-
Predicting discrete-time bifurcations with deep learning
Authors:
Thomas M. Bury,
Daniel Dylewsky,
Chris T. Bauch,
Madhur Anand,
Leon Glass,
Alvin Shrier,
Gil Bub
Abstract:
Many natural and man-made systems are prone to critical transitions -- abrupt and potentially devastating changes in dynamics. Deep learning classifiers can provide an early warning signal (EWS) for critical transitions by learning generic features of bifurcations (dynamical instabilities) from large simulated training data sets. So far, classifiers have only been trained to predict continuous-tim…
▽ More
Many natural and man-made systems are prone to critical transitions -- abrupt and potentially devastating changes in dynamics. Deep learning classifiers can provide an early warning signal (EWS) for critical transitions by learning generic features of bifurcations (dynamical instabilities) from large simulated training data sets. So far, classifiers have only been trained to predict continuous-time bifurcations, ignoring rich dynamics unique to discrete-time bifurcations. Here, we train a deep learning classifier to provide an EWS for the five local discrete-time bifurcations of codimension-1. We test the classifier on simulation data from discrete-time models used in physiology, economics and ecology, as well as experimental data of spontaneously beating chick-heart aggregates that undergo a period-doubling bifurcation. The classifier outperforms commonly used EWS under a wide range of noise intensities and rates of approach to the bifurcation. It also predicts the correct bifurcation in most cases, with particularly high accuracy for the period-doubling, Neimark-Sacker and fold bifurcations. Deep learning as a tool for bifurcation prediction is still in its nascence and has the potential to transform the way we monitor systems for critical transitions.
△ Less
Submitted 8 February, 2024; v1 submitted 16 March, 2023;
originally announced March 2023.
-
Flavor-dependent b-profiles from Drell-Yan spectra at low transverse momenta
Authors:
M. Bury,
F. Hautmann,
S. Leal-Gomez,
I. Scimemi,
A. Vladimirov,
P. Zurita
Abstract:
We discuss our recent study of the impact of collinear PDFs and their uncertainties on the determination of transverse momentum dependent (TMD) distributions and the description of Drell-Yan (DY) production measurements at low transverse momenta. Using QCD factorization and evolution in transverse coordinate b space, this study takes into account for the first time flavor-dependent non-perturbativ…
▽ More
We discuss our recent study of the impact of collinear PDFs and their uncertainties on the determination of transverse momentum dependent (TMD) distributions and the description of Drell-Yan (DY) production measurements at low transverse momenta. Using QCD factorization and evolution in transverse coordinate b space, this study takes into account for the first time flavor-dependent non-perturbative b-profiles. It illustrates that collinear PDF uncertainties and non-perturbative TMD flavor dependence are both essential to obtain reliable TMD determinations.
△ Less
Submitted 9 November, 2022;
originally announced November 2022.
-
Universal Early Warning Signals of Phase Transitions in Climate Systems
Authors:
Daniel Dylewsky,
Timothy M. Lenton,
Marten Scheffer,
Thomas M. Bury,
Christopher G. Fletcher,
Madhur Anand,
Chris T. Bauch
Abstract:
The potential for complex systems to exhibit tip** points in which an equilibrium state undergoes a sudden and often irreversible shift is well established, but prediction of these events using standard forecast modeling techniques is quite difficult. This has led to the development of an alternative suite of methods that seek to identify signatures of critical phenomena in data, which are expec…
▽ More
The potential for complex systems to exhibit tip** points in which an equilibrium state undergoes a sudden and often irreversible shift is well established, but prediction of these events using standard forecast modeling techniques is quite difficult. This has led to the development of an alternative suite of methods that seek to identify signatures of critical phenomena in data, which are expected to occur in advance of many classes of dynamical bifurcation. Crucially, the manifestations of these critical phenomena are generic across a variety of systems, meaning that data-intensive deep learning methods can be trained on (abundant) synthetic data and plausibly prove effective when transferred to (more limited) empirical data sets. This paper provides a proof of concept for this approach as applied to lattice phase transitions: a deep neural network trained exclusively on 2D Ising model phase transitions is tested on a number of real and simulated climate systems with considerable success. Its accuracy frequently surpasses that of conventional statistical indicators, with performance shown to be consistently improved by the inclusion of spatial indicators. Tools such as this may offer valuable insight into climate tip** events, as remote sensing measurements provide increasingly abundant data on complex geospatially-resolved Earth systems.
