PDF Issues in Faserν and SND - Robert Thorne January 15th 2021 University College London - CERN Indico

Page created by Eduardo King
 
CONTINUE READING
PDF Issues in Faserν and SND - Robert Thorne January 15th 2021 University College London - CERN Indico
PDF Issues in Faserν and SND
           Robert Thorne

         January 15th 2021

      University College London

          CERN – January 2021
PDF Issues in Faserν and SND - Robert Thorne January 15th 2021 University College London - CERN Indico
I will discuss what opportunities we envision for learning about PDFs
from potential data at the FASERν and SND in the context of what has
happened at previous DIS experiments.

Essentially all related to heavy quarks, i.e. charm, in one manner or
another.

Possible future interest in broader area, e.g. nuclear PDFs, but not
concentrated on here - though many issues common to nucleon and
nuclear PDFs in practice.

CERN – January 2021                                                 1
PDF Issues in Faserν and SND - Robert Thorne January 15th 2021 University College London - CERN Indico
Comparison of neutrino flux/interactions - FASERν 1908.02310.

FASERν flux dominated by νµ(ν̄µ) coming from pion and kaon decay,
i.e. not charm mesons in the initial state.

CERN – January 2021                                             2
PDF Issues in Faserν and SND - Robert Thorne January 15th 2021 University College London - CERN Indico
At low energies νµ(ν̄µ) dominates flux at SND (2002.08722), but at
energies up to Eν ∼ 104GeV σν ∝ Eν .

At high Energies SND flux dominated by νe(ν̄e), and with larger
contribution from ντ (ν̄τ ). These are coming from charm meson decay.

CERN – January 2021                                                 3
PDF Issues in Faserν and SND - Robert Thorne January 15th 2021 University College London - CERN Indico
Consequence of distance from beam axis of respective detectors, with
shorter decay time for heavy measons facilitating more contribution
away from beam axis.

CERN – January 2021                                                4
PDF Issues in Faserν and SND - Robert Thorne January 15th 2021 University College London - CERN Indico
Can see this also in numerical comparison of CC interaction expected,
and mean energies.

                      FASERν – 1908.02310

                        SND – 2002.08722

CERN – January 2021                                                 5
PDF Issues in Faserν and SND - Robert Thorne January 15th 2021 University College London - CERN Indico
Hence, more scope for FASERν in terms of physics where the (largely
muon) neutrino is the probe of the nucleons in the FASERν target.

Fig. from 1908.02310.

Spectrum extends to a slightly higher energy than previous fixed-target
neutrino DIS experiments, e.g NuTeV.

CERN – January 2021                                                   6
PDF Issues in Faserν and SND - Robert Thorne January 15th 2021 University College London - CERN Indico
Neutrino Cross section.

Very firmly in the regime of Deep-Inelastic Scattering, and determined
essentially entirely by PDFs. At LO in QCD

Where q = (u + d)/2 + s + b and q̄ = (d¯ + ū)/2 + c

Where for ν → ν̄ then q → q̄.

y is inelasticity.

CERN – January 2021                                                  7
PDF Issues in Faserν and SND - Robert Thorne January 15th 2021 University College London - CERN Indico
Q0 is a cut-off for the “non-perturbative regime”, plot by P. Plesniak.

At most a few % contribution from nonperturbative regime, mainly near
E = 100GeV.

CERN – January 2021                                                       8
PDF Issues in Faserν and SND - Robert Thorne January 15th 2021 University College London - CERN Indico
PDFs regime of 0.001 < x < 0.03 and 1GeV2 < Q2 < 50GeV2.
Dominant contributions from u, d which are known to couple of percent
level.

Cross section and uncertainty (at LO for now) using MMHT2014 PDFs
of order 2% and similar for MSHT2020 (plot by T Cridge).

CERN – January 2021                                                 9
Nuclear PDF corrections add additional (essentially uncorrelated)
uncertainty of ∼ 3 − 4%.

Plot from 2001.03073.

