-- AmnonHarel - 12-Sep-2010

Relevant hyperlinks

• Responses to comments from the CWR, Our take on comparing with the inclusive jet preliminary plots
• Wiki page for the dijet centrality ratio post-PRL
• AmnonHarelJetRatio - an older page from before the publication push
• Details on particular subjects: AmnonHarelCLsInDCR, AmnonHarelEnsembleTestsWithFits
• analysis management page - with PAS link, PAS (link broken?), Official analysis Wiki page - EXO10002, Exo-10-002 hyper news
• Jim's input fits:
  • With data: Jim's data plots, data in .root file, data with various models
  • Without data: Signal fits (after inner mass fix), plot for Pythia ratio fit, extended contact interaction MC and fit, demonstrating need for turn offs there, plots of total uncertainties.
• Other webpages from Jim: http://home.fnal.gov/~jhirsch/dijet_hadEinHO/ , http://home.fnal.gov/~jhirsch/dijet_ratio_ptdif_in/, pT balancing studies, the pretty "z" plot, 14ipb, angular cut optimization studies
• Jason's web page for the dijet ratio kept orthogonal to the dijet mass analysis
• AmnonHarelDCRHowToShortenPaper

Alternative sideband fits

	nominal - 489-788GeV	full - 156GeV and up	156-788GeV
NLO with NP corrections	-0.050 +/- 0.013	-0.037 +/- 0.005	-0.036 +/- 0.005
Pythia	0.011 +/- 0.013	0.033 +/- 0.005	0.033 +/- 0.005

The errors shown here are the statistical errors reported by the fitter (ROOT's standard chi^2 fit). They are superceded by the statistical errors from the ensemble testing, which are significantly bigger.

Consistency statistic

Plots

NLO+NPcorr
C with systematics (the usual statistic)		C without systematics
	plot_fg.root
NLO+NPcorr - but without modeling uncertainties (NP corr & PDF)
	plot_fg.root_nm
Pythia
C with systematics (the usual statistic)		C without systematics
	plot_fg_pyt.root

This shows that the data is only consistent with our model when the systematic uncertainties are taken into account. Arguably, the model uncertainties should not be included in this comparison, how ever this has little effect. The choice of SM model is not crucial. All the above are not surprising given the input data and SM models - see Jim's data plot #1, data plot #2, and uncertainties page. On the other hand, some features are noteworthy for what they tell us of the data. Namely, the large difference between the two test statistics in the data (indicating that the data comes from a different distribution?) and the overall level of agreements between data and SM models.

BTW: We should improve data plot #2,

Correlations between feel good statistics

• Note the black triangle that indicates the values in our data
• The black dashed line is x=y

