Research

ALICE Run 3 jets, photon-tagged observables, calorimetry.

ALICE Run 3 inclusive charged-particle jet measurements, PWG-JE tracking-efficiency systematics, ePIC Barrel Imaging Calorimeter R&D, and central-framework engineering in ALICE O²Physics. Forward program centered on bias-robust radial and photon-tagged observables in O+O and Pb+Pb.

§1 Doctoral analysis · ALICE Run 3 charged-particle jets

Full analysis chain, pp 13.6 TeV.

I am the primary analyzer of the first ALICE Run 3 inclusive charged-particle jet cross-section measurement, R-dependent charged-particle jet production in pp at √s = 13.6 TeV with ALICE (in preparation). The analysis spans the full chain — detector-level QA on continuous-readout data, jet reconstruction in O²Physics, response-matrix construction, Bayesian and SVD unfolding, and a complete systematic-uncertainty program covering tracking effects and unfolding stability. Preliminary results are PWG-JE-approved; presented on behalf of ALICE at EPS-HEP 2025 as the PWG-JE merge talk.

Varying the jet radius R from 0.2 to 0.6 probes the balance between collinear and large-angle emissions where parton-shower tunes disagree, and provides the pp baseline for the 2025 O+O and Ne+Ne running.

I implemented the Run 3 charged-jet cross-section normalization from the ALICE TVX visible cross section under continuous readout, now becoming the standard normalization for Run 3 charged-jet analyses in the group.

§2 Tracking-efficiency systematics · Hard Probes 2026

pp tracking-efficiency systematics at 13.6 and 5.36 TeV.

I derived the pp charged-particle tracking-efficiency uncertainty at √s = 13.6 and 5.36 TeV (p_T-binned, 2022–2024 periods) for the PWG-JE Hard Probes 2026 preparation, combining the track-selection component with the ITS–TPC matching component.

I also studied bunch-crossing and track-matching inefficiencies arising from continuous TPC readout and time-frame-based asynchronous reconstruction, together with the corresponding signal-loss corrections, and track-p_T resolution (covariance-matrix propagation, MC residuals, V0 mass resolution), feeding the p_T-dependent momentum-resolution smearing used for Run 3 charged-jet systematics.

§3 Detector R&D · ePIC Barrel Imaging Calorimeter

Pb/scintillating-fiber prototype, three test-beam campaigns.

CERN PS T10 (July 2025) — lead analyzer for the campaign: energy response, resolution, and linearity. KEK AF–AR (March 2025) — on-site DAQ shifter, equalization, prompt analysis. CERN PS T10 (August 2024) — first campaign; module performance characterization, published as a co-author: submitted to Nucl. Instrum. Meth. A (NIMA-D-26-00482), arXiv:2604.22647.

Earlier hardware work: I ran the PMT–module QA for the prototype modules prior to the test beams — signal characterization with cosmic-ray and radioactive-source data, timing-coincidence checks, optical-coupling stability, applied per-module acceptance criteria. On the simulation/reconstruction side I contribute energy- and η-dependent sampling-fraction studies in JANA2 / EIC-recon and Geant4 prototype modeling for the Korea-BIC software effort.

§4 Software & infrastructure · ALICE O²Physics

O²Physics framework contributions.

Contributor since 2023 to AliceO2Group/O2Physics: jet-finding and response-matrix tasks, systematic-uncertainty pipelines, particle-level cross-section normalization, and shared track-smearing tools.

§5 Mentoring & community

Mentoring and tutorials.

Co-mentor of the MSc main analyzer on the ALICE Run 3 O+O charged-jet measurement (SKKU, 2024–present): O²Physics onboarding, jet reconstruction methodology, response-matrix workflow, and systematic-uncertainty estimation; co-author with W. Ham on the corresponding PWG-JE Analysis Note.
Co-organizer and instructor, KoALICE O²Physics Tutorial (SKKU, January 2024; with H. Lee) — two-day Run 3 analysis-framework tutorial for the Korean ALICE community (~22 participants from SKKU, Inha University, and other Korean institutes); instructor on Running O²Physics with Hyperloop (solo) and on Practical PWG-JE tasks.

§6 Current study · Validating learned reconstruction

Calibration transfer on public benchmarks.

A calibration-transfer study on the public sPHENIX TPCpp-10M spacepoint benchmark and the released FM4NPP foundation-model backbones (4.9M–175M parameters), plus companion measurements on paired Pythia/Herwig quark–gluon jet samples. Classifiers are evaluated at fixed working points under a signal-loss budget, differentially in occupancy, with event mixing reaching Au+Au-like densities.

A working point calibrated on p+p is exceeded by a factor of ~30 at central-Au+Au-like density, while aggregate metrics change little. Across the released backbone ladder, discrimination improves with model size while high-density background rejection does not. A head built on 21 truth-free features doubles high-density rejection; fine-tuning does not. Score migration runs in opposite directions under density shift and generator shift. Data and model checkpoints are public.

§7 What I am thinking about

Open questions.

Whether jet quenching is present in small systems, which the 2025 O+O and Ne+Ne data can test.
Radial, bias-robust substructure (jet shape ρ(r), fragmentation function D(z)) as observables that make selection bias directly testable.
Photon-tagged jets as a cleaner recoil probe — γ-tagged ρ(r) in small systems and γ+jet Δφ acoplanarity / γ+jet correlations in Pb+Pb for medium-induced k_T broadening and the diffusion-wake search.
ML methods (OmniFold-style unfolding) that preserve correlations and control bias when extracting weak high-dimensional signals.
Detector R&D and performance characterization through test-beam data and simulation, including the ePIC Barrel Imaging Calorimeter.

Publications, talks, and the CV.

Full publication record on INSPIRE-HEP; talks history at Talks; CV downloads at CV.

Publications ↗ Talks ↗ CV ↗ jbae@cern.ch