21 October 2005 Top 10 Fest Some kind of overview Had a lengthy “interview” with the organizers:...
-
Upload
clare-garrett -
Category
Documents
-
view
214 -
download
1
Transcript of 21 October 2005 Top 10 Fest Some kind of overview Had a lengthy “interview” with the organizers:...
21 October 200521 October 2005 Top 10 FestTop 10 Fest
Some kind of overviewSome kind of overview
Had a lengthy “interview” with the organizers:Had a lengthy “interview” with the organizers:
Told to be funnyTold to be funny
Asked to be non-TechnicalAsked to be non-Technical
All of these, in 20 minutesAll of these, in 20 minutes
Required to be P.C.Required to be P.C.
Begged not to offend anyoneBegged not to offend anyone
Instructed to have many pictures (and be funny)Instructed to have many pictures (and be funny)
Sure to give credit to everyoneSure to give credit to everyone
19pb19pb
57pb57pb
Top Turns 10Top Turns 10 (or 11)(or 11)
Au
g 9
3A
ug
93
Dec 9
4D
ec 9
4
21 October 200521 October 2005 Top 10 FestTop 10 Fest
Run 0 - Brief race with UA2 Run 0 - Brief race with UA2 (88/89)(88/89)
21 October 200521 October 2005 Top 10 FestTop 10 Fest
Realizing that Top is heavier than W Realizing that Top is heavier than W (89/90)(89/90)
Mtop > MW + Mb
• Old trick does not work in this region• Much dirtier environment• different sources of BG - not well understood• A lot less signal (10 ev/yr for Mtop=175 GeV)
Our last limit was: Mtop > 91 GeV
– Based mostly on Di-Lepton analysis– And first (μ−only) SLT
21 October 200521 October 2005 Top 10 FestTop 10 Fest
After Run 0After Run 0
a few years with lots of work on detector and software:
- Detector upgrades (muons, VTX, SVX, Trigger)
- Many lessons from Run 0 --> Run 1 software improvements
==> We all knew: #1 mission is search for the top quark
We knew it was going to be a competitive analysis, but only later we fully realized what it takes.
To really "nail" it one must demonstrate having identified:
- 2 W bosons
- at least one of the two b jets
- consistent masses of two combinations
All at onceAll at once
21 October 200521 October 2005 Top 10 FestTop 10 Fest
Between the Runs Between the Runs (90/92)(90/92)
lengthy preparations, various groups tooling with taggers, mass fitters, working groups forming, splitting…
• Atmosphere was intense. Fierce competition - internal. Lots of discussions, comparisons. Bunch of kids, really. – One main mtg, bi-weekly. (for both top and b physics!!!) – Rather loosely organized and very few sub-groups– Groupings of 2-4 students and post-docs (some very young profs)– Lots of corridor talks, some easily qualified as trash-talk...
• sometime negative…– “Secret analysis ideas…” - Everyone looking for a special angle– We all wanted to be the first to “do it”– Some were worried they might get scooped…– All convinced that it can and will be done by one analysis!
21 October 200521 October 2005 Top 10 FestTop 10 Fest
In the meantimeIn the meantime
- data sample in hand small (4 pb), nonetheless re-analyzed many times and some early discoveries made.- Finding. Un-finding. Always fun…- Some good ideas during this time- Importance of mass measurement, tools- Position on kinematical analyses established
- continuously, we all asked ourselves things like:– how heavy will it be?– how much data do we need to see it?– what will be the killer measurement that will lead to the
"smoking gun”-quality evidence. – Lots of running around with ideas…
21 October 200521 October 2005 Top 10 FestTop 10 Fest
Top Production and decay - a quick overviewTop Production and decay - a quick overview
DILDIL - both W decay to leptons - both W decay to leptons
Low rate, Very cleanLow rate, Very clean
SVXSVX - one of the b-jets is identified - one of the b-jets is identified using a displaced vertex tagusing a displaced vertex tag
SVX detectorSVX detector
SLTSLT - one of the b-jets is identified - one of the b-jets is identified using a leptonic decay productusing a leptonic decay product
lepton I.D. in jetslepton I.D. in jets
21 October 200521 October 2005 Top 10 FestTop 10 Fest
What Was the “Plan”?What Was the “Plan”?
