Samantha Abbott
UC Davis
https://lpc.fnal.gov/programs/graduate-scholars/2025/Samantha_Abbott.shtml
At the LPC, I am contributing to developing the upgrade from file-based storage to a more flexible and scalable object-based storage model using S3 (Simple Storage Service) technology for the CMS experiment. The Ceph S3 object storage model breaks down event data into discrete, independent objects, reducing storage requirements and streamlining data processing. Adopting this model will enable CMS to handle HL-LHC data in a way that is scalable, flexible, and cost-effective. Additionally, I will assist in integrating other CMSSW data handling upgrades with the object-based storage model: First, the migration away from TTree to RNTuple, a new I/O format for ROOT, provides increased granularity that perfectly complements the object storage model. Second, the potential shift towards a Structure of Arrays (SoA) format, which generally offers much better efficiency for vectorized CPUs and GPUs, optimized with the object-based data storage model would further improve data retrieval and storage efficiency for large, complex datasets.
Finally, I will continue to develop the Boosted Event Shape Tagger (BEST), the focus of my thesis. BEST is a multi-object classifier that uses machine learning to tag jets and distinguish signal particles from QCD background. BEST analyzes jet substructure by calculating known observables from the jet constituents, which are boosted into seven different hypothetical rest frames. I am now working on upgrading the neural network architecture in preparation for a future Run 3 analysis by adding a layer that understands physics operations, such as Lorentz Boosts.
I joined CMS when I began my PhD at UC Davis over the summer of 2019. When I was based at CERN from 2021-2022, I contributed to CMS hardware upgrades by working on GEMs (Gas Election Multipliers), a Muon subdetector. My primary focus was assembly and commissioning of the GE 2/1 Demonstrator, a single chamber test-run for a new layer of GEMs aimed to be fully installed by Run 4. I also contributed to operations through regular GEM and CSC DOC shifts leading up to and during Run 3. After returning home, when not mentoring new students at UC Davis, I have taken several online DQM shifts on site at the FNAL ROC and participate in the CMS Data Analysis School at the FNAL LPC. I look forward to many more opportunities to contribute to the collaborative environment at the LPC!