arXiv subjects

Michael Park

Publications and source records attributed to Michael Park.

17 recordsLinked to original sources

arXiv2026

Is Innovation Becoming Less Disruptive? An Inventory of the Literature

A growing literature has examined whether innovation is becoming less disruptive, spanning diverse domains and data sources and using a range of methodologies. This paper provides an inventory of 105 studies exploring this question. The evidence is largely consistent in direction. Studies spanning scientific papers, patents, products, legal cases, music, and visual art consistently report evidence of a decline. This pattern holds not only for citation-based measures, but also for text-based approaches, firm displacement rates, product similarity networks, and audio and visual embeddings. The literature has also identified notable exceptions, including rebounds in specific domains and predictable variation across field lifecycles. We catalog each study's data, methods, and findings to provide a resource for researchers and policymakers seeking to understand the current state of the evidence.

physics.soc-ph↗

arXiv2025

Robust Evidence for Declining Disruptiveness: Assessing the Role of Zero-Backward-Citation Works

We respond to Holst et al.'s critique that the decline in scientific disruptiveness documented in Park et al. (Nature, 2023) is an artifact of including works with zero backward citations. Using their advocated dataset, metric, and exclusion criteria, we find declines equivalent to major benchmark transformations in science. Their own regression model--designed to address their concerns about zero-citation works--yields large and significant declines for both papers and patents (p<0.001), a result found in their supplementary tables yet left unaddressed, despite directly contradicting their central claim. Their critique is further undermined by severe quality issues in their data, which contain three times more zero-citation works than ours. We trace this excess to their inclusion of at least 2.8 million editorials, obituaries, and comments, 1.5 million books and proceedings, and 254,000 product and artistic reviews--in all, 20% of their sample is non-research content that almost by definition lacks backward citations. Simple keyword searches confirm the problem's severity, identifying among others 456 For Dummies guides, 50 Dr. Seuss and Curious George books, and the Captain Underpants series--all zero-citation entries in their sample. Applying granular document type classification to their data reveals that such non-research content fell from 40% to 8% of their sample between 1945 and 2010--a shift sufficient to generate the decline in zero-citation prevalence they attribute to metadata errors in our study. Standard practice excludes such content to guard against the metadata quality concerns at the center of their critique--concerns their dataset exemplifies rather than addresses. Declining disruptiveness has been documented in nearly 100 studies across multiple databases, metrics, and non-citation-based measures. The weight of evidence does not support an artifact-based explanation.

cs.SI↗

arXiv2023

Serendipity in Science

Serendipity plays an important role in scientific discovery. Indeed, many of the most important breakthroughs, ranging from penicillin to the electric battery, have been made by scientists who were stimulated by a chance exposure to unsought but useful information. However, not all scientists are equally likely to benefit from such serendipitous exposure. Although scholars generally agree that scientists with a prepared mind are most likely to benefit from serendipitous encounters, there is much less consensus over what precisely constitutes a prepared mind, with some research suggesting the importance of openness and others emphasizing the need for deep prior experience in a particular domain. In this paper, we empirically investigate the role of serendipity in science by leveraging a policy change that exogenously shifted the shelving location of journals in university libraries and subsequently exposed scientists to unsought scientific information. Using large-scale data on 2.4 million papers published in 9,750 journals by 520,000 scientists at 115 North American research universities, we find that scientists with greater openness are more likely to benefit from serendipitous encounters. Following the policy change, these scientists tended to cite less familiar and newer work, and ultimately published papers that were more innovative. By contrast, we find little effect on innovativeness for scientists with greater depth of experience, who, in our sample, tended to cite more familiar and older work following the policy change.

econ.EM↗

arXiv2022

The decline of disruptive science and technology

Theories of scientific and technological change view discovery and invention as endogenous processes, wherein prior accumulated knowledge enables future progress by allowing researchers to, in Newton's words, "stand on the shoulders of giants". Recent decades have witnessed exponential growth in the volume of new scientific and technological knowledge, thereby creating conditions that should be ripe for major advances. Yet contrary to this view, studies suggest that progress is slowing in several major fields of science and technology. Here, we analyze these claims at scale across 6 decades, using data on 45 million papers and 3.5 million patents from 6 large-scale datasets. We find that papers and patents are increasingly less likely to break with the past in ways that push science and technology in new directions, a pattern that holds universally across fields. Subsequently, we link this decline in disruptiveness to a narrowing in the use of prior knowledge, allowing us to reconcile the patterns we observe with the "shoulders of giants" view. We find that the observed declines are unlikely to be driven by changes in the quality of published science, citation practices, or field-specific factors. Overall, our results suggest that slowing rates of disruption may reflect a fundamental shift in the nature of science and technology.