△ Less
Submitted 5 December, 2022; v1 submitted 31 May, 2022;
originally announced June 2022.
-
PDF bias and flavor dependence in TMD distributions
Authors:
Marcin Bury,
Francesco Hautmann,
Sergio Leal-Gomez,
Ignazio Scimemi,
Alexey Vladimirov,
Pia Zurita
Abstract:
Transverse momentum dependent (TMD) distributions match collinear parton density functions (PDF) in the limit of small transverse distances, which is accounted for by global extractions of TMD distributions. We study the influence of the collinear PDF value and uncertainties on the determination of unpolarized TMD distributions and the description of Drell-Yan (DY) and Z-boson production measureme…
▽ More
Transverse momentum dependent (TMD) distributions match collinear parton density functions (PDF) in the limit of small transverse distances, which is accounted for by global extractions of TMD distributions. We study the influence of the collinear PDF value and uncertainties on the determination of unpolarized TMD distributions and the description of Drell-Yan (DY) and Z-boson production measurements at low transverse momenta. We take into account, for the first time, flavor-dependent non-perturbative TMD profiles. We carry out a Bayesian analysis to incorporate the propagation of PDF uncertainties into TMD extractions. We find that collinear PDF uncertainties and non-perturbative TMD flavor dependence are both essential to obtain reliable TMD determinations, and should be included in future global analyses.
△ Less
Submitted 27 July, 2022; v1 submitted 18 January, 2022;
originally announced January 2022.
-
TMDlib2 and TMDplotter: a platform for 3D hadron structure studies
Authors:
N. A. Abdulov,
A. Bacchetta,
S. Baranov,
A. Bermudez Martinez,
V. Bertone,
C. Bissolotti,
V. Candelise,
L. I. Estevez Banos,
M. Bury,
P. L. S. Connor,
L. Favart,
F. Guzman,
F. Hautmann,
M. Hentschinski,
H. Jung,
L. Keersmaekers,
A. Kotikov,
A. Kusina,
K. Kutak,
A. Lelek,
J. Lidrych,
A. Lipatov,
G. Lykasov,
M. Malyshev,
M. Mendizabal
, et al. (13 additional authors not shown)
Abstract:
A common library, TMDlib2, for Transverse-Momentum-Dependent distributions (TMDs) and unintegrated parton distributions (uPDFs) is described, which allows for easy access of commonly used TMDs and uPDFs, providing a three-dimensional (3D) picture of the partonic structure of hadrons. The tool TMDplotter allows for web-based plotting of distributions implemented in TMDlib2, together with collinear…
▽ More
A common library, TMDlib2, for Transverse-Momentum-Dependent distributions (TMDs) and unintegrated parton distributions (uPDFs) is described, which allows for easy access of commonly used TMDs and uPDFs, providing a three-dimensional (3D) picture of the partonic structure of hadrons. The tool TMDplotter allows for web-based plotting of distributions implemented in TMDlib2, together with collinear pdfs as available in LHAPDF.
△ Less
Submitted 16 August, 2021; v1 submitted 17 March, 2021;
originally announced March 2021.
