Home Page of Steve Ellis

1. Talk at QCD/CDF/Theory Jet Workshop at Fermilab, 12/16/02 (PowerPoint file, HTML file, click on animations to start them – if they don’t start automatically). Note that the HMTL version is pretty slow over the NET due to the (huge) AVI files for the animations (there is always a price for new technology). It may be better to download either the ppt file or the HTML file with the associated directory of graphics (located here) and run them locally.

2. Talk at the TeV-Scale Physics Workshop, Cambridge, 7/18/02 (PowerPoint file, HTML file, the HTML version does not handle the animations well and downloads slowly).

3. Talk at Run II Jet Workshop, Fermilab, 1/23/02 (htm file from PowerPoint)

4. Snowmass 2001 talks: mine (PowerPoint); Walter Giele (Postscript).

5. Run II Jet Workshop, Fermilab, 2/8/01 (does not include all figures)

6. Summary Talk at IPPP Workshop on Matrix Elements and Parton Showers, St John’s College, Durham, UK, 12/13-12/15/00 (PowerPoint)

7. Building Better Jet Algorithms, talk to CDF (4/7/00) and MSU (6/19/00), (does not include all figures)

8. CTEQ Summer School 2000 lectures (in postscript): Lecture 1, Lecture 2, Lecture 3, Lecture 4.

Physics related:

Second cut (1/23/04) at Single Top Note in TeX, PDF
8/16/03 version of the “Better Jets” manuscript in PDF formatted (now TeXed); TeX file.
Notes on W decay & constrained fit issues (PDF file)
Draft manual for ME-PS issues
Here are some Flash “movies” illustrating how the kT algorithm combines calorimeter cells, i.e., the movies illustrate the order in which cells (of size 0.1 x 0.1 in a symmetric region around the 2 partons indicated by the contours) are clustered (contiguous cells of the same colors are in a single cluster; the same color can be used by more than one cluster). These are all for the case of large smearing (s = 0.25) with varying values of z, the ratio of the ET’s of the 2 partons, and d, the separation of the 2 partons. First consider the case z = 1.0 and d = 0.71 (all with R = 0.7), which illustrates the generic behavior that the first cells to be clustered are at the periphery and that the clustering tends to grow inward along radial boundaries. This inward growth continues, with only a small amount of clustering in the angular direction, until all of the cells are in a cluster. The final stage involves the clustering in the angular direction, which in the next-to-final step yields 2 clusters with essentially the kinematics of the initial 2 partons. Thus if the 2 partons have d > R, there are 2 final jets (as in the movie). If d < R, there is one final jet. For a very asymmetric situation, z < 0.2, the final stages are more irregular and the partons must be further apart to avoid completer merging. These is illustrates by movies for z = 0.1, d = 0.8 (1 final jet) and z = 0.1, d = 0.9 (2 final jets).
Here are some notes (9/20/02) on the issues of kinematics and jet definitions in PDF format with figures.
Here are some recent (9/3/02) figures concerning the structure of stable cones in the (d,z) plane (the subscripts refer to whether stable cones are centered over the parton at the origin = L, at d = R, or over the central stable cone with both partons = C. The various choices are NLO, Gaussian smearing (s = 0.1), Gaussian smearing (s = 0.25), D0 ET profile, D0 ET profile with 2_LR merging boundary (f_merge = 0.5) indicated, Gaussian smearing (s = 0.1) with ratcheting, Gaussian smearing (s = 0.25) with ratcheting, D0 ET profile with ratcheting, D0 ET profile with ratcheting with 2_LR merging boundary (f_merge = 0.75) indicated. Here are some notes (in PDF format) that try to explain these figures.
Comments on comparing cone algorithms to k_T algorithms at NLO (pdf format, postscript format). – Revised 11/20/20
2/19/02 draft of Snowmass – JEF paper (postscript version). Also available are the TeX file, Fig 1, Fig 2, Fig 3, Fig 4, Fig 5 and Fig 6.
11/18/01 draft of Snowmass – Ratchet paper (postscript version). Here is a PDF version and the tarred, gziped version. Here is a preprint version for the archives.
Here are some comments on Matthias’ graphs and the various algorithms (in postscript) and some notes about simulating ratcheting in JETCLU using the simple 1-D model in the Snowmass note.
New results on Dijet mass distributions: Snowmass cone jets with both jets in the rapidity range 0.1 < | y1, y2 | < 0.7; 0 < | y1, y2 | < 3.0 (note that there is a jet ET cut in half of the curves).
Compare the Cone Algorithm jet rate using Snowmass kinematics with the rate using 4-vector kinematic both to find the cone and to define P_J = P_T; the model result assumes that the difference between the 2 cross section is proportional to the (calculated) log-log derivative of the cross section (the <n> in E_T^-<n>) times a_s with a fitted coefficient of 0.035 . Here is a second graph making a similar comparison including also 2 other algorithms, one finds cones with 4-vector kinematics and the other with Snowmass kinematics; both use the CDF form for the jet E_T, P_J = E_T (CDF) = E P_T/P. Clearly the results depend on the choice of P_J and not the variables used to find the cones, i.e., the two 2 curves are nearly identical.
Comparison to 2001 CDF data for EKS/Snowmass jets with CTEQ4M, CTEQ5M and CTEQ5HJ PDFs.
Jet rates (number of jets per GeV) for RUN IIB compared in Run IB for bins in rapidity: bin 1 (0.1<|y|<0.7) [also a version with Run IIA curves], bin 2 (0.7<|y|<1.4), bin 3 (1.4<|y|<2.1) and bin 4 (2.1<|y|<3.0).
Comparison of CTEQ5M and CTEQ5HJ in the 4 rapidity bins in the form (5HJ-5M)/5M: bin1 (0.1<|y|<0.7), bin 2 (0.7<|y|<1.4), bin 3 (1.4<|y|<2.1) and bin 4 (2.1<|y|<3.0).
Seedless Studies – here are some plot of found jets for event 24749 (Steve, Matthias), 5715 (Steve, Matthias), 9937 (Steve, Matthias); next are some corresponding plots for event 9937 with a range of starting grids for the trial cones – (labeled in the notation delta eta x delta phi): 0.1 x 0.13, 0.1 x 0.26, 0.1 x 0.52, 0.2 x 0.26, 0.2 x 0.52, 0.4 x 0.26, 0.4 x 0.52.
More Seedless – two versions of event 59683 with regular initial grid – mine, Matthias’; then for event 71876 (here is a Mma program for this event, ~ 2 MB) – mine, Matthias’; then for event 16517 – mine, Matthias’.
EKS QCD jet codes
CTEQ Summer School 2000 lectures (in postscript): Lecture 1, Lecture 2, Lecture 3, Lecture 4.
ET weighted flow vector in two shower event
three shower event
well separated three shower event
two shower event with random background ET
ET distribution per 0.1x0.1 cell in two shower event
three shower event
well separated three shower event
two shower event with random background ET
Now to look at some events generated with Pythia to correspond to jets with nominal ET's of 20, 80 and 160 Gev. In each case there is a plot of the ET weighted flow vector and a bar chart of the ET in cones of R=0.7 as a function of the eta and phi of the cone center (so you can see the jets) and also a bar chart of the ET in each tower.
20 GeV - Event 1 ("clean" 2 jets): vector flow, ET in cone and ET in towers.
20 GeV - Event 9 (2 jets + "mini"jet): vector flow, ET in cone and ET in towers.
20 GeV - Event 10 ("clean" 3 jets): vector flow, ET in cone and ET in towers.
160 GeV - Event 1 (2 jets, 1 is "fat"): vector flow, ET in cone and ET in towers.
160 GeV - Event 5 (2 jets + "mini"jet or shoulder): vector flow, ET in cone and ET in towers.
160 GeV - Event 10 (2 jets + "mini" jet): vector flow, ET in cone and ET in towers.
Now let's look at contour plots for Event 1 at 160 GeV showing the regions where the magnitude of the ET weighted flow vector is less than 0.1 (and < 0.05) plus where the ET in towers is greater than 1.0 GeV and where the ET in cones of R=0.7 is greater than 80 GeV .
Simlarly for Event 1 at 20 GeV showing the regions where the magnitude of the ET weighted flow vector is less than 0.1 (and < 0.05) plus where the ET in towers is greater than 1.0 GeV and where the ET in cones of R=0.7 is greater than 15 GeV .
Here is another way to look at these Pythias generated jet events. These plots use color to indicate the level of activity in each tower. The double graphs compare the magnitude of the flow vector (jets show up as "atolls") with the ET in the towers or the ET in cones of R = 0.7. In either case jets are clearly identified by the overlap of small magnitude flow vectors, ringed by large magnitude, with high ET towers or cones.
20 GeV - Event 1 ("clean" 2 jets): towers and cones.
20 GeV - Event 9 (2 "marginal" jets + "mini"jet): towers and cones.
20 GeV - Event 10 ("clean" 3 jets): towers and cones.
160 GeV - Event 1 (2 jets, 1 is "fat"): towers and cones.
160 GeV - Event 5 (2 jets + "mini"jet or shoulder, or maybe 3 jets, plus a "mini" jet): towers and cones. Looking at this event with the "seedless" jet algorithm produces 3 jets with sizeable ET, 2 of which overlap slightly. The participating towers are indicated in the corresponding flow vector plot and combined flow vector and tower ET plot.
160 GeV - Event 10 (2 jets + "mini" jet): towers and cones.
Here are some observations on the application of various related versions of the seedless algorithm to the 160 GeV - Event 5 data in postscript form.

Math