Update on the HZZ 250 GeV +80/-30 250/fb Analysis for the
...
5 September 2013 SiD Meeting:
Archive of the preparation and analysis code:
^subproclst-data8-delay-e250-350Gflav-4jets|\
^flavortag-all-batch-revtx-350-4jets.xml|\ ^revertex-all-batch.xml|\
^runitallbatch2|\ ^suballbatch2|\ ^allSME250p80m30.lst|\ ^ffhzzE250p80m30.lst|\
All difference in the LCFIPlus routines were commented out; not including these should make no difference in the output.
^algoEtc.cc|\ ^FlavorTag.cc|\ ^flavtag.cc|\
Analysis code:
...
Table of Contents | ||
---|---|---|
|
...
Analysis:
- divide into 4 jet / 6 jet topologies
- apply preselection depending on topology
- train/apply TMVA's
- validate with cut table
- check remaining backgrounds
...
TMVA driver routine and cut-table code:
...
Preselection variables:
Code Block |
---|
Evis<140.0: (4 Jet category) # nTrks>10 && # EvisJETS>60.0 && # PTvisJETS>5.0 && # hmass>50. && hmass<140. && # fabs(fabs(dph)-3.14159)>0.15 && # ej1>40. && # jetthrust<0.99 Evis>140: (6 jet category) # nTrks>20 && # EvisJETS<220.0 && # PTvisJETS>5.0 && # hmass>100. && hmass<150. && # fabs(fabs(dph)-3.14159)>0.05 && # ej1<130. && # jetthrust<0.99 |
...
Code Block |
---|
[neal@localhost weights]$ grep -A 8 -i optimal- ../hzzv14-presel-v4-new-common0410results.txt --- Classifier ( #signal, #backgr.) Optimal-cut S/sqrt(S+B) NSig NBkg EffSig EffBkg -- --------------------------------------------------------------------------------------------- --- Cuts: ( 664.0744,1312202.5) -0.0050 0 0 0 0 0 --- Likelihood: ( 664.0744,1312202.5) 1.0000 0.728991 644.0848 779980.1 0.9699 0.5944 --- Fisher: ( 664.0744,1312202.5) 0.0120 1.70335 258.2099 22721.13 0.3888 0.01732 --- BDTG: ( 664.0744,1312202.5) -0.9823 1.17754 538.5353 208621.3 0.811 0.159 --- BDT: ( 664.0744,1312202.5) -0.0462 3.42709 100.6751 762.2891 0.1516 0.0005809 -- --------------------------------------------------------------------------------------------- |
Combined 4-jet 6-jet performance using a cut at EvisJETS of 140 GeV to determine the preselection cuts to be used:
...
Cut table for BDT:
Code Block |
---|
isigall= 1345.984602 isigpresel= 1460.101057 igood= 114.116455 ibad= 762.500000 intot= 77692005.984714 ipre= 876.616455 cuts__ (Cut Name ): all others 2f 4f 6f aa 1f 3f 5f q e+mu tau nu Signif. cut #0 (all ): 7.769e+07 625000 14507062 1019938 0 54073784 0 7464875 0 948 91 45 262 0.15 Delta(sig*BR) = 6.548594 +/- 0.046879 cut #1 (20.<PTvisJETS ): 1.283e+07 37500 5525512 746188 0 2968250 0 3546875 0 806 69 44 259 0.33 Delta(sig*BR) = 3.042092 +/- 0.024172 cut #2 (EvisJETS<220. ): 4.637e+06 25000 3287550 514400 0 130196 0 678875 0 608 42 41 219 0.42 Delta(sig*BR) = 2.365748 +/- 0.020248 cut #3 (50.<hmass<140. ): 3.306e+06 0 2542562 378675 0 60426 0 323875 0 536 38 35 206 0.45 Delta(sig*BR) = 2.233245 +/- 0.019503 cut #4 (||dph|-3.14159|>0.15 ): 2.772e+06 0 2083912 364150 0 57097 0 266000 0 514 36 33 202 0.47 Delta(sig*BR) = 2.120439 +/- 0.018965 cut #5 (nTrks>10 ): 1.426e+06 0 1136375 169088 0 30283 0 89875 0 466 10 19 177 0.56 Delta(sig*BR) = 1.777889 +/- 0.017994 cut #6 (jetthrust<0.99 ): 1.325e+06 0 1044700 165588 0 30033 0 84125 0 463 10 19 173 0.58 Delta(sig*BR) = 1.730704 +/- 0.017728 cut #7 (ej1>40. ): 1.313e+06 0 1035112 164625 0 29215 0 83250 0 462 10 19 173 0.58 Delta(sig*BR) = 1.725415 +/- 0.017704 cut #8 (MVA ): 8.766e+02 0 138 625 0 0 0 0 0 51 0 1 62 3.85 Delta(sig*BR) = 0.259452 +/- 0.015456 |
Remaining backgrounds:
Code Block |
---|
