Yellow Report Physics Jets-HF

From eicug
Jump to navigation Jump to search


The Jets and Heavy Flavor working group meets Mondays at 12:00 PM Eastern. Meeting agendas can be found on our indico page

Kinematic coverage files

This section summarizes the kinematics studies of the Jets-HF working group. We have simulated the following processes:


Simulation Software
PythiaeRHIC (aka ePythia)
Simulation Setup Files
steerfile_e10_p100 (Electron Energy = 10 GeV Hadron Energy = 100 GeV)
steerfile_e18_p100 (Electron Energy = 18 GeV Hadron Energy = 100 GeV)
steerfile_e18_p275 (Electron Energy = 18 GeV Hadron Energy = 275 GeV)
Selected Particle Species
π+ / π-
K+ / K-
P+ / P-
Cuts / Data Selection
Drawing Macro and Script



These kinematic plots are plotted based on the PYTHIA8.240 simulation.

1, the macro to run the pythis8.240 is This macrot will produce the output file: kin_hist_s63.root

2, to plot the kinematic distributions, the macro is Plot_kin_eic.C

root -l Plot_kin_eic.C

will produce the D-meson angle and momentum distribution as suggested by the EICUG and the full momentum versus pseudorapidity distributions as well. You could switch to the B-meson to produce the relevant kinematic distributions as well.

For example: D_etap_s63.gif is the full momentum versus the pseudoradity for D-mesons produced in 10GeV election and 100 GeV proton collsions with integrated luminotiy at 10 fb^{-1}.

Tarfile: EIC_kin_XuanLi.tar

Photoproduction Jet Kinematics

This section contains kinematic plots for photoproduction jets (10-5 < Q2 < 1 GeV2) with transverse momenta greater than 10 GeV. The kinematics of the jet constituents are also shown. Currently, only the highest COM energy has been simulated. Input to the jet finder was all stable particles (except the scattered electron) with |η| < 3.5.

Please contact Brian Page ( with any questions.

PYTHIA6 (pythiaeRHIC implementation)
Event Cuts
10-5 < Q2 < 1 GeV2
0.01 < y < 0.95
Electron Energy = 18 GeV
Proton Energy = 275 GeV
Particle Cuts
|η| < 3.5
Exclude scattered electron
Jet Cuts and Parameters
R = 0.8 and R = 0.4
Jet pT > 10 GeV

Example jet kinematic plots can be seen below for transverse momentum vs theta and energy vs theta for a jet radius of 0.8. Plots for a jet radius of 0.4 as well as pT and energy vs theta for the jet constituents can be found in this zip file: This archive also contains a root file of all the plots, a light-weight tree with all relevant kinematic variables, and a macro which can read the tree and produce output plots.

Jet pT vs theta R = 0.8
Jet energy vs theta R = 0.8

Heavy-Flavor Kinematics

This section gives the kinematics of heavy-flavor hadrons and their decay particles. These distributions are generated using the pythiaeRHIC implementation of PYTHIA 6. Events are required to have Q2 > 1 GeV2. The simulated energy is 18 GeV electrons on 275 GeV protons. Contact person: Matthew Kelsey (

Zip file:

Within the file, all plots can be found in /AccPlots/. The code in the root directory is used with pythiaeRHIC to create the trees, and in /ANA/ is the code to fill histograms with these trees. In /ANA/Plots/ is the plot.C macro used to create the final plots.

Global event shape 1-jettiness kinematics

This section shows the kinematic plots for global event shape observable 1-jettiness. The shape of 1-jettiness vastly depends on the region in Q^2 vs. x phase space. Electrons with an energy of 18 GeV on protons with an energy of 275 GeV were simulated using PYTHIA6 (pythiaeRHIC implementation) to test the most challenging scenario. Four regions are selected based on preferred Q^2 resolution, theory uncertainties, etc. In four regions selected, normalized angular distributions of the constituent particle's contribution to 1-jettiness are histogrammed.

Please contact Sookhyun Lee ( with any questions.

