arXiv subjects

Jesse Thaler

Publications and source records attributed to Jesse Thaler.

At least 73 records · Page 4Linked to original sources

arXiv2021

Scaffolding Simulations with Deep Learning for High-dimensional Deconvolution

A common setting for scientific inference is the ability to sample from a high-fidelity forward model (simulation) without having an explicit probability density of the data. We propose a simulation-based maximum likelihood deconvolution approach in this setting called OmniFold. Deep learning enables this approach to be naturally unbinned and (variable-, and) high-dimensional. In contrast to model parameter estimation, the goal of deconvolution is to remove detector distortions in order to enable a variety of down-stream inference tasks. Our approach is the deep learning generalization of the common Richardson-Lucy approach that is also called Iterative Bayesian Unfolding in particle physics. We show how OmniFold can not only remove detector distortions, but it can also account for noise processes and acceptance effects.

stat.ML↗

arXiv2021

Mapping Machine-Learned Physics into a Human-Readable Space

We present a technique for translating a black-box machine-learned classifier operating on a high-dimensional input space into a small set of human-interpretable observables that can be combined to make the same classification decisions. We iteratively select these observables from a large space of high-level discriminants by finding those with the highest decision similarity relative to the black box, quantified via a metric we introduce that evaluates the relative ordering of pairs of inputs. Successive iterations focus only on the subset of input pairs that are misordered by the current set of observables. This method enables simplification of the machine-learning strategy, interpretation of the results in terms of well-understood physical concepts, validation of the physical model, and the potential for new insights into the nature of the problem itself. As a demonstration, we apply our approach to the benchmark task of jet classification in collider physics, where a convolutional neural network acting on calorimeter jet images outperforms a set of six well-known jet substructure observables. Our method maps the convolutional neural network into a set of observables called energy flow polynomials, and it closes the performance gap by identifying a class of observables with an interesting physical interpretation that has been previously overlooked in the jet substructure literature.

hep-ph↗

arXiv2021

The LHC Olympics 2020: A Community Challenge for Anomaly Detection in High Energy Physics

A new paradigm for data-driven, model-agnostic new physics searches at colliders is emerging, and aims to leverage recent breakthroughs in anomaly detection and machine learning. In order to develop and benchmark new anomaly detection methods within this framework, it is essential to have standard datasets. To this end, we have created the LHC Olympics 2020, a community challenge accompanied by a set of simulated collider events. Participants in these Olympics have developed their methods using an R&D dataset and then tested them on black boxes: datasets with an unknown anomaly (or not). This paper will review the LHC Olympics 2020 challenge, including an overview of the competition, a description of methods deployed in the competition, lessons learned from the experience, and implications for data analyses with future datasets as well as future colliders.

hep-ph↗

arXiv2020

A Neural Resampler for Monte Carlo Reweighting with Preserved Uncertainties

Monte Carlo event generators are an essential tool for data analysis in collider physics. To include subleading quantum corrections, these generators often need to produce negative weight events, which leads to statistical dilution of the datasets and downstream computational costs for detector simulation. Building on the recent proposal of a positive resampler method to rebalance weights within histogram bins, we introduce neural resampling: an unbinned approach to Monte Carlo reweighting based on neural networks that scales well to high-dimensional and variable-dimensional phase space. We pay particular attention to preserving the statistical properties of the event sample, such that neural resampling not only maintains the mean value of any observable but also its Monte Carlo uncertainty. This uncertainty preservation scheme is general and can also be applied to binned (non-neural network) resampling. To illustrate our neural resampling approach, we present a case study from the Large Hadron Collider of top quark pair production at next-to-leading order matched to a parton shower.

hep-ph↗

arXiv2020

Data-driven quark and gluon jet modification in heavy-ion collisions

Whether quark- and gluon-initiated jets are modified differently by the quark-gluon plasma produced in heavy-ion collisions is a long-standing question that has thus far eluded a definitive experimental answer. A crucial complication for quark-gluon discrimination in both proton-proton and heavy-ion collisions is that all measurements necessarily average over the (unknown) quark-gluon composition of a jet sample. In the heavy-ion context, the simultaneous modification of both the fractions and substructure of quark and gluon jets by the quark-gluon plasma further obscures the interpretation. Here, we demonstrate a fully data-driven method for separating quark and gluon contributions to jet observables using a statistical technique called topic modeling. Assuming that jet distributions are a mixture of underlying "quark-like" and "gluon-like" distributions, we show how to extract quark and gluon jet fractions and constituent multiplicity distributions as a function of the jet transverse momentum. This proof-of-concept study is based on proton-proton and heavy-ion collision events from the Monte Carlo event generator Jewel with statistics accessible in Run 4 of the Large Hadron Collider. These results suggest the potential for an experimental determination of quark and gluon jet modifications.