-
Extraction of the Sivers function from SIDIS, Drell-Yan, and $W^\pm/Z$ boson production data with TMD evolution
Authors:
Marcin Bury,
Alexei Prokudin,
Alexey Vladimirov
Abstract:
We perform a global fit of the available polarized Semi-Inclusive Deep Inelastic Scattering (SIDIS), polarized pion-induced Drell-Yan (DY) and $W^\pm/Z$ boson production data at N$^3$LO and NNLO accuracy of the Transverse Momentum Dependent (TMD) evolution, and extract the Sivers function for $u$, $d$, $s$ and for sea quarks. The Qiu-Sterman function is determined in a model independent way via th…
▽ More
We perform a global fit of the available polarized Semi-Inclusive Deep Inelastic Scattering (SIDIS), polarized pion-induced Drell-Yan (DY) and $W^\pm/Z$ boson production data at N$^3$LO and NNLO accuracy of the Transverse Momentum Dependent (TMD) evolution, and extract the Sivers function for $u$, $d$, $s$ and for sea quarks. The Qiu-Sterman function is determined in a model independent way via the operator product expansion from the extracted Sivers function. The analysis is supplemented by additional studies, such as the estimation of applicability region, the impact of the unpolarized distributions' uncertainties, the universality of the Sivers functions, positivity constraints, the significance of the sign-change relation, and the comparison with the existing extractions
△ Less
Submitted 4 March, 2021;
originally announced March 2021.
-
N$^3$LO extraction of the Sivers function from SIDIS, Drell-Yan, and $W^\pm/Z$ data
Authors:
Marcin Bury,
Alexei Prokudin,
Alexey Vladimirov
Abstract:
We perform the global analysis of polarized Semi-Inclusive Deep Inelastic Scattering (SIDIS), pion-induced polarized Drell-Yan (DY), and $W^\pm/Z$ boson production data and extract the Sivers function for $u$, $d$, $s$ and for sea-quarks. We use the framework of transverse momentum dependent factorization at N$^3$LO accuracy. The Qiu-Sterman function is determined in a model-independent way from t…
▽ More
We perform the global analysis of polarized Semi-Inclusive Deep Inelastic Scattering (SIDIS), pion-induced polarized Drell-Yan (DY), and $W^\pm/Z$ boson production data and extract the Sivers function for $u$, $d$, $s$ and for sea-quarks. We use the framework of transverse momentum dependent factorization at N$^3$LO accuracy. The Qiu-Sterman function is determined in a model-independent way from the extracted Sivers function. We also evaluate the significance of the predicted sign change of Sivers function in DY with respect to SIDIS.
△ Less
Submitted 9 December, 2020;
originally announced December 2020.
-
Forward trijet production in p-p and p-Pb collisions at LHC
Authors:
Marcin Bury,
Andreas van Hameren,
Piotr Kotko,
Krzysztof Kutak
Abstract:
We calculate various azimuthal angle distributions for three jets produced in the forward rapidity region with transverse momenta $p_T>20\,\mathrm{GeV}$ in proton-proton (p-p) and proton-lead (p-Pb) collisions at center of mass energy $5.02\,\,\mathrm{TeV}$. We use the multi-parton extension of the so-called small-$x$ Improved Transverse Momentum Dependent factorization (ITMD). We study effects re…
▽ More
We calculate various azimuthal angle distributions for three jets produced in the forward rapidity region with transverse momenta $p_T>20\,\mathrm{GeV}$ in proton-proton (p-p) and proton-lead (p-Pb) collisions at center of mass energy $5.02\,\,\mathrm{TeV}$. We use the multi-parton extension of the so-called small-$x$ Improved Transverse Momentum Dependent factorization (ITMD). We study effects related to change from the standard $k_T$-factorization to ITMD factorization as well as changes as one goes from p-p collision to p-Pb. We observe rather large differences in the distribution when we change the factorization approach, which allows to both improve the small-$x$ TMD gluon distributions as well as validate and improve the factorization approach. We also see significant depletion of the nuclear modification ratio, indicating a possibility of searches for saturation effects using trijet final states in a more exclusive way than for dijets.
△ Less
Submitted 10 September, 2020; v1 submitted 23 June, 2020;
originally announced June 2020.