Tar file: kin_tau1.tar

Q2 resolution in Q2 vs. x phase space and four selected regions.
Normalized angular distributions of 'constituent's contribution to 1-jettiness'. (Relative event count density per 'Constant Log' binning in x-Q^2: A (1) , B (10^2), C (10) and D (10^2))
Region A
Region B
Region C
Region D

Detector Requirements

The EIC detector requirements identified in the Yellow Report effort are cataloged in an interactive Detector Matrix maintained by the detector working group conveners. A spreadsheet containing the initial reasonable performance characteristics can be seen here:

This section summarizes the detector parameters needed to perform the measurements studied by the Jets and Heavy Flavor working group.

Simulation Samples

A variety of event generation, fast smearing, and detector simulation frameworks were used in the studies below. Several analyses, including groomed jet substructure, 1-subjettiness, and angularity use pythiaeRHIC (Pythia6) and eic-smear. The pythiaeRHIC implementation has been seen to reproduce e+p (di)jet cross sections and profiles in both the photoproduction and electroproduction regions (see for example here, here, and here). Other jet analyses, such as the TMD and charm jet tagging studies utilize Pythia8 + DELPHES to generate and smear events. Validation plots showing Pythia8/ZEUS and Pythia8/theory jet cross section comparisons can be seen below. Finally, the heavy flavor reconstruction studies often use more advanced detector modeling packages such as eic-root or GEANT in Fun4All.

PYTHIA 8.244 Data / Simulation / Theory Comparisons)
Data / simulation (Pythia8 / ZEUS) comparison of lab-frame NC DIS inclusive jet cross section vs jet pT
Data / theory comparison comparison of lab-frame NC DIS inclusive jet cross section for EIC kinematics as a function of Q2


Momentum Resolution

The following table summarizes the track momentum resolutions found to be necessary by our working group compared to the 'default' values found in the original detector matrix.

Track Momentum Resolution
Eta Range Default Resolution (σP/P)% Requested (σP/P)%
-3.5 < η < -2.5 0.1%*P + 0.5% Same
-2.5 < η < -2.0 0.1%*P + 0.5% Same
-2.0 < η < -1.0 0.05%*P + 0.5% Same
-1.0 < η < 1.0 0.05%*P + 0.5% Same
1.0 < η < 2.5 0.05%*P + 1.0% Same
2.5 < η < 3.5 0.1%*P + 2.0% Same

Physics Drivers

Charged particle momentum resolution is fundamentally important to all analyses considered here, from jets to heavy flavor reconstruction and precision event shape measurements. Roughly two-thirds of the energy present in a jet is composed of charged hadrons, and because tracking resolution is superior to hadron calorimeter resolution over a wide range of the EIC acceptance, the tracking resolution will set the lower bound for achievable jet energy resolution. Most studies performed by our group assumed the track momentum resolutions listed in the original detector matrix, or values very similar, and found them to be sufficient. Thus, we conclude the parameters listed above are adequate for our needs.

Additional Considerations

While the momentum resolutions listed in the detector matrix were found to be adequate, there are several points not directly addressed in the matrix we would like clarified:

  • What is the assumed B-field for these resolutions?
  • Closely related to the above, what is the minimum transverse momentum a particle can have and still be tracked? This will of course affect acceptances for soft decay products from heavy flavor decays and play a role in how well measured jet substructure and global event shapes can reproduce the particle level equivalents. The soft pion from D*-meson decays is of particular relevance; see Matt Kelsey's study here.
  • What is the expected tracking efficiency (as a function of transverse momentum)? Several efficiency scenarios were looked at in the 1-jettiness analysis (see here), but for the most part, were not considered in our analyses.
  • At what momentum do the resolution parameterizations listed in the table break down? This is especially relevant in the forward region where particle momenta can become quite large. The figure below (see this talk for more details) shows the expected performance of a silicon tracker as implemented in the Fun4All framework. It is seen that momentum resolution rapidly deteriorates for high momentum tracks in the η > ~2.5 to 3.0 region. Tracking performance here will have implications on the required hadron calorimetry energy resolution in this region.
Si Tracker Momentum Resolution (π-)
Negative pion momentum resolution as a function of eta for different momentum ranges and B-fields

Vertex Resolution

The following table summarizes the vertex resolutions found to be necessary by our working group compared to the 'default' values found in the original detector matrix.