CERN – January 2021                                            10
Total uncertainty in normalization of order ∼ 5%. Better than knowledge
of flux, and of FASERν sensitivity.

Plot from 1908.02310.

CERN – January 2021                                                  11
Charged Current Charm production
However, if heavy flavour detected in the final state more interesting for
PDFs
Long existing measurements of dimuon production in neutrino DIS on
heavy targets CCFR, NuTeV, (Phys.Rev.D 74 (2006) 012008).

Dominated by strange (anti)quark initial state - traditionally main
constraint on strangeness.

CERN – January 2021                                                     12
Measurements using both νµ and ν̄µ beams at different energies.

Possibility of reaching up to ∼ 5 times greater energy at FASERν, i.e.
∼ 5 times lower in x. Also no dependence on D → µ branching ratio,
which gives O(10%) uncertainty

CERN – January 2021                                                 13
Extremely high precision data on W, Z from ATLAS

Difficulties in fitting both W and
Z distributions fixed by increase
in strange quark fraction
        s+s̄
RS =    ū+d¯

in ATLAS study (PRL 109
(2012) 012001).

CERN – January 2021                                14
MSHT2020 PDFs compared to MMHT2014 at NNLO.
                      s+s̄
Look at Rs =          ū+d¯
                           .

Currently an increase in the strange quark below x = 0.1 due to W, Z
data (mainly ATLAS 7, 8 TeV).
Still significant uncertainty, and some tension, at higher x. Important
constraint from FASERν possible.

CERN – January 2021                                                  15
Prediction (at LO for now) for charm fraction in final state from
MMHT2014 PDFs and MSHT2020 PDFs (plot by T Cridge).

Need measurement of precision better than ∼ 10% for constraint, but
currently tension between LHC and DIS measurements.

CERN – January 2021                                              16
Charm meson production probed by neutrino production.

Charm mesons dominant source of neutrino flux, or more specifically
interactions at SND.

Charm produced in LHC collisions
at pseudo-rapidity ∼ 7 − 9.

Corresponds to x1 ∼ 0.04 − 0.8
and x2 ∼ 10−7 − 5 × 10−5.

On the edge of LHCb lower
limit.

CERN – January 2021                                              17
Theoretical Issues in cross section calculation.

One parton at potentially very high x and another at very small x leads
to many complications/opportunities.

At high x there is the question of
intrinsic charm.

What is the status of intrinsic
charm?

Plot from 1908.02310.

CERN – January 2021                                                  18
Intrinsic charm

Formally of higher twist, i.e. O(Λ2/m2c ).

Proposed that it could be enhanced at high x by Brodsky et al in 1983.

               ĉ(x) = Ax2[6x(1 + x) ln x + (1 − x)(1 + 10x + x2)]

Possible enhancement at high-x, similar to the large higher twist
expected at low W 2.

Therefore no expected constraint from HERA data.

Potentially of relevance for LHC neutrino data.

CERN – January 2021                                                  19
Investigations in PDF fits

CT add this type of intrinsic charm model, and sea quark type intrinsic
charm, and see the effect on global fits. Some limited evidence for
preference, e.g JHEP 02 (2018) 059.

CERN – January 2021                                                  20
NNPDF recently started including “fitted charm” (not identical to
“intrinsic charm”) as their default, e.g. EPJC 76 (2016) 11, 647.

Very different to perturbative only charm, with much bigger uncertainty.

Larger than perturbative charm at highest x but smaller for x ∼ 0.01.

CERN – January 2021                                                     21
Also very different to CT instrinsic charm models.

Tends to be lower that perturbative charm at lowish x.

Possibility of large terms in g → c in transition matrix.

Really higher order theory correction?

CERN – January 2021                                         22
Small-x issues.

However, independent of the issues associatted with intrinsic charm at
high-x, there are numerous uncertainties associatted with convergence
of perturbation theory, particularly related to small-x.

To begin with there is simply very large scale uncertainty (approximate
indication of theory uncertainty) from the NLO calculation (Bai et al.
2002.03012 ).