hep-ph↗

arXiv2020

A Robust Measure of Event Isotropy at Colliders

We introduce a new event shape observable -- event isotropy -- that quantifies how close the radiation pattern of a collider event is to a uniform distribution. This observable is based on a normalized version of the energy mover's distance, which is the minimum "work" needed to rearrange one radiation pattern into another of equal energy. We investigate the utility of event isotropy both at electron-positron colliders, where events are compared to a perfectly spherical radiation pattern, as well as at proton-proton colliders, where the natural comparison is to either cylindrical or ring-like patterns. Compared to traditional event shape observables like sphericity and thrust, event isotropy exhibits a larger dynamic range for high-multiplicity events. This enables event isotropy to not only distinguish between dijet and multijet processes but also separate uniform N-body phase space configurations for different values of N. As a key application of this new observable, we study its performance to characterize strongly-coupled new physics scenarios with isotropic collider signatures.

hep-ph↗

arXiv2020

The Hidden Geometry of Particle Collisions

We establish that many fundamental concepts and techniques in quantum field theory and collider physics can be naturally understood and unified through a simple new geometric language. The idea is to equip the space of collider events with a metric, from which other geometric objects can be rigorously defined. Our analysis is based on the energy mover's distance, which quantifies the "work" required to rearrange one event into another. This metric, which operates purely at the level of observable energy flow information, allows for a clarified definition of infrared and collinear safety and related concepts. A number of well-known collider observables can be exactly cast as the minimum distance between an event and various manifolds in this space. Jet definitions, such as exclusive cone and sequential recombination algorithms, can be directly derived by finding the closest few-particle approximation to the event. Several area- and constituent-based pileup mitigation strategies are naturally expressed in this formalism as well. Finally, we lift our reasoning to develop a precise distance between theories, which are treated as collections of events weighted by cross sections. In all of these various cases, a better understanding of existing methods in our geometric language suggests interesting new ideas and generalizations.

hep-ph↗

arXiv2020

Quantum Algorithms for Jet Clustering

Identifying jets formed in high-energy particle collisions requires solving optimization problems over potentially large numbers of final-state particles. In this work, we consider the possibility of using quantum computers to speed up jet clustering algorithms. Focusing on the case of electron-positron collisions, we consider a well-known event shape called thrust whose optimum corresponds to the most jet-like separating plane among a set of particles, thereby defining two hemisphere jets. We show how to formulate thrust both as a quantum annealing problem and as a Grover search problem. A key component of our analysis is the consideration of realistic models for interfacing classical data with a quantum algorithm. With a sequential computing model, we show how to speed up the well-known O(N^3) classical algorithm to an O(N^2) quantum algorithm, including the O(N) overhead of loading classical data from N final-state particles. Along the way, we also identify a way to speed up the classical algorithm to O(N^2 log N) using a sorting strategy inspired by the SISCone jet algorithm, which has no natural quantum counterpart. With a parallel computing model, we achieve O(N log N) scaling in both the classical and quantum cases. Finally, we consider the generalization of these quantum methods to other jet algorithms more closely related to those used for proton-proton collisions at the Large Hadron Collider.

hep-ph↗

arXiv2020

OmniFold: A Method to Simultaneously Unfold All Observables

Collider data must be corrected for detector effects ("unfolded") to be compared with many theoretical calculations and measurements from other experiments. Unfolding is traditionally done for individual, binned observables without including all information relevant for characterizing the detector response. We introduce OmniFold, an unfolding method that iteratively reweights a simulated dataset, using machine learning to capitalize on all available information. Our approach is unbinned, works for arbitrarily high-dimensional data, and naturally incorporates information from the full phase space. We illustrate this technique on a realistic jet substructure example from the Large Hadron Collider and compare it to standard binned unfolding methods. This new paradigm enables the simultaneous measurement of all observables, including those not yet invented at the time of the analysis.