Vertex Resolution
Eta Range Default Resolution Requested Resolution
-3.5 < η < -3.0 TBD N/A
-3.0 < η < -2.5 σxy = 30/pT + 40 μm
-2.5 < η < -1.0 σxy = 30/pT + 20 μm
-1.0 < η < 1.0 σxyz~20μm, σxy ~ σz ~ 20μm/pT + 5μm Same
1.0 < η < 2.5 TBD σxy = 30/pT + 20 μm
2.5 < η < 3.0 σxy = 30/pT + 40 μm
3.0 < η < 3.5 σxy = 30/pT + 60 μm

Physics Drivers

Precision vertexing is prerequisite for high statistics measurements of charmonium and bottomonium states, which will have many uses at the EIC from studies of hadronization and cold nuclear matter properties to determinations of the strangeness content of the nucleon and gluon TMDs. The requirements on 2D DCA resolution quoted above have been shown by the LANL group to allow high precision measurements of the nuclear modification ReA for open charm and bottom mesons over a wide pseudorapidity range (see figure below), which will discriminate between different models of parton energy loss and hadronization (see this paper for more details). A full technical note describing the results from the LANL group can be found here

Heavy Meson Reconstruction and ReA (√s = 63 GeV, BeAST B-Field)
Reconstructed charged D and B mesons with resulting ReA assuming an integrated luminosity of 10 fb-1
Reconstructed D0 mesons with resulting ReA assuming an integrated luminosity of 10 fb-1 in 4 pseudorapidity bins

While the parameters above will allow precision measurements of ReA, more differential measurements may benefit from higher pointing resolutions, which could allow for higher significance measurements for the same integrated luminosity. In a separate study, the LBNL group observed that decreasing the pointing resolution from 20 microns to 30 microns at a pT of 1 GeV reduced the D0 significance by 10%, while increasing the resolution to 10 microns led to 20% increase in significance. Details can be found in these slides.

The effects of vertex resolution were also studied in the context of charm jet tagging, where for the procedure used, a reduction in σxy = σz from 20 microns to 100 microns resulted in a 60% reduction of charm jet tagging efficiency. On the other hand improving the pointing resolution from 20 microns to 10 microns resulted in a 30% gain in charm jet tagging efficiency. Details can be found in this paper.


The following table summarizes the track momentum for which particle identification at the 3σ level was found to be necessary by our working group compared to the 'default' values found in the original detector matrix.

PID Momentum Coverage
Eta Range Default Momentum Coverage Requested Momentum Coverage
-3.5 < η < -1.0 ≤ 7 GeV Same
-1.0 < η < 0.0 ≤ 5 GeV ≤ 10 GeV
0.0 < η < 0.5
0.5 < η < 1.0 ≤ 15 GeV
1.0 < η < 1.5 ≤ 8 GeV ≤ 30 GeV
1.5 < η < 2.0 ≤ 50 GeV
2.0 < η < 2.5 ≤ 20 GeV
2.5 < η < 3.0 ≤ 30 GeV
3.0 < η < 3.5 ≤ 45 GeV Can tolerate ≤ ~20 GeV

Physics Drivers

The analyses with the most demanding PID requirements are the hadron-in-jet fragmentation measurements and the Collins asymmetry measurements. The figures below show the pseudorapidity vs momentum of charged particles within jets for electron x hadron beam energy combinations of 18x275 GeV and 10x100 GeV.