$ sed 's/\./,/g' zzhpassing.dat | awk '{if (NF==18) print $3,$7,$15}' zzhpassing.dat | sort -n | uniq -c | awk '{if (NF==4) aa+=($1*$4);print $1*$2*$4,$0}"\t"$0} END{print "Sum of weights = "aa}' | sort -k 1,1 -n 1278744 Sum of weights = 762,5 12,5 1 106562 4f_sz 12.,500000 1278792 12,5 1 106566 4f_sw 12.,500000 1278864 12,5 1 106572 4f_sz 12.,500000 1278900 12,5 1 106575 4f_zz 12.,500000 1279284 12,5 1 106607 2f_z_ 12.,500000 2557224 25 2 106551 4f_ww 12.,500000 2557272 25 2 106553 4f_zz 12.,500000 2557704 25 2 106571 4f_sz 12.,500000 2557752 25 2 106573 4f_zz 12.,500000 3836664 37,5 3 106574 4f_zz 12.,500000 3836808 37,5 3 106578 4f_ww 12.,500000 12792960 125 10 106608 2f_z_ 12.,500000 16625856 162,5 13 106576 4f_zz 12.,500000 24299556 237,5 19 106577 4f_ww 12.,500000 |
Variable distributions without any preselection:
^variables_id_c1-nopresel.png|\ ^variables_id_c2-nopresel.png|\ ^variables_id_c3-nopresel.png|\
nu nu H only:
qqH,llH only:
|
Background contributions for Evis>140:
Code Block |
---|
[neal@localhost weights]$ sed 's/\./,/g' zzhpassing.dat | awk '{if ($18>=140. && NF==18) print $3,$7,$15}' | sort -n | uniq -c | awk '{if (NF==4) aa+=($1*$4);print $1*$4"\t"$0} END{print "Sum of weights = "aa}' | sort -k 1,1 -n
Sum of weights = 187,5
12,5 1 106551 4f_ww 12,500000
12,5 1 106577 4f_ww 12,500000
25 2 106553 4f_zz 12,500000
25 2 106573 4f_zz 12,500000
25 2 106576 4f_zz 12,500000
37,5 3 106574 4f_zz 12,500000
50 4 106608 2f_z_ 12,500000
|
Background contributions for Evis<140:
Code Block |
---|
[neal@localhost weights]$ sed 's/\./,/g' zzhpassing.dat | awk '{if ($18<140. && NF==18) print $3,$7,$15}' | sort -n | uniq -c | awk '{if (NF==4) aa+=($1*$4);print $1*$4"\t"$0} END{print "Sum of weights = "aa}' | sort -k 1,1 -n
Sum of weights = 575
12,5 1 106551 4f_ww 12,500000
12,5 1 106562 4f_sz 12,500000
12,5 1 106566 4f_sw 12,500000
12,5 1 106572 4f_sz 12,500000
12,5 1 106575 4f_zz 12,500000
12,5 1 106607 2f_z_ 12,500000
25 2 106571 4f_sz 12,500000
37,5 3 106578 4f_ww 12,500000
75 6 106608 2f_z_ 12,500000
137,5 11 106576 4f_zz 12,500000
225 18 106577 4f_ww 12,500000 <***************
|
Code Block |
---|
--- Classifier ( #signal, #backgr.) Optimal-cut S/sqrt(S+B) NSig NBkg EffSig EffBkg
--- --------------------------------------------------------------------------------------------------
--- Cuts: (309.89453, 1128865) -0.0050 0 0 0 0 0
--- Likelihood: (309.89453, 1128865) 1.0000 0.378385 299.2573 625191.7 0.9657 0.5538
--- Fisher: (309.89453, 1128865) 0.0088 1.63217 150.0501 8301.587 0.4842 0.007354
--- BDTG: (309.89453, 1128865) -0.9950 1.40141 134.4158 9065.234 0.4337 0.00803
--- BDT: (309.89453, 1128865) 0.0177 3.63521 38.5486 73.90136 0.1244 6.547e-05
--- --------------------------------------------------------------------------------------------------
after raising reco higgs mass cut from 50 GeV to 100 GeV:
--- Classifier ( #signal, #backgr.) Optimal-cut S/sqrt(S+B) NSig NBkg EffSig EffBkg
--- --------------------------------------------------------------------------------------------------
--- Cuts: (209.30833, 51800) -0.0050 0 0 0 0 0
--- Likelihood: (209.30833, 51800) 0.9105 1.08931 191.9554 30860.64 0.9171 0.5958
--- Fisher: (209.30833, 51800) 0.0176 1.78558 116.8125 4162.945 0.5581 0.08037
--- BDTG: (209.30833, 51800) -0.5098 1.00598 204.0195 40926.24 0.9747 0.7901
--- BDT: (209.30833, 51800) 0.0361 2.92564 72.9263 548.4119 0.3484 0.01059
--- --------------------------------------------------------------------------------------------------
|
Evis<140 GeV and only idrup = 106577 for the background:
Potential tools for reducing the 4f_WW background:
- higher cut on number of charged tracks (unlikely to help S)
- extra cuts but keep any event that has high B-likeness
- redoing n-tuples adding variables for jet combos consistent with W decays (IN PROGRESS)
Note concerning ZZ background: wider cut on acoplanarity should help^TrackSubdetectorHitNumbersDriver.java|\