cs.SI↗

arXiv2021

FNetAR: Mixing Tokens with Autoregressive Fourier Transforms

In this note we examine the autoregressive generalization of the FNet algorithm, in which self-attention layers from the standard Transformer architecture are substituted with a trivial sparse-uniformsampling procedure based on Fourier transforms. Using the Wikitext-103 benchmark, we demonstratethat FNetAR retains state-of-the-art performance (25.8 ppl) on the task of causal language modelingcompared to a Transformer-XL baseline (24.2 ppl) with only half the number self-attention layers,thus providing further evidence for the superfluity of deep neural networks with heavily compoundedattention mechanisms. The autoregressive Fourier transform could likely be used for parameterreduction on most Transformer-based time-series prediction models.

cs.CL↗

arXiv2020

SyDeBO: Symbolic-Decision-Embedded Bilevel Optimization for Long-Horizon Manipulation in Dynamic Environments

This study proposes a Task and Motion Planning (TAMP) method with symbolic decisions embedded in a bilevel optimization. This TAMP method exploits the discrete structure of sequential manipulation for long-horizon and versatile tasks in dynamically changing environments. At the symbolic planning level, we propose a scalable decision-making method for long-horizon manipulation tasks using the Planning Domain Definition Language (PDDL) with causal graph decomposition. At the motion planning level, we devise a trajectory optimization (TO) approach based on the Alternating Direction Method of Multipliers (ADMM), suitable for solving constrained, large-scale nonlinear optimization in a distributed manner. Distinct from conventional geometric motion planners, our approach generates highly dynamic manipulation motions by incorporating the full robot and object dynamics. Furthermore, in lieu of a hierarchical planning approach, we solve a holistically integrated bilevel optimization problem involving costs from both the low-level TO and the high-level search. Simulation and experimental results demonstrate dynamic manipulation for long-horizon object sorting tasks in clutter and on a moving conveyor belt.

cs.RO↗

arXiv2020

Moment-Matching Graph-Networks for Causal Inference

In this note we explore a fully unsupervised deep-learning framework for simulating non-linear structural equation models from observational training data. The main contribution of this note is an architecture for applying moment-matching loss functions to the edges of a causal Bayesian graph, resulting in a generative conditional-moment-matching graph-neural-network. This framework thus enables automated sampling of latent space conditional probability distributions for various graphical interventions, and is capable of generating out-of-sample interventional probabilities that are often faithful to the ground truth distributions well beyond the range contained in the training set. These methods could in principle be used in conjunction with any existing autoencoder that produces a latent space representation containing causal graph structures.

stat.ML↗

arXiv2019

Quantum Circuits for Scale Transformations

In this note we develop quantum circuits for exactly simulating the thermal properties of the quantum XY/Ising chain. These circuits are applicable to the simplest integrable lattice models for which the exact momentum-space scattering-matrix can be constructed in terms of local two-body gates. We discuss a circuit analogy for Wilsonian renormalization group flow and show how it can be used to simulate a global coarse graining transformation. We then construct a circuit that transforms the ground state of these models into the thermo-field double state, and show how it can be used to implement a local scale transformation.

quant-ph↗

arXiv2018

Composite Higgs Models with a Hidden Sector

We discuss the phenomenology of composite Higgs models that naturally produce a Standard Model-like Higgs boson with a mass of $126$ GeV. The effective theory below the compositeness scale is weakly coupled in these models, and the goldstone sector acts as a portal between the third generation of quarks and a hidden gauge sector. The addition of hidden-sector fermions gives rise to a calculable effective scalar potential with a naturally light scalar resonance. The generic prediction of these theories is the existence of additional pseudo-Nambu Goldstone bosons with electroweak-scale cross sections and masses. In this paper we analyze the collider signatures for some simple concrete realizations of this framework. We find that despite the existence of additional weakly and strongly coupled particles that are kinematically within reach of current experiments, the generic signatures are difficult to resolve at the LHC, and could remain well-hidden in the absence of an $e^+ e^-$ Higgs factory such as the CEPC or a surface detector such as MATHUSLA.