Q2 > 16 GeV2; 0.05 ≤ y ≤ 0.95; Anti-kT; R = 1.0; Jet pT > 5 GeV
Pseudorapidity vs momentum of charged particles within jets for 10x100 GeV beam energies
Pseudorapidity vs momentum of charged particles within jets for 18x275 GeV beam energies

It is evident from the above plots that in order to sample even moderate z values (z = pH/pJet) for identified hadrons within a jet, PID up to much higher values than initially listed in the detector matrix will be necessary. This is made more explicit in the following figures which show the population of pions in jT - z space assuming our group's requested PID coverage as well as the ratio matrix PID / requested PID. It is immediately evident how much the identified pion phase space is restricted with the default PID parameters.

Phasespace for identified pions inside jets
jT vs z distribution of pions from jets which could be identified given the requested PID coverage
Ratio of yields of pions which could be identified using the default PID coverage to those which could be identified using the requested coverage

Another driver of PID capabilities is the identified hadron-in-jet Collins asymmetry at moderate-to-high x and Q2. The three sets of figures below show how the ability to measure the Collins asymmetry as a function of z in three x bins is affected by varying the PID coverage. The original detector matrix coverage excluded a large fraction of the z range for moderate to high x, which is largely restored when using the requested PID performance. It should be noted that requirements on kaon tagging will be even more stringent as both the kaon yield and asymmetry will be smaller than those of charged pions. Please see this presentation and this paper for more details.

Effect of PID on Collins Asymmetry Measurement
Collins asymmetry reach assuming perfect PID
Collins asymmetry reach assuming PID ranges currently in detector matrix
Collins asymmetry reach assuming PID requested PID ranges

PID is also a crucial component of heavy flavor reconstruction where, for example, open charm mesons are tagged using the presence of a kaon. It has also been shown that the addition of PID can moderately improve the ability to tag charm jets. (See also this paper). The PID requirements outlined above have been found to be adequate for the heavy flavor program.

Additional Considerations

PID has also been shown to be an important capability in the precision measurement of global event shapes such as 1-jettiness, as the distortion caused by the assumption that all charged tracks have a pion mass can become sizable in certain regions of x-Q2 phase space. In addition, tracking and PID coverage are seen to be important as extending both from η < 3.5 to η < 4.0 substantially improves the degree to which the reconstructed event shape matches the particle level event shape. Please see slides here and here for more details.

As tracking and PID performance will be very challenging for pseudorapidities greater than 3.5 and some details of the 1-jettiness analysis are still being investigated, we do not include requirements on tracking and PID in this region in our official request, but just catalog them here for potential future discussions on performance or complementarity.


Electromagnetic Calorimetry

The following table summarizes the Electromagnetic calorimeter energy resolutions found to be necessary by our working group compared to the 'default' values found in the original detector matrix.

EMCal Energy Resolution
Eta Range Default Resolution (σE/E) Requested (σE/E)
-4.5 < η < -2.5 2%/√E Same (1-3% constant term acceptable)
-2.5 < η < -2.0 2%/√E Same (1-3% constant term acceptable)
-2.0 < η < -1.5 7%/√E Same (1-3% constant term acceptable)
-1.5 < η < -1.0 7%/√E Same (1-3% constant term acceptable)
-1.0 < η < 4.5 10-12%/√E Same (1-3% constant term acceptable)

Physics Drivers

The primary role of the ECal for our group is to measure the electromagnetic component of the jet. As such, the energy resolution will have a direct effect on the overall jet energy resolution. Many of the jet studies carried out, in both the eic-smear and Delphes frameworks, used the energy resolutions listed above (or very similar) and found acceptable jet energy resolutions. We therefore conclude that the default values are sufficient for our purposes. We do note that the default values did not include constant terms, which is unrealistic, but we do not foresee the addition of small, realistic constant terms of between 1 and 3% changing our basic conclusions.

Additional Considerations

While the ECal energy resolutions were found to be sufficient, there are several performance / design considerations that were not addressed in the detector matrix for which we would like some clarification or insight.

