PDFs. Dan Clements – Jan 2006 Glasgow ATLAS physics meeting

7. Grids For NLO Jet Cross-Section Calculations (5) Progress • Studies indicate that a (30×30×30) grid in (x1,x2,Q2) can create inclusive jet cross-sections at ~0.5% accuracy. Step 1 generation takes a long time due to statistical errors from NLOJET. • Code has been developed to perform step 1 on multiple nodes for efficient production. • Variable histogram bins in pT implemented to reduce statistical errors in NLOJET. • Error analysis on grids under development. • Trial PDF fitting being carried out by Claire Gwenlan. Dan Clements – Jan 2006 Glasgow ATLAS physics meeting

8. Grids For NLO Jet Cross-Section Calculations (6) • LHAPDF interface to (step 2) created agreement with pure NLOJET (older version) good ~1% CTEQ6.1 Dan Clements – Jan 2006 Glasgow ATLAS physics meeting

9. Grids For NLO Jet Cross-Section Calculations (6) CTEQ6.1 Dan Clements – Jan 2006 Glasgow ATLAS physics meeting