Download RHZYAM (Red Herring ZYAM)

RHZYAM
(Red Herring ZYAM)
Why I think ZYAM is not the
most important problem for us to
be focusing on right now
Or
Dr. (Strange)Cole: How I learned
to stop worrying and love ZYAM
Why are we having this discussion?
We’re here
because of
this feature
in the data
•Issue to put to rest: is it an artifact of the
background subtraction procedure?
My final (penultimate) word on subject
BAC
QM08
Seeing the “Mach Cone” without subtraction
Pure correlation
function, no
subtraction.
0-5% centrality –
large background
but modest flow
• Can see the distortion in the di-hadron correlation
function without ever subtracting.
– Of course, there is modulated background
– Subtraction will decrease yield at ∆ϕ = π relative to π ± 1.
Seeing the “Mach Cone” without subtraction(2)
Use dependence
of correlation
function on angle
of “trigger” wrt
reaction plane
“Trigger” here is
an analysis cut
on data
• Characteristic variation depends on angle of trigger
bin and resolution of reaction plane measurement.
• In one bin, flow effects are weak.
Seeing the “Mach Cone” without subtraction(3)
PHENIX
Run 5
4th reaction
plane bin w/
weakest flow
Pure
correlation
function, no
subtraction.
• Without subtraction, the “mach cone” is clear
• Critical point:
– Modification of ∆ϕ distribution at π ± ≈1 far more important
than the behavior at ∆ϕ= π.
Seeing the “Mach Cone” without subtraction(4)
PHENIX Run 7
0-5 centrality
correlation
functions
(not optimal plot
but only approved
corr. func. plot)
• One important point to take away:
– Dramatic change in shape of C2 vs ϕtrig.
– Substantial errors in v2, v4, background would cause flow
variations to swamp any jet / “mach cone” signal.
Reaction Plane Dependence After Subtraction
Centrality
0-5%
• Many details to control in the subtraction.
– v2(pt), v4(pt), Ψ resolution, background level
• But, see same “Mach Cone” while flow-modulated
background changes by 10x signal at ∆ϕ = π ± ≈1
Reaction Plane Dependence After Subtraction
Centrality
30-40%
• Flow modulation ~ 4-5 x larger here than in 0-5%.
• Nonetheless, see ~ identical shape for “Mach Cone”
Conical? flow – RP dependence (STAR)
From parallel talk by A. Feng
• PHENIX & STAR results on RP dependence in
excellent qualitative agreement.
Background Normalization
• The problem:
– How to find the level of the
combinatoric background in
di-hadron correlations.
• In principle – can measure
yield, v2 of each hadron,
calculate background level.
– “absolute normalization”
– But, not so simple (e.g.
multiplicity fluctuations).
• ZYAM:
– Use feature of jet C2 to
determine/constraint
background level.
• Comparison used for
systematic errors in
Two-Component Model?
But wait, wasn’t previous slide predicated on
“two-component” assumption?
• What is the two-component “model”?
– Some hadrons correlated with jet (trigger hadron).
– Some hadrons not correlated with jet (trigger hadron).
⇒These have the “inclusive” ϕ - Ψ distribution.
• I refuse to accept that every hadron produced within
∆η=1 of a jet is affected by the presence of jet.
– e.g. in 0-5% centrality
– “jet” correlation/total ~ 0.01
• So how much is correlated?
• We don’t know.
Two-component model? (2)
So, what to do?
• One reasonable way to proceed:
– Assume that the maximum # of hadrons consistent with C2 are
uncorrelated w/ jet (minimal jet-medium interaction).
– ZYAM assumption.
• Pro’s:
– ZYAM gives (~ unambiguous) estimate of background.
– Relatively insensitive to details of modifications of jet shape.
– Lower limit on conditional yield.
• Con’s:
– “Ad hoc”
– We know it’s wrong (but how wrong?)
– Makes it easy to generate “hidden” bias.
Why ZYAM is wrong
PHENIX paper (arXiv:0801.4545v1 [nucl-ex])
p-p di-hadron pair
yield per event
(NOT conditional
yield)
•Even in p-p
collisions ∃ no
region in ∆ϕ
where di-hadron
yield is zero
“Underlying Event” Does it Matter?
• Underlying event in p-p is ~ 1/100th of jet signal
• Jet signal in Au-Au is ~ 1/100 of inclusive pairs.
• Jet-associated underlying event ~ 10-4 inclusive pairs.
– Would have to be enhanced by a factor of 100 to have a
significant impact on background normalization.
• Forget it. The underlying event (IS radiation, etc.) not
relevant (yet) to this discussion.
Unknown Jet-medium correlations?
• So, ZYAM assumption is reasonable based on p-p
data (also d-Au data).
• But, what if there are unknown jet-medium
correlations that contribute near ∆ϕ = π/2?
– Then we will over-subtract.
– If it does not follow flow pattern, we will subtract
too much with the wrong shape.
But, we see “Mach Cone”
before subtraction where
we can make flow
modulation small.
Everywhere else we get
consistent results after
subtraction.
My conclusions
• The “Mach Cone” is absolutely, positively not an
artifact of background subtraction.
• ZYAM is not correct, not ideal, would prefer
something much better.
– (e.g.) absolute subtraction – but also depends on the socalled “two-component” assumption.
– No evidence from data for strong violation of “twocomponent” assumption.
My final (I hope) word on subject
BAC
QM08