  • Minimum energy threshold: Several analyses assumed energy thresholds of 100 to 200 MeV for a particle to be 'detected' by the ECal, but it would be good to have some official guidance. The thresholds used were generally found to be sufficient. These thresholds will affect how much energy from the hadronizing parton is collected in the jet and therefore the jet energy scale, they will also be relevant for the accuracy of missing transverse energy measurements.
  • Pointing resolution and cluster separation: Observables such as jet substructure and global event shapes depend on both the energy and position of the particles in an event. Information on how well the positions of particles can be resolved by the ECal will be useful for evaluating the accuracy of such measurements.
  • Non-hermiticity: We understand that gaps in calorimeter coverage (likely around the barrel - endcap interface) may be necessary in order to provide services to the inner detectors or due to interface issues between barrel and endcap detector components. As these gaps will distort jet energy scales and resolutions as well as missing energy measurements and hadronic kinematic reconstruction techniques such as Jacquet-Blondel, we encourage detector designs which minimize the extent of such holes in coverage. A quick study showing the effect of two gap sizes (for both ECal and HCal) on reconstructed vs particle jet properties can be found here.
  • Large η coverage: We wonder if the ECal coverage to η of 4.5 is compatible with current beam pipe designs. Regardless of the actual proposed η coverage, we request that it be the same as for the ECal.

Hadron Calorimetry

The following table summarizes the Hadron calorimeter energy resolutions found to be necessary by our working group compared to the 'default' values found in the original detector matrix.

HCal Energy Resolution
Eta Range Default Resolution (σE/E) Requested (σE/E)
-3.5 < η < -1.0 50%/√E Same (~10% constant term is acceptable)
-1.0 < η < 1.0 N/A 85%/√E + 10%
1.0 < η < 3.0 50%/√E 50%/√E + 10%
3.0 < η < 3.5 50%/√E + 5%
3.5 < η < 4.0 N/A

Physics Drivers

As with the electromagnetic calorimeters, the primary purpose of the hadron calorimeters for our group is to measure the energy carried by hadrons. However, because of the generally low momentum / energy of particles produced at an EIC, the tracking resolution will be much better than the HCal energy resolution for all but the highest momentum particles and jets, which generally lie in the very forward (hadron going) region. Thus, the primary contribution from the HCal for most of the EIC detector acceptance will be measuring neutral hadrons such as neutrons and K0Ls.

The η < -1.0 region is does not contain a large number of jets with significant transverse momentum or energy and the original detector matrix resolution of 50%/√E (plus a constant term of ~10%) should be adequate for our needs.

While the original detector matrix did not contain a specification for an HCal in the barrel region, the jets and heavy flavor group request that one be included. A study of missing transverse energy (MET) for tagging charged current DIS events was carried out in the Delphes framework and decent MET resolution was seen even assuming a relatively poor HCal resolution of 100%/√E + 10% (see this paper for more details). It was seen, however, that the lack of a barrel HCal led to large asymmetric tails in the MET distribution that would complicate unfolding procedures and reduce photoproduction and NC DIS background rejection.

A barrel HCal with resolution 85%/√E + 7% was also considered in the eic-smear framework for photoproduction (Q2 < 1 GeV2) jets. When selecting smeared jets and comparing them to their matched particle level counterparts, a large bias in average transverse momentum / energy difference is seen as one moves to lower pseudorapidities, while no such bias is seen if the HCal information is not used in the jet reconstruction (see left hand panel of the figure below). This effect arises because a certain fraction of the jets in a smeared pT or energy bin actually arose from a particle level jet with much lower pT or energy which contained a neutral hadron whose energy was smeared to a much higher value, and because of the steeply falling spectra, even small smearing at low pT or energy can contribute to higher smeared bins. This effect becomes more pronounced at lower eta where jets have smaller energies. This bias can be reduced by improving the resolution of the barrel HCal as shown in the middle panel below. However, a more effective method would be to select only those jets which do not contain a neutral hadron, and thus do not suffer from the large energy distortion, by using the HCal as a neutral hadron veto. This is illustrated in the right panel of the below figure where the red curve, which is for jets which do not contain a neutral hadron, shows no bias and much better resolution. Despite the bias, we request a barrel HCal resolution of 85%/√E + 10% with the expectation that the resolution will be somewhat better when combined with the ECal and/or a hadron veto method can be implemented.