CERN – January 2021                                                  23
PDF uncertainty.
As well as (perturbative theory uncertainties) there is a huge intrinsic
PDF uncertainty at this very low x value.
                                 xg(x,Q), comparison
               10
                                                   MSHT2020NNLO
                 8                                 NNPDF3.1
                 6                                 CT18ANNLO
                                                   Q = 2.00e+00 GeV

                                                                            Generated with APFEL 2.7.1 Web
                 4
                 2
                 0
               −2
               −4
               −6
               −8
              −10
                10−7   10−6   10−5   10−4       10−3    10−2     10−1   1
                                            x

There is literally no constraining data in this regime in PDF fits, so any
genuine uncertainty estimate explodes here.

CERN – January 2021                                                                                          24
Large scale uncertainty a feature of low physical scale ∼ mc and
relatively slowly convergent series.

However, also the problem of small-x logarithms

                             ∞
                         αS2 X αSn lnn(1/x)
                      σ∝       σn
                         x n=0       n!

 i.e. in terms of the fractional momentum x of the one of partons there
is an extra power of ln(1/x) for each power of αS

In this case αS ≈ 0.35 while ln(1/x) can be as low as ln(1/x) ∼ 15 (the
cross section involves a convolution so this is the maximum value).

Scale variation at fixed order gives no information at all about missing
higher power terms of ln(1/x).

CERN – January 2021                                                   25
Fits to LHCb data including ratios (both from different energies and
different rapidities) can reduce effect of scale uncertainties (Gauld et
al 1610.09373).

Not guaranteed to be so stable under genuine higher-order corrections.

CERN – January 2021                                                   26
Long-known and hugely studied problem mainly in the context of
structure functions.

Basic resummation of leading and next-to-leading logarthims known -
BFKL resummation.

Far more required for realistic and accurate resummation including
running coupling effects, momentum conservation .......

Known moderately well (we think) for structure functions.

CERN – January 2021                                              27
For example, the fit to final HERA inclusive cross sections is not optimal
at NNLO at low x and high y, where

                      σ ∝ F2(x, Q2) − y 2FL(x, Q2)

Improved by small-x resummation, as shown by e.g.             the xFitter
collaboration 1802.00064 and NNPDF 1710.05935.

CERN – January 2021                                                     28
Improvement due to increase in FL(x, Q2) for the same fitted F2(x, Q2)
which dominates the cross section in most regions.

i.e. small-x resummation leads to different changes for different physical
quantities.
Could alter charm cross section at the LHC for a given fit gluon, but this
is not calculated at all precisely yet.

CERN – January 2021                                                     29
Note the same sort of small-x conclusions were already made in 2007
(hep-ph/0611204) with less precise earlier HERA data and more basic,
but qualitatively very similar resummation calculations.

CERN – January 2021                                               30
Fits with resummation in PDFs and LHCb data performed (Gauld et al
1808.02034).

However, resummation not performed in cross section.

CERN – January 2021                                             31
Beyond Leading Twist Perturbation Theory.

In this regime can also get potential non-universal corrections at higher
twist O(Λ2QCD /Q2) corrections at small x.

Expected (with some uncertainty/disagreement) due to saturation of
cross section and PDF growth - this can start to occur roughly when
xg(x, Q2) > Q2R2, where R2 ∼ (0.2GeV)2.

For example, improvement in fit to high-y, low-x, Q2 HERA data can be
explained phenomenologically by O(Λ2QCD /Q2) corrections to FL(x, Q2)
(1601.03413, 1704.03187 – plots from latter).

CERN – January 2021                                                    32
Conclusions

FASERν is particularly well suited to use the dominantly muon neutrino
flux from light mesons to measure charm production in the final state.

This could provide a good constraint on the (relatively uncertain) strange
quark in a x-range extending a little beyond previous DIS experiments.
Also in a region where less direct LHC - ATLAS constraints are in
tension to some degree with DIS constraints.