hep-ph↗

arXiv2020

Navigating Collinear Superspace

We introduce a new set of effective field theory rules for constructing Lagrangians with $\mathcal{N} = 1$ supersymmetry in collinear superspace. In the standard superspace treatment, superfields are functions of the coordinates $(x^μ,θ^α, θ^{\dagger \dotα})$, and supersymmetry preservation is manifest at the Lagrangian level in part due to the inclusion of auxiliary $F$- and $D$-term components. By contrast, collinear superspace depends on a smaller set of coordinates $(x^μ,η,η^\dagger)$, where $η$ is a complex Grassmann number without a spinor index. This provides a formulation of supersymmetric theories that depends exclusively on propagating degrees of freedom, at the expense of obscuring Lorentz invariance and introducing inverse momentum scales. After establishing the general framework, we construct collinear superspace Lagrangians for free chiral matter and non-Abelian gauge fields. For the latter construction, an important ingredient is a superfield representation that is simultaneously chiral, anti-chiral, and real; this novel object encodes residual gauge transformations on the light cone. Additionally, we discuss a fundamental obstruction to constructing interacting theories with chiral matter; overcoming these issues is the subject of our companion paper, where we introduce a larger set of superfields to realize the full range of interactions compatible with $\mathcal{N} = 1$. Along the way, we provide a novel framing of reparametrization invariance using a spinor decomposition, which provides insight into this important light-cone symmetry.

hep-th↗

arXiv2020

Circumnavigating Collinear Superspace

In this paper, we extend the collinear superspace formalism to include the full range of $\mathcal{N} = 1$ supersymmetric interactions. Building on the effective field theory rules developed in a companion paper - "Navigating Collinear Superspace" - we construct collinear superspace Lagrangians for theories with non-trivial $F$- and $D$-term auxiliary fields. For (massless) Wess-Zumino models, the key ingredient is a novel type of Grassmann-valued supermultiplet whose lowest component is a (non-propagating) fermionic degree of freedom. For gauge theories coupled to charged chiral matter, the key ingredient is a novel type of vector superfield whose lowest component is a non-propagating gauge potential. This unique vector superfield is used to construct a gauge-covariant derivative; while such an object does not appear in the standard full superspace formalism, it is crucial for modeling gauge interactions when the theory is expressed on a collinear slice. This brings us full circle, by showing that all types of $\mathcal{N} = 1$ theories in four dimensions can be constructed in collinear superspace from purely infrared considerations. We speculate that supersymmetric theories with $\mathcal{N} > 1$ could also be implemented using similar collinear superspace constructions.

hep-th↗

arXiv2020

Cutting Multiparticle Correlators Down to Size

Multiparticle correlators are mathematical objects frequently encountered in quantum field theory and collider physics. By translating multiparticle correlators into the language of graph theory, we can gain new insights into their structure as well as identify efficient ways to manipulate them. In this paper, we highlight the power of this graph-theoretic approach by "cutting open" the vertices and edges of the graphs, allowing us to systematically classify linear relations among multiparticle correlators and develop faster methods for their computation. The naive computational complexity of an $N$-point correlator among $M$ particles is $\mathcal O(M^N)$, but when the pairwise distances between particles can be cast as an inner product, we show that all such correlators can be computed in linear $\mathcal O(M)$ runtime. With the help of new tensorial objects called Energy Flow Moments, we achieve a fast implementation of jet substructure observables like $C_2$ and $D_2$, which are widely used at the Large Hadron Collider to identify boosted hadronic resonances. As another application, we compute the number of leafless multigraphs with $d$ edges up to $d = 16$ (15,641,159), conjecturing that this is the same as the number of independent kinematic polynomials of degree $d$, previously known only to $d=8$ (279).

hep-ph↗

arXiv2020

Exploring the Space of Jets with CMS Open Data

We explore the metric space of jets using public collider data from the CMS experiment. Starting from 2.3/fb of 7 TeV proton-proton collisions collected at the Large Hadron Collider in 2011, we isolate a sample of 1,690,984 central jets with transverse momentum above 375 GeV. To validate the performance of the CMS detector in reconstructing the energy flow of jets, we compare the CMS Open Data to corresponding simulated data samples for a variety of jet kinematic and substructure observables. Even without detector unfolding, we find very good agreement for track-based observables after using charged hadron subtraction to mitigate the impact of pileup. We perform a range of novel analyses, using the "energy mover's distance" (EMD) to measure the pairwise difference between jet energy flows. The EMD allows us to quantify the impact of detector effects, visualize the metric space of jets, extract correlation dimensions, and identify the most and least typical jet configurations. To facilitate future jet studies with CMS Open Data, we make our datasets and analysis code available, amounting to around two gigabytes of distilled data and one hundred gigabytes of simulation files.

hep-ph↗

arXiv2019

Measurements of two-particle correlations in $e^+e^-$ collisions at 91 GeV with ALEPH archived data