Bias between smeared and particle level jets vs pseudorapidity
(Smeared - Particle)/Particle Jet pT vs eta for all detector components (blue) and no HCal (green). Error bars are RMS of distribution.
Same as left panel but now for three different HCal resolutions.
Same as left panel but now showing offset and resolution for subsample of smeared jets which do not contain a neutral hadron (red).

An HCal resolution of 50%/√E (plus a 10% constant term) was found to be adequate in the region 1.0 < η < 3.0 where tracking performance is generally good and jet energies are relatively low. For η > 3, we argue that a constant term of roughly 5% is needed as jet energies rapidly increase in this region while tracking resolution significantly degrades, enhancing the importance of the HCal energy resolution, which is dominated by the constant term at these energies (see figures below). See also this presentation for more details.

Relationship between jet energy, pseudorapidity, and HCal performance
Jet energy spectrum along with the average pseudorapidity in each energy bin
Comparison of resolution vs energy for various stochastic and constant terms. The constant term dominates the resolution budget at high energy

We also request that the HCal acceptance be extended from η < 3.5 to η < 4.0. This will provide access to the highest bjorken x regions as seen in the figures below. As tracking will be absent in for η > 3.5, good HCal resolution will be imperative for maintaining good overall jet energy resolution. This will be necessary to allow differential TMD measurements with jets such as the electron-jet Sivers asymmetry in the valence region and mid to high Q2. The phasespace gained with HCAL up to 4.0 is crucial given that existing DIS data with transversely-polarized targets stops at about x=0.3.

x-Q2 phasespace accessible for different jet pseudorapidity
Phasespace accessible for jet η < 3.0 (R = 0.5, HCal coverage up to η = 3.5)
Phasespace accessible for jet η < 3.5 (R = 0.5, HCal coverage up to η = 4.0)

Additional Considerations

  • Position resolution: As discussed above, the ability to identify jets which contain neutral hadrons can greatly improve jet energy resolution in certain regions of phasespace. In order to work as a neutral hadron veto, the HCal will need to have a reasonable ability to separate energy clusters and match them to incoming charged hadrons. While detailed investigations of this capability will need full GEANT simulations, it should be kept in mind when considering HCal parameters such as granularity or the inclusion of a highly segmented preshower.
  • Linearity: As the HCal systems will see a wide variety of particle energies, especially in the forward region, the linearity of response will be important to understand.
  • ECal + HCal integration: The resolutions discussed here do not take into account being able to utilize both ECal and HCal information which of course any realistic calorimeter system will do.

Physics Analyses

This section briefly summarizes the analyses performed by members of the Jets and Heavy flavor working group. Analysis descriptions, points of contact, and links to additional material are provided.

The indico page containing all presentations made to the working group can be found here. Relevant presentations were also given at the Temple and Pavia meetings.

TMDs with Jets

The study of the 3D structure of the nucleon via Transverse Momentum Dependent parton distributions and fragmentation functions (TMD PDFs and FFs) is an important program for the EIC. The use of jets to study these objects will provide an important compliment to studies carried out with single hadron SIDIS measurements.

Point of Contact: Miguel Arratia (


Heavy Flavor Reconstruction

LANL Group

Point of Contact: Xuan Li (


LBNL Group

Point of contact: Xin Dong (

EICUG-YR Workshop presentations:

EICUG-YR Jets and Heavy Quarks WG presentations:

Strange PDF via Charm Jet Tagging

Points of Contact:


(Groomed) Jet Substructure

Points of Contact: Joe Osborn () and Dillon Fitzgerald ()


Photoproduction Jets and Jet Angularity

Point of Contact: Brian Page (


Global Event Shapes


Transverse Energy-Energy Correlators