hep-ph↗

arXiv2015

Revealing Compressed Stops Using High-Momentum Recoils

Searches for supersymmetric top quarks at the LHC have been making great progress in pushing sensitivity out to higher mass, but are famously plagued by gaps in coverage around lower-mass regions where the decay phase space is closing off. Within the common stop-NLSP / neutralino-LSP simplified model, the line in the mass plane where there is just enough phase space to produce an on-shell top quark remains almost completely unconstrained. Here, we show that is possible to define searches capable of probing a large patch of this difficult region, with S/B ~ 1 and significances often well beyond 5 sigma. The basic strategy is to leverage the large energy gain of LHC Run 2, leading to a sizable population of stop pair events recoiling against a hard jet. The recoil not only re-establishes a MET signature, but also leads to a distinctive anti-correlation between the MET and the recoil jet transverse vectors when the stops decay all-hadronically. Accounting for jet combinatorics, backgrounds, and imperfections in MET measurements, we estimate that Run 2 will already start to close the gap in exclusion sensitivity with the first few 10s of inverse-fb. By 300/fb, exclusion sensitivity may extend from stop masses of 550 GeV on the high side down to below 200 GeV on the low side, approaching the "stealth" point at m(stop) = m(top) and potentially overlapping with limits from top pair cross section and spin correlation measurements.

hep-ph↗

arXiv2013

Multi-Lepton Signals of Top-Higgs Associated Production

We evaluate the potential to measure ttbar H associated production at the LHC using non-resonant multi-lepton final states in conjunction with two or more b-tags. The multi-lepton ttbar H signal arises predominantly from H to tau tau and H to WW^* alongside the semi-leptonic or fully leptonic decay of the ttbar pair. We demonstrate the power of a multi-lepton search for ttbar H associated production by recasting the CMS b-tagged multi-lepton search with 19.5 fb^-1 of 8 TeV data to obtain an observed (expected) limit of 4.7 (6.6) times the Standard Model rate, comparable to ongoing searches in 4b and bb gamma gamma final states. Sensitivity can be further improved by the addition of exclusive channels involving same-sign dileptons. We recast the CMS b-tagged same-sign dilepton search with 10.5 fb^-1 of 8 TeV data to set limits on ttbar H associated production, and approximately combine the two searches by calculating the fraction of same-sign dilepton signal events which do not satisfy multi-lepton selection criteria. We estimate an expected total non-resonant leptonic reach of mu < 5.0 times the Standard Model rate in 20 fb^-1 of 8 TeV data, with improvements possible.

hep-ph↗

arXiv2012

Searching for $t \to c h$ with Multi-Leptons

The results of a multi-lepton search conducted by the CMS collaboration with 5 fb$^{-1}$ of data collected from 7 TeV $pp$ collisions are used to place the first bound on the rare flavor-changing decay of the top quark to a Higgs boson and charm quark. Combining results from a number of exclusive three- and four-lepton search channels yields an estimated upper limit of ${\rm Br}(t \to ch) < 2.7%$ for a Higgs boson mass of 125 GeV. The sensitivity of future dedicated searches for $t \to c h$ could be improved by adding exclusive same sign di-lepton channels, as well as by sub-dividing channels based on $b$-quark tagging and partial kinematic top quark and Higgs boson tagging. This bound may be interpreted more widely within a range of new physics processes that yield final states with a $W$-boson in association with a Higgs boson. For such processes with kinematics that are similar to top--anti-top production and decay, the estimated limit on cross section times branching ratio corresponds to roughly $σ\cdot {\rm Br}(pp \to WhX) < 9$ pb.

hep-ph↗

arXiv2012

Multi-Lepton Signals of Multiple Higgs Bosons

We identify and investigate novel multi-lepton signatures of extended Higgs sectors at the LHC in the guise of CP- and flavor-conserving two-Higgs-doublet models (2HDMs). Rather than designing individual searches tailored to specific 2HDM signals, we employ the combination of many exclusive multi-lepton search channels to probe the collective signal from the totality of production and decay processes. Multi-lepton signals of 2HDMs can arise from a variety of sources, including Standard Model-like production of the CP-even scalars, $h$ and $H$, through gluon-fusion with $h,H \to ZZ^{(*)}$, or associated production with vector bosons or top quarks, with $h,H \to WW^{(*)}, ZZ^{(*)},ττ$. Additional sources include gluon-fusion production of the heavy CP-even scalar with decays through the light CP-even scalar, the CP-odd scalar, $A$, or the charged scalar, $H^\pm$, such as $H \to hh$, $H \to AA$, $H \to H^+ H^-$, $H \to ZA$, with $A \to Zh, ττ$, $H^\pm \to Wh$, and $h \to WW^*, ZZ^*,ττ$. Altogether, the combined multi-lepton signal may greatly exceed that of the Standard Model Higgs boson and provides a sensitive probe of extended Higgs sectors. As a proof of principle, we use a factorized mapping procedure between model parameters and signatures to determine multi-lepton sensitivities in four different flavor conserving 2HDM parameter spaces by simulating the acceptance times efficiency in 20 exclusive multi-lepton channels for 222 independent production and decay topologies that arise for four benchmark 2HDM spectra within each parameter space. A comparison of these sensitivities with the results of a multi-lepton search conducted by the CMS collaboration using 5 fb$^{-1}$ of data collected from 7 TeV $pp$ collisions yields new limits in regions of 2HDM parameter space that have not previously been covered by other types of direct experimental searches.