-
TMD gluon distributions for multiparton processes
Authors:
Marcin Bury,
Piotr Kotko,
Krzysztof Kutak
Abstract:
We derive gauge invariant operators entering definitions of the Transverse Momentum Dependent (TMD) gluon distributions, for all five and six parton processes. Our calculations utilize color decomposition of amplitudes in the color flow basis. In addition, we find the general result for multi-gluon process (with arbitrary number of gluons) at large $N_{c}$. On phenomenological ground our results m…
▽ More
We derive gauge invariant operators entering definitions of the Transverse Momentum Dependent (TMD) gluon distributions, for all five and six parton processes. Our calculations utilize color decomposition of amplitudes in the color flow basis. In addition, we find the general result for multi-gluon process (with arbitrary number of gluons) at large $N_{c}$. On phenomenological ground our results may be used for multi-jet production in the small-$x$ regime, where the TMD gluon distributions can be derived from the Color Glass Condensate effective theory.
△ Less
Submitted 21 July, 2020; v1 submitted 24 September, 2018;
originally announced September 2018.
-
Single inclusive jet production and the nuclear modification ratio at very forward rapidity in proton-lead collisions with $\sqrt{s_{NN}}$ = 5.02 TeV
Authors:
Marcin Bury,
Hans Van Haevermaet,
Andreas Van Hameren,
Pierre Van Mechelen,
Krzysztof Kutak,
Mirko Serino
Abstract:
We present calculations of single inclusive jet transverse momentum and energy spectra at forward rapidity ($5.2\!<\!y\!<\!6.6$) in proton-lead collisions with $\sqrt{s_{NN}}$ = 5.02 TeV. The predictions are obtained with the KaTie Monte Carlo event generator, which allows to calculate interactions within the High Energy Factorisation framework. The tree-level matrix element results are subsequent…
▽ More
We present calculations of single inclusive jet transverse momentum and energy spectra at forward rapidity ($5.2\!<\!y\!<\!6.6$) in proton-lead collisions with $\sqrt{s_{NN}}$ = 5.02 TeV. The predictions are obtained with the KaTie Monte Carlo event generator, which allows to calculate interactions within the High Energy Factorisation framework. The tree-level matrix element results are subsequently interfaced with the CASCADE Monte Carlo event generator to account for hadronisation. The effects of the saturation of the gluon density, leading to suppression of the cross section, are investigated.
△ Less
Submitted 21 December, 2017;
originally announced December 2017.
-
Calculations with off-shell matrix elements, TMD parton densities and TMD parton showers
Authors:
Marcin Bury,
Andreas van Hameren,
Hannes Jung,
Krzysztof Kutak,
Sebastian Sapeta,
Mirko Serino
Abstract:
A new calculation using off-shell matrix elements with TMD parton densities supplemented with a newly developed initial state TMD parton shower is described. The calculation is based on the KaTie package for an automated calculation of the partonic process in high-energy factorization, making use of TMD parton densities implemented in TMDlib. The partonic events are stored in an LHE file, similar…
▽ More
A new calculation using off-shell matrix elements with TMD parton densities supplemented with a newly developed initial state TMD parton shower is described. The calculation is based on the KaTie package for an automated calculation of the partonic process in high-energy factorization, making use of TMD parton densities implemented in TMDlib. The partonic events are stored in an LHE file, similar to the conventional LHE files, but now containing the transverse momenta of the initial partons. The LHE files are read in by the CASCADE package for the full TMD parton shower, final state shower and hadronization from PYTHIA where events in HEPMC format are produced. We have determined a full set of TMD parton densities and developed an initial state TMD parton shower, including all flavors following the TMD distribution. As an example of application we have calculated the azimuthal de-correlation of high pt dijets as measured at the LHC and found very good agreement with the measurement when including initial state TMD parton showers together with conventional final state parton showers and hadronization.
△ Less
Submitted 16 December, 2017;
originally announced December 2017.
