EPSCI

Group Meeting�

August 18, 2026

1

Coupling Experiment to Accelerator Controls

2

CEAC-Moller

1. ODU collaborators starting to discuss control algorithms based on tree-based models trained on CRex data
2. Weekly meeting is being rescheduled due to faculty and student time constraints

TwinD

• Fixed some bugs in the two person teams
• Began iterating to get the harp scan working
• Got claude to help make an autojudging script to begin quantifying trials
• David mentioned a possible desire to “harden” the agents. Have begun thinking about that…

CEAC-GlueX

• Validation of GNN for FCAL cluster classification (neutral vs hadronic showers) in progress
• ECAL/FCAL calibrations integrated into SMOCS
• Possible JANA2 producer integration (need to strategize with Nathan Brei)
• David updated the data challenge roadmap

EPSCI Group Meeting August 18, 2026

EIC Computing and Software

3

Validation

1. See ePIC integration notes on Hydra slide
2. New meeting time needed to incorporate co-conveners schedule
3. Stalled effort on validating tracking-level campaign plots because everyone is so busy

Accomplishments

• one …
• two …
• three …

Reconstruction

• Timeframe splitter PR with Takuya
• Two rounds of reviews and responses to comments
• Prepared a talk at ePIC workshop at ODU (Wed–Fri this week)
• “JANA2 Updates: Hierarchical vs Batched Events”
• Contributed to a paper for NIM-A
• “ePIC Software and Simulations”
• Sections on framework, timeframe splitting, and future work (e.g. heterogeneous hardware offloading)
• Resumed work on BatchedArrow

EPSCI Group Meeting August 18, 2026

Hydra

4

Accomplishments

1. one …
2. two …
3. three …

OPS

• Beam off early due to a compressor failure.
• GlueX and Hall C online operations have stopped.
• ePIC, GSI, GlueX and HallC offline operations are ongoing
• VM request ticket is in to get a VM to run ePIC’s deployment
• Hydra summary stats provided for David L’s lab update

Development

• CILogon almost done and ready
• Deployed for ePIC
• Restful api to allow ePIC production to integrate hydra
• Notify of data and readback results
• A number of merge requests in the queue due to CI/CD issues

EPSCI Group Meeting August 18, 2026

DAQ: Streaming Readout + CODA

5

CODA + ERSAP

1. Began drafting the design requirements for a Virtual Crate CODA component
2. Addressing new CODA requirements from Hall D
3. Resuming work on the extended ERSAP paper
4. Nathan Baltzell’s request to move EVIO’s java to Java 21

Streaming Readout

• Initiated a GEMC digitization project to simulate VTP output for the CLAS12 detector.
• Writing the CHEP paper.
• Preparing the SC26 poster and accompanying paper: � submitted Aug 8
• Had several chats on shift w/ Hall D and EIC groups about SRO@Hall D
• Test platform (PS) supports SRO, in use
• GlueX Inner ECAL: VTP/fADCs should support SRO(?)
• LDRD (clas sro trigger): no update yet
• News on JLab ejfat LBs?

EPSCI Group Meeting August 18, 2026

Genesis Mission / American Science Cloud

6

RFA

1. Both of the RFAs Thomas is on have begun to move forward
• Infrastructure, SOWs etc
2. Torri unofficially advising on tasks, milestones, SOWs, and data sharing plans for polarized target RFA

Discovery Engine

• Thomas drafted some content on how to access AmSC and MAG and has submitted a PR to include it in the knowledgebase

HAIDIS

1. Implemented ERSAP observability support for the training set preparation workflow
2. Workings on paper in progress

Nathan: AmSC-SUF

• Attempted to reproduce EJFAT streaming from BNL IC2 for a SC paper
• BNL->JLab, BNL->Perlmutter both failed
• Quickly tested JLab EJFAT locally (JLab->JLab), worried that it might not be working at all right now

​

Nathan: MOSAIC

1. Brainstormed ways to substantially contribute, received positive feedback from Torre on all counts

EPSCI Group Meeting August 18, 2026

7

LD2606 and AmSC Data Services - Data Movement

Aug/5 – Aug/18 Accomplishments

1. IPDPS’27 submission (ddl: Oct 1, 10-page excluding References)
• GPU variability (ongoing) - Internship & Me
• 8-page draft (Background, Methodology, Microbenchmarking and initial AI training results).
• Proof-write to redo the logic and confirm the findings.
• LDRD paper (initiated).
2. LDRD
• DAOS (TODO) - Not working since Aug. Hardware failure?
• DPU testbed (Ongoing)
• DOCA v3.4 flashed to BF-3 and BF-2 but failed to SSH to them. BF-3’s default mode is NIC.
• No IP for either DPU. 100+ Gbps Ethernet link not installed?
• Raiqa: end-to-end data collection and visualization workflows.
3. SC26 workshop
• PDSW submitted on Aug/7. AD submitted on Aug/12.
• This workshop has 25 full-paper submissions in total and it might only accept < 10.
• W&M professor explored HPC-Quantum I/O problem.
4. QCD-ZGEMM benchmarking (for Amitoj)
• Ongoing - Meetings AMD (predict MI430 perf with ROCm profiling on MI350), NVIDIA and Google (using BF16 hardware doing FP64 computing)

EPSCI Group Meeting August 18, 2026

JANA2

8

LD2613: Distributed workflow steering

1. Jon Zarling site visit!�
2. Refactored Conductor to use Reactor/Listener abstractions�
3. Ported the ERSAP BaseOrchestrator to C++, working on the CoreOrchestrator
• Base <: Core <: Abstract <: Generic
• Adventures leveraging Claude for this�
4. Ongoing: Running the Conductor from a VM, communicating with MCP and code.jlab.org

JANA2

• Issue: LLMs are complaining about nonexistent race conditions because they can’t figure out which JANA2 callbacks are called sequentially�
• Issue: Input<>s are accessible from JEventProcessor::ProcessParallel, which _is_ a race condition if somebody uses them

EPSCI Group Meeting August 18, 2026

Miscellaneous

9

Accomplishments

• one …
• two …
• three …

MCwrapper

• Chewing through a backlog caused by outage
• Discovered a bug (URL not updated) in the csh version (and fixed it)
• ​

SPINDLE

• ERSAP + JANA2 integration discussions
• No response yet from lab leadership about status of LDRD

AIOP-Target

• Monibor (data science) is finishing up his portions of edits to the paper draft this week

EPSCI Group Meeting August 18, 2026