There was no plan - not in the sense of planning.
We sort of wandered/staggered into one, as we were winging it…
Counting experiments:
DIL: focus on the two W's as a selector GOLDEN MODE
SLT: focus on one W and a b jet identified by its decay mode
SVX: focus on one W and a b jet identified by its lifetime
NONE OF WHICH IS CONCLUSIVE! (on its own)
– do we see top? Need at least two of the above
• Our goal: cross-section measurement
- Mass peak– mass determination
21 October 200521 October 2005 Top 10 FestTop 10 Fest
The “Plan”The “Plan”
Did not plan on, or studied combining channels. – always thought it will be light - like everyone else did before – each of us "went for it”– did not think we could observe such a heavy top! (120-130 max) with
such a small sample size (20pb)
Did not think the mass or kinematical information can be a part of the evidence for top
(concerned about theoretical biases, simulation errors…)– We kind of thought that we will: – observe top via counting experiments (a counting
experiment)– Measure mass, study kinematics - afterwards
21 October 200521 October 2005 Top 10 FestTop 10 Fest 12
First HintFirst Hint
21 October 200521 October 2005 Top 10 FestTop 10 Fest
The first hint The first hint (92)(92)
DPF event– beautiful e-mu , two energetic jets, large missing Et– a bit strange, but hey, nobody’s perfect– everyone studied it in great detail. From various perspectives.
>> “Bambi’s mother was shot” syndrome…– and a b-tagged jet too! (but we did not want to show it...)
Showed-up very early in the run
So naturally, we expected - many more soon - large number of “relatives” in other channels
long silent period.
Back to work…
21 October 200521 October 2005 Top 10 FestTop 10 Fest
What did What did youyou do when the DPF event came in? do when the DPF event came in?
21 October 200521 October 2005 Top 10 FestTop 10 Fest 15
First Hint:First Hint:As shownAs shown during the Fall 1992 DPF mtg at during the Fall 1992 DPF mtg at
FermilabFermilab
Alvin’s Swiss army knife was here…
21 October 200521 October 2005 Top 10 FestTop 10 Fest
Top Discovery - What does it take? Top Discovery - What does it take? (92/93)(92/93)
- Hugely debated, lots of soul searching, long before any (real) signal showed up…
- Discussions always open and exchanges passionate…- Many forks along the road
--> CDF always chose the more conservative approach
A few “case studies”:1. Usage of MC or theoretical inputs to estimate backgrounds• Blind analysis, or analysis-cuts “adjustment”• Poking holes• What did you know, and when did you know it?
21 October 200521 October 2005 Top 10 FestTop 10 Fest
1. Usage of MC/Theory1. Usage of MC/Theory
UA1 - a lesson from our immediate predecessors.
1984:– Using isolated high transverse momentum lepton
– 2 or 3 hadron jets– Observed 5 events (e+ >=2 jets); 4 events (μ+ >=2 jets)– Expected background: 0.2 events
• fake leptons dominate
• bb & cc production negligible– Conclude: results consistent with M top = 40 ±10 GeV
1988:– x6 the data– much better understanding of backgrounds– M top > 44
QuickTime™ and aTIFF (LZW) decompressor
are needed to see this picture.
21 October 200521 October 2005 Top 10 FestTop 10 Fest
2. Blind analysis - what blind analysis?2. Blind analysis - what blind analysis?
We were not doing Blind analysis. At least not consciously…
Had to somehow control our quick fingers - self and mutual policing & enforcement
Example 2a. Pseudo e-mu event- Event clearly hi Pt di-lepton event- Electron slightly non-fiducial- Such where included in previous and later analyses!- Impact of assignment on overall significance of result
- Painful, long arguments.- Decided to stick to the “no changing cuts” rule
21 October 200521 October 2005 Top 10 FestTop 10 Fest
Blind analysis - what blind analysis? (part ii)Blind analysis - what blind analysis? (part ii)
Example 2b. SLT PT cut choice:
- Chosen to be 2 GeV at start of search (when top could have been anywhere between 120-200).