Table

bin #	data	SM	Statistics only			With systematics
		(NLO w. NPcorr)	stat. error	contribution	total	total error	contribution	total
1	0.50672	0.517433	0.0165322	0.419871	0.419871	0.0333583	0.103126	0.103126
2	0.49558	0.520195	0.0218004	1.27485	1.69472	0.0359925	0.467699	0.570824
3	0.505894	0.522922	0.0288795	0.347644	2.04237	0.0405666	0.176189	0.747013
4	0.493892	0.525597	0.0385961	0.674781	2.71715	0.0478999	0.438106	1.18512
5	0.482169	0.528155	0.0135796	11.4679	14.185	0.0312853	2.1606	3.34572
6	0.517212	0.530546	0.0187997	0.503015	14.688	0.033901	0.154688	3.50041
7	0.494711	0.532747	0.0224031	2.88256	17.5706	0.0361143	1.10927	4.60968
8	0.496565	0.534778	0.0285179	1.79545	19.366	0.0401775	0.904578	5.51426
9	0.511478	0.536579	0.0109993	5.20767	24.5737	0.0303682	0.683185	6.19744
10	0.492249	0.53819	0.0131865	12.1377	36.7114	0.0312656	2.15905	8.35649
11	0.48489	0.539546	0.0162432	11.322	48.0334	0.0325728	2.81552	11.172
12	0.485567	0.540677	0.0200651	7.54349	55.5769	0.0345943	2.53772	13.7097
13	0.477836	0.541601	0.0241731	6.95822	62.5351	0.0371366	2.94821	16.658
14	0.540691	0.542334	0.0322465	0.0025955	62.5377	0.0428829	0.00146764	16.6594
15	0.475483	0.542908	0.034561	3.80603	66.3438	0.0447181	2.27341	18.9328
16	0.51357	0.543346	0.0435832	0.46676	66.8105	0.0521629	0.325843	19.2587
17	0.460177	0.543671	0.0490584	2.89657	69.7071	0.0570348	2.14304	21.4017
18	0.443089	0.543914	0.0572215	3.10466	72.8118	0.0644529	2.44708	23.8488
19	0.560694	0.544103	0.070781	0.0549417	72.8667	0.0792203	0.0438594	23.8927
20	0.449153	0.544256	0.0877034	1.17588	74.0426	0.0931079	1.04333	24.936
21	0.488095	0.544394	0.112687	0.249601	74.2922	0.117313	0.230303	25.1663
22	0.469697	0.544524	0.127399	0.344977	74.6372	0.1318	0.322322	25.4886
23	0.487805	0.54465	0.175409	0.105022	74.7422	0.17887	0.100997	25.5896
24	0.576923	0.544764	0.181538	0.031381	74.7736	0.186334	0.0297863	25.6194
25	0.625	0.544852	0.199324	0.161681	74.9352	0.204018	0.154328	25.7737
26	0.473684	0.544892	0.289776	0.060385	74.9956	0.292354	0.0593251	25.833
27	0.625	0.544851	0.240542	0.111023	75.1067	0.24495	0.107063	25.9401
28	0.3	0.544691	0.382066	0.410166	75.5168	0.38436	0.405283	26.3454
29	0.8	0.544365	0.473428	0.291562	75.8084	0.475979	0.288446	26.6338
31	0.25	0.543001	0.851947	0.118281	75.9267	0.853182	0.117939	26.7518
33	0.333333	0.540285	1.28723	0.0258481	75.9525	1.28814	0.0258117	26.7776

• Note that only bins which have at least one entry in both inner and outer bins are used.

Limits on contact interactions

Final results - more CLs limiting values visible due to outragously large ensembles

with systematics
zoomed in	full
clim1n_25KI2_88d.root
without systematics
zoomed in	full
clim2n_30KI2_88d.root

BTW: the points missing from the no-systematics plot are missing because I didn't generate them. Can be easily fixed if there's any reason to do so.

Z plots

cz1n_25KI2_88d.root

Notes:

• the CLs limit at 4.2TeV is very doubtful (~50% chance of it not really being there, in which case it may lie lower and may not exist at all).
• the CLs limit at 4.05TeV appears low in this plot. In fact, there's about a 50% chance that it belongs at -2.645 (which is still below the data, but makes a nicer straight line).
• the existence of the CLs limit at 4.1TeV is driven by the point plotted - our confidence is about 1.5 sigma in MC statistic - again, this is a harder question than whether we exclude this lambda value)
• These and other details are available in https://twiki.cern.ch/twiki/bin/view/Main/AmnonHarelCLsInDCR#Final_results

Presentation:

Present as a Z-plot with only the point at 4.05TeV? More conservative - show 4.05 and 4.1. We have an answer ready if people within and outside CMS want to know more. No one said we got to have a short answer to people who are trying to 2nd guess us.

Technical tests

Systematic variations output from the statistics code

The following plots shows the systematic variations possible in the statistics code. In the limit setting, NPcorr and PDF are used as flat uncertainties. Their full shapes are only used in the QCD ensembles for the p-values. Similarly, the QCD model systematic is only used in limit setting (it's |Pythia-NLOcorr|).

plot_sm_syst.root	plot_sm_syst2.root

Unsorted Attachments