Dominant flux produced by charm mesons from SND can be a useful
constraint on charm production from high-x and very low-x PDFs.
FASERν elecron and tau contributions sensitive at even more extreme
values.

Impact on study of intrinsic charm at high-x probably limited.

PDF, particularly gluon uncertainty so large for x < 10−6 even though
there are a wide range of large theoretical uncertainties – fixed order
(scale), small-x resummation, higher-twist, saturation – a measurement
will give very useful constraints. Uncertainties likely minimised by ratios
in which systematic uncertainties (theory and experiment) cancel.

CERN – January 2021                                                      33
Back-up

CERN – January 2021   34
High-x Strange Quark.

There is also the possibility
of looking at the less                     νe
well know strange quark             e
via charm quark jets
at the EIC.

Requires dealing with
fragmentation (but so
                                           W
do some current methods
at some level).
                                                    jet
                                p

Plot from https://arxiv.org/abs/2006.12520 Arratia et al..

CERN – January 2021                                          35
From a CT study one
can see the likely x                                                    1

                         reco Q2JB [GeV2]
and Q2 range likely to                                                  0.9

be covered.                                                             0.8
                                                                        0.7
Higher x than the                           103                         0.6
main constraints from                                                   0.5
the LHC, and from the                                                   0.4
most precise dimuon                                                     0.3
measurements.                                 2
                                                                        0.2
                                            10
                                                                        0.1
                                                                        0
                                                  −2    −1
                                             10        10           1
                                                             reco xJB

Plot from https://arxiv.org/abs/2006.12520.

CERN – January 2021                                                     36
Inclusion of new NNLO corrections.

NNLO corrections to dimuon production (Phys. Rev. Lett. 116 (2016),
Berger et al., J. Gao, arXiv:1710.04258).

NNLO correction negative, but larger in size at lower x

Impacts on tension between dimuon data and LHC W, Z data.

CERN – January 2021                                              37
In some NNPDF fits try fitting old EMC data (NPB 213 (1983) 31-64).

Clearly prefers supression at smaller x, and also some enhancement
as x → 1.

Consistent with suggestions from LHC data sensitive to flavours (W, Z).

Note however, EMC data relied on large corrections using LO theory
generators and extremely basic PDFs. Significant question mark.

CERN – January 2021                                                   38
MSTW tried fitting EMC data.
Overshoot lower x data even at
NLO with dynamical charm.

High-x intrinsic charm with modified
coefficient functions,
m2c → m2c + Λ2, at threshold
works ok.

At low Q2 and W 2 we likely need
to worry about nonperturbative
or higher twist effects beyond
just that of intrinsic charm.

Similar with hadronization issues
in Monte Carlos for higher
energy production.

CERN – January 2021                    39
Choices for Heavy Flavours in DIS.

Near threshold Q2 ∼ m2H massive quarks not partons. Created in final
state.

Described using Fixed Flavour Number Scheme FFNS.

Variable Flavour - at high scales Q2  m2H heavy quarks behave like
massless partons. Sum ln(Q2/m2H ) terms via evolution.

Partons in different number regions related to each other perturbatively.
                       n +1                            n
                      fj f    (Q2) = Ajk (Q2/m2H ) ⊗ fk f (Q2),

Perturbative matrix elements Ajk (Q2/m2H ) (Buza et al.) containing
                          n             n +1
ln(Q2/m2H ) terms relate fi f (Q2) and fi f (Q2) → correct evolution for
both.

CERN – January 2021                                                    40
Affects distinction between “intrinsic charm” and “fitted charm”.

CERN – January 2021                                                 41
Higher orders

At O(αS3 ) similar form seemingly, but AHg not yet fully known.

Calculation of matrix elements part of an enormous project by Blümlein
et al. e.g (Nucl.Phys.B 890 (2014) 48-151).

Goes into PDFs or into NNLO FFNS cross sections.

CERN – January 2021                                                  42
Level of uncertainty in AHg (top) and Agg,H (bottom) (McGowan).

CERN – January 2021                                               43
You can also read