- After looking at the Pt spectrum of our tags, was obvious that a 4GeV cut will be slightly better…
- Painful, long arguments.
- Decided to stick to the “no changing cuts” rule
21 October 200521 October 2005 Top 10 FestTop 10 Fest
3. …and others do not!3. …and others do not!
Looking for holes–It was FUN to find a problem–We competed to do so. (Always nice to break something)–Extremely healthy atmosphere of continuous scrutiny and criticism–Not exactly for the faint of heart, though…
- Z+jets: - too many tags (expected 0.6 see 2 events)
- W+4jet events: - background “deficit”
Inserted the following text into the abstract: “We present several cross checks. Some support this
hypothesis, others do not.”
21 October 200521 October 2005 Top 10 FestTop 10 Fest
4. a-priori Vs. a-posteriori knowledge4. a-priori Vs. a-posteriori knowledge
What are we "allowed" to use?
2.8 sigma from counting exps
> Additional 2.3 sigma --> mass
> Not used,
because it was not in the “plan”
CDF measured:
(1994) Mtop = 174 +- 10 +- 13 GeV
(1995) Mtop = 176 +- 8 +- 10 GeV
(2005) Mtop = 172 +- 3.7 GeV
World average, as of today:
172.7 +-2.9 GeV
CDF 1994CDF 1994
21 October 200521 October 2005 Top 10 FestTop 10 Fest
Putting it all together - take 1 Putting it all together - take 1 (93)(93)
• Consensus building - how did we decide on things:– Do we have it?
• We all tell our tales. None of us actually have it.
• Hmmm, shall we combine?• Very tentative get-together: Wrote 4 PRL’s• God-Parent committee assigned shortly after that• Plan was to present papers in the next collaboration mtg. Oct 93.
Channel/(# of ev)
Est. BG Observed
DIL 0.6 2
SVX 2.3 6
SLT 3.1 7
2315.4
276.7
61.3
ObservedEst. BG
199519952005
Est. BG Observed
15 46
32 138
16 34
21 October 200521 October 2005 Top 10 FestTop 10 Fest
Putting it all together - the real thing Putting it all together - the real thing (94)(94)
4PRL’s --> 1 PRD • OK, so each counting experiment writes up their work in a draft PRL• Add it all up in a fourth one• Collaboration shreds the whole thing. Paper writers momentarily
crushed…• A new “Plan” is put in motion• Six months later…
Title of the PRD - the name of the beast• Hours of public debate, dictionary usage, joking, looking at the data,
joking. • In a way, establishing a joint, common “comfort level” with our
analysis. • Obvious proposals:
– Search for…– Evidence for…– Observation of…
• Some, more offbeat, interesting proposals offered as well==> “Evidence” was the consensus
21 October 200521 October 2005 Top 10 FestTop 10 Fest
Last Slide Last Slide (95)(95)
• Had a few very scary weeks at the start of run 1b - no top?
• Statistics played us a bit
• SVX replacement, tagging algorithm re-visited
• Introduced a much more sophisticated (and more MC dependent) BG estimate method - The famous “Method 2” (x-check with old one)
• Gave a heads-up to D0 (as agreed to before the run)
SUBMITTED PAPER - very clear confirmation (see plots)
Since then:
• Mass determination became a small industry
• Studies of properties of top production and decays
• Soon: 1Hz production rates @ LHC
21 October 200521 October 2005 Top 10 FestTop 10 Fest 25
Figures from the 1995 Discovery paperFigures from the 1995 Discovery paper
QuickTime™ and aTIFF (LZW) decompressor
are needed to see this picture.