Measurements of two-particle angular correlations of charged particles emitted in hadronic $Z$ decays are presented. The archived $e^+e^-$ annihilation data at a center-of-mass energy of 91 GeV were collected with the ALEPH detector at LEP between 1992 and 1995. The correlation functions are measured over a broad range of pseudorapidity and full azimuth as a function of charged particle multiplicity. No significant long-range correlation is observed in either the lab coordinate analysis or the thrust coordinate analysis, where the latter is sensitive to a medium expanding transverse to the color string between the outgoing $q\bar{q}$ pair from $Z$ boson decays. The associated yield distributions in both analyses are in better agreement with the prediction from the PYTHIA v6.1 event generator than from HERWIG v7.1.5. They provide new insights to showering and hadronization modeling. These results serve as an important reference to the observed long-range correlation in proton-proton, proton-nucleus, and nucleus-nucleus collisions.

hep-ex↗

arXiv2019

Sorting out quenched jets

We introduce a new 'quantile' analysis strategy to study the modification of jets as they traverse through a droplet of quark-gluon plasma. To date, most jet modification studies have been based on comparing the jet properties measured in heavy-ion collisions to a proton-proton baseline at the same reconstructed jet transverse momentum ($p_T$). It is well known, however, that the quenching of jets from their interaction with the medium leads to a migration of jets from higher to lower $p_T$, making it challenging to directly infer the degree and mechanism of jet energy loss. Our proposed quantile matching procedure is inspired by (but not reliant on) the approximate monotonicity of energy loss in the jet $p_T$. In this strategy, jets in heavy-ion collisions ordered by $p_T$ are viewed as modified versions of the same number of highest-energy jets in proton-proton collisions, and the fractional energy loss as a function of jet $p_T$ is a natural observable ($Q_{\rm AA}$). Furthermore, despite non-monotonic fluctuations in the energy loss, we use an event generator to validate the strong correlation between the $p_T$ of the parton that initiates a heavy-ion jet and the $p_T$ of the vacuum jet which corresponds to it via the quantile procedure ($p_T^{\rm quant}$). We demonstrate that this strategy both provides a complementary way to study jet modification and mitigates the effect of $p_T$ migration in heavy-ion collisions.

hep-ph↗

arXiv2019

Searching for Axion Dark Matter with Birefringent Cavities

Axion-like particles are a broad class of dark matter candidates which are expected to behave as a coherent, classical field with a weak coupling to photons. Research into the detectability of these particles with laser interferometers has recently revealed a number of promising experimental designs. Inspired by these ideas, we propose the Axion Detection with Birefringent Cavities (ADBC) experiment, a new axion interferometry concept using a cavity that exhibits birefringence between its two, linearly polarized laser eigenmodes. This experimental concept overcomes several limitations of the designs currently in the literature, and can be practically realized in the form of a simple bowtie cavity with tunable mirror angles. Our design thereby increases the sensitivity to the axion-photon coupling over a wide range of axion masses.

hep-ph↗

arXiv2019

Searching in CMS Open Data for Dimuon Resonances with Substantial Transverse Momentum

We study dimuon events in 2.11/fb of 7 TeV pp collisions, using CMS Open Data, and search for a narrow dimuon resonance with moderate mass (14-66 GeV) and substantial transverse momentum (pT). Applying dimuon pT cuts of 25 GeV and 60 GeV, we explore two overlapping samples: one with isolated muons, and one with prompt muons without an isolation requirement. Using the latter sample requires information about detector effects and QCD backgrounds, which we obtain directly from the CMS Open Data. We present model-independent limits on the product of cross section, branching fraction, acceptance, and efficiencies. These limits are stronger, relative to a corresponding inclusive search without a pT cut, by factors of as much as nine. Our "pT-enhanced" dimuon search strategy provides improved sensitivity to models in which a new particle is produced mainly in the decay of something heavier, as could occur, for example, in decays of the Higgs boson or of a TeV-scale top partner. An implementation of this method with the current 13 TeV data should improve the sensitivity to such signals further by roughly an order of magnitude.

hep-ph↗

arXiv2019

The Metric Space of Collider Events

When are two collider events similar? Despite the simplicity and generality of this question, there is no established notion of the distance between two events. To address this question, we develop a metric for the space of collider events based on the earth mover's distance: the "work" required to rearrange the radiation pattern of one event into another. We expose interesting connections between this metric and the structure of infrared- and collinear-safe observables, providing a novel technique to quantify event modifications due to hadronization, pileup, and detector effects. We showcase how this metrization unlocks powerful new tools for analyzing and visualizing collider data without relying upon a choice of observables. More broadly, this framework paves the way for data-driven collider phenomenology without specialized observables or machine learning models.

hep-ph↗