publications & presentations

2020

1. Simulation-Assisted Decorrelation for Resonant Anomaly Detection

   Le Pottier, Luc

   NeurIPS MP4PS Workshop Accepted Papter & Poster Presentation

   dec 2020

   Abs HTML PDF

   A growing number of weak- and unsupervised machine learning approaches to anomaly detection are being proposed to significantly extend the search program at the Large Hadron Collider and elsewhere. One of the prototypical examples for these methods is the search for resonant new physics, where a bump hunt can be performed in an invariant mass spectrum. A significant challenge to methods that rely entirely on data is that they are susceptible to sculpting artificial bumps from the dependence of the machine learning classifier on the invariant mass. We explore two solutions to this challenge by minimally incorporating simulation into the learning. In particular, we study the robustness of Simulation Assisted Likelihood-free Anomaly Detection (SALAD), one of our recently developed anomaly detection methods, to correlations between the classifier and the invariant mass. Next, we propose a new approach that only uses the simulation for decorrelation but the Classification without Labels (CWoLa) approach for achieving signal sensitivity.

2. Simulation-Assisted Decorrelation for Resonant Anomaly Detection

   Benkendorfer, Kees, Le Pottier, Luc, and Nachman, Benjamin

   arXiv:2009.02205 [hep-ex, physics:hep-ph, physics:physics, stat]

   sep 2020

   Abs arXiv

   A growing number of weak- and unsupervised machine learning approaches to anomaly detection are being proposed to significantly extend the search program at the Large Hadron Collider and elsewhere. One of the prototypical examples for these methods is the search for resonant new physics, where a bump hunt can be performed in an invariant mass spectrum. A significant challenge to methods that rely entirely on data is that they are susceptible to sculpting artificial bumps from the dependence of the machine learning classifier on the invariant mass. We explore two solutions to this challenge by minimally incorporating simulation into the learning. In particular, we study the robustness of Simulation Assisted Likelihood-free Anomaly Detection (SALAD) to correlations between the classifier and the invariant mass. Next, we propose a new approach that only uses the simulation for decorrelation but the Classification without Labels (CWoLa) approach for achieving signal sensitivity. Both methods are compared using a full background fit analysis on simulated data from the LHC Olympics and are robust to correlations in the data.

3. Machine Learning for Physics Undergraduates: Introductory Workshop

   Le Pottier, Luc

   University of Michigan SPS Physics Workshop Series

   jan 2020

   Abs Code

   This workshop aims to provide a 2 day,  4 hour introduction to advanced python programming for physicists, graduate students, and Undergraduates. The workshop covers some of the most useful advanced data processing, visualization, and storage methods for ML in a physics context on the first day. On the second day, it walks through the theoretical and motivational bases of machine learning in physics, culminating in a step-by-step walkthrough in which participants build a basic neural network in Keras and apply it to a trivial particle physics example. Implemented via Jupyter Notebooks in Google Colab, this workshop was funded by the UM Physics department.

4. Unsupervised Machine Learning for Rare Signal Detection in CMS Detector Data

   Le Pottier, Luc

   APS March Meeting Poster Presentation

   mar 2020

   Abs HTML PDF Code

   We present here a method for identifying high-pt beyond the Standard Model (BSM) signals, with no dependence on the expected physics signatures of these models. We do this by constraining our detector data to the expected signal region of a given model and training a deep neural autoencoder to ’recognize’ jets belonging to the Standard Model (SM) background data. In this way, our models learn the background signature of SM jets in our signal region. We then evaluate this model on a mixture of SM and BSM jets, flagging jets with a high autoencoder reconstruction error as ’anomalous’ or signal jets. This method of searching independently of any physics model is especially useful when the BSM model in question involves particles of unknown mass, branching fraction, etc. This is especially relevant in Dark Matter searches, such as the search for Semi-Visible jets in missing-ET events at the LHC. In these cases, our models are capable of flagging a range of BSM models with different parameters within our signal region, as opposed to a supervised model (such as a Boosted Decision Tree) which is only able to flag the exact signal it was trained on. We show that this technique is a promising method of identifying arbitrary BSM signals to high precision in LHC data.

5. A liquid hydrogen target for the MUSE experiment at PSI

   Roy, P., Corsetti, S., Dimond, M., Kim, M., Le Pottier, L., Lorenzon, W., Raymond, R., Reid, H., Steinberg, N., Wuerfel, N., Deiters, K., Briscoe, W.J., Golossanov, A., and Rostomyan, T.

   Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment

   jan 2020

   Abs arXiv HTML

   A 280 ml liquid hydrogen target has been constructed and tested for the MUSE experiment at PSI to investigate the proton charge radius via simultaneous measurement of elastic muon-proton and elastic electron-proton scattering. To control systematic uncertainties at a sub-percent level, strong constraints were put on the amount of material surrounding the target and on its temperature stability. The target cell wall is made of 120μm-thick Kapton, while the beam entrance and exit windows are made of 125μm-thick aluminized Kapton. The side exit windows are made of Mylar laminated on aramid fabric with an areal density of 368g/m2. The target system was successfully operated during a commissioning run at PSI at the end of 2018. The target temperature was stable at the 0.01 K level. This suggests a density stability at the 0.02% level, which is about a factor of ten better than required.

2017

1. LH_2 Target Design Position Survey Techniques for the MUSE Experiment for Precise Proton Radius Measurement

   Le Pottier, Luc

   APS Division of Nuclear Physics Poster Presentation

   oct 2017

   Abs HTML PDF Code

   The proton radius puzzle is a currently unresolved problem which has intrigued the scientific community, dealing with a 7 σ discrepancy between the proton radii determined from muonic hydrogen spectroscopy and electron scattering measurements. The MUon Scattering Experiment (MUSE) aims to resolve this puzzle by performing the first simultaneous elastic scattering measurements of both electrons and muons on the proton, which will allow the comparison of the radii from the two interactions with reduced systematic uncertainties. The data from this experiment is expected to provide the best test of lepton universality to date. The experiment will take place at the Paul Scherrer Institute in Switzerland in 2018. An essential component of the experiment is a liquid hydrogen (LH2) cryotarget system. Our group at the University of Michigan is responsible for the design, fabrication and installation of this system. Here we present our LH2 target cell design and fabrication techniques for successful operation at 20 K and 1 atm, and our computer vision-based target position survey system which will determine the position of the target, installed inside a vacuum chamber, with 0.01 mm or better precision at the height of the liquid hydrogen target and along the beam direction during the experiment.

2. Cafe Shapiro Anthology, Selected Poems and Short Stories from the 20th Annual Cafe Shapiro

   Alberta, C., Arrington, S., Aulepp, A., Ayalp, E., Baker, A., Beglin, R., Campbell, A., Cantuba, D., Cardenas, M., Chu, A., Duan, K., Dzubay, L., Esquivel, A.V., Fall, S., Gurm, Z., Hao, P., Hawthorne, T., Heng, J.A., James, G., Janeschek, K., Jaracz, S., Le Pottier, L., Lu, L., Mandilk, J., McLinden, P., Morris, G., Mota, N., Nair, B., Orwig, K., Ramirez-Gorski, E., Rao, S., Rebera, K., Robins-Somerville, M., Staten, M., Steers, N., Veasey, E., Wilson, E., Wood, C., Wybrecht, M., Yang, S., and Zielinski, A.

   University of Michigan Undergraduate Library Publications

   feb 2017

   HTML PDF