hep-ph↗

arXiv2011

Multi-Lepton Signals of the Higgs Boson

The possibility of searching for the Higgs boson in channels with multiple non-resonant leptons is evaluated in light of recent advances in multi-lepton search techniques at the LHC. The total multi-lepton Higgs signal exceeds the four lepton gold-plated resonant mode, but is spread over many channels with same-sign di-lepton, tri-lepton, and four lepton final states. While any individual channel alone is not significant, the exclusive combination across multiple channels is shown to provide a sensitivity competitive with other discovery level searches for the Higgs boson. We estimate that with 5 inverse femtobarn of data, existing non-optimized multi-lepton searches at the LHC could exclude the Higgs boson to 95% CL at a few times the predicted Standard Model cross section in the mass range 120-150 GeV. Refinements focused specifically on the Higgs boson signal are suggested that would further increase sensitivity. We illustrate the possibility of discerning patterns in production and decay modes using correlations across multiple channels by comparing sensitivities to Standard Model, Fermi-phobic, and b-phobic Higgs bosons.

hep-ph↗

arXiv2011

Backgrounds To Higgs Boson Searches from Asymmetric Internal Conversion

A class of potential backgrounds for Higgs boson searches in the W W to di-lepton channel at both the Tevatron and Large Hadron Collider is presented. Backgrounds from W photon production with external conversion of the on-shell photon in detector material to an asymmetric electron-positron pair, with loss of the trailing electron or positron has been treated adequately in Higgs searches. Here we consider analogous backgrounds from the production of a W with an off-shell photon and with the internal conversion of the off-shell photon in vacuum to an asymmetric lepton-anti-lepton pair. While the former process yields almost entirely electrons or positrons, the latter can give electron, positron, muon, and anti-muon backgrounds in roughly equal amounts. We estimate that asymmetric internal conversion backgrounds are comparable to the Higgs boson signal in the standard signal region of phase space. These processes also represent potential backgrounds for new physics searches in same-sign di-lepton channels. Some data driven methods to characterize asymmetric internal conversion backgrounds are suggested.

hep-ph↗

arXiv2011

Recovering Particle Masses from Missing Energy Signatures with Displaced Tracks

If physics beyond the Standard Model contains metastable new particle states, its signatures may manifest themselves at the LHC through the presence of displaced vertices or displaced tracks. Such signatures are common in low-scale supersymmetry breaking scenarios as well as a host of other well-studied new physics models. Here we analyze the kinematics of dual cascade decays with missing energy resulting from a stable non-interacting new particle state at the bottom of the decay chain. We find that if the final step of the decay chain involves a metastable particle, the resulting presence of displaced tracks provides strong constraints with which to solve for the unknown kinematic quantities lost through missing energy. In addition we develop techniques with which to recover all kinematic unknowns, even in situations where the unknown quantities outnumber the constraints. These techniques can all be performed using a very small number of events and can thus be applied to very early discovery level searches at the LHC.

hep-ph↗

arXiv2011

Simplified Models for LHC New Physics Searches

This document proposes a collection of simplified models relevant to the design of new-physics searches at the LHC and the characterization of their results. Both ATLAS and CMS have already presented some results in terms of simplified models, and we encourage them to continue and expand this effort, which supplements both signature-based results and benchmark model interpretations. A simplified model is defined by an effective Lagrangian describing the interactions of a small number of new particles. Simplified models can equally well be described by a small number of masses and cross-sections. These parameters are directly related to collider physics observables, making simplified models a particularly effective framework for evaluating searches and a useful starting point for characterizing positive signals of new physics. This document serves as an official summary of the results from the "Topologies for Early LHC Searches" workshop, held at SLAC in September of 2010, the purpose of which was to develop a set of representative models that can be used to cover all relevant phase space in experimental searches. Particular emphasis is placed on searches relevant for the first ~50-500 pb-1 of data and those motivated by supersymmetric models. This note largely summarizes material posted at http://lhcnewphysics.org/, which includes simplified model definitions, Monte Carlo material, and supporting contacts within the theory community. We also comment on future developments that may be useful as more data is gathered and analyzed by the experiments.

hep-ph↗