-
Efficient Similarity Search in Dynamic Data Streams
Authors:
Marc Bury,
Chris Schwiegelshohn,
Mara Sorella
Abstract:
The Jaccard index is an important similarity measure for item sets and Boolean data. On large datasets, an exact similarity computation is often infeasible for all item pairs both due to time and space constraints, giving rise to faster approximate methods. The algorithm of choice used to quickly compute the Jaccard index $\frac{\vert A \cap B \vert}{\vert A\cup B\vert}$ of two item sets $A$ and…
▽ More
The Jaccard index is an important similarity measure for item sets and Boolean data. On large datasets, an exact similarity computation is often infeasible for all item pairs both due to time and space constraints, giving rise to faster approximate methods. The algorithm of choice used to quickly compute the Jaccard index $\frac{\vert A \cap B \vert}{\vert A\cup B\vert}$ of two item sets $A$ and $B$ is usually a form of min-hashing. Most min-hashing schemes are maintainable in data streams processing only additions, but none are known to work when facing item-wise deletions. In this paper, we investigate scalable approximation algorithms for rational set similarities, a broad class of similarity measures including Jaccard. Motivated by a result of Chierichetti and Kumar [J. ACM 2015] who showed any rational set similarity $S$ admits a locality sensitive hashing (LSH) scheme if and only if the corresponding distance $1-S$ is a metric, we can show that there exists a space efficient summary maintaining a $(1\pm \varepsilon)$ multiplicative approximation to $1-S$ in dynamic data streams. This in turn also yields a $\varepsilon$ additive approximation of the similarity. The existence of these approximations hints at, but does not directly imply a LSH scheme in dynamic data streams. Our second and main contribution now lies in the design of such a LSH scheme maintainable in dynamic data streams. The scheme is space efficient, easy to implement and to the best of our knowledge the first of its kind able to process deletions.
△ Less
Submitted 8 March, 2021; v1 submitted 12 May, 2016;
originally announced May 2016.
-
Single and double inclusive forward jet production at the LHC at $\sqrt{s}$ = 7 and 13 TeV
Authors:
Marcin Bury,
Michal Deak,
Krzysztof Kutak,
Sebastian Sapeta
Abstract:
We provide a description of the transverse momentum spectrum of single inclusive forward jets produced at the LHC, at the center-of-mass energies of 7 and 13 TeV, using the high energy factorization (HEF) framework. We subsequently study double inclusive forward jet production and, in particular, we calculate contributions to azimuthal angle distributions coming from double parton scattering. We a…
▽ More
We provide a description of the transverse momentum spectrum of single inclusive forward jets produced at the LHC, at the center-of-mass energies of 7 and 13 TeV, using the high energy factorization (HEF) framework. We subsequently study double inclusive forward jet production and, in particular, we calculate contributions to azimuthal angle distributions coming from double parton scattering. We also compare our results for double inclusive jet production to those obtained with the Pythia Monte Carlo generator. This comparison confirms that the HEF resummation acts like an initial state parton shower. It also points towards the need to include final state radiation effects in the HEF formalism.
△ Less
Submitted 11 August, 2016; v1 submitted 5 April, 2016;
originally announced April 2016.
-
Sublinear Estimation of Weighted Matchings in Dynamic Data Streams
Authors:
Marc Bury,
Chris Schwiegelshohn
Abstract:
This paper presents an algorithm for estimating the weight of a maximum weighted matching by augmenting any estimation routine for the size of an unweighted matching. The algorithm is implementable in any streaming model including dynamic graph streams. We also give the first constant estimation for the maximum matching size in a dynamic graph stream for planar graphs (or any graph with bounded ar…
▽ More
This paper presents an algorithm for estimating the weight of a maximum weighted matching by augmenting any estimation routine for the size of an unweighted matching. The algorithm is implementable in any streaming model including dynamic graph streams. We also give the first constant estimation for the maximum matching size in a dynamic graph stream for planar graphs (or any graph with bounded arboricity) using $\tilde{O}(n^{4/5})$ space which also extends to weighted matching. Using previous results by Kapralov, Khanna, and Sudan (2014) we obtain a $\mathrm{polylog}(n)$ approximation for general graphs using $\mathrm{polylog}(n)$ space in random order streams, respectively. In addition, we give a space lower bound of $Ω(n^{1-\varepsilon})$ for any randomized algorithm estimating the size of a maximum matching up to a $1+O(\varepsilon)$ factor for adversarial streams.
△ Less
Submitted 9 July, 2015; v1 submitted 8 May, 2015;
originally announced May 2015.
-
OBDDs and (Almost) $k$-wise Independent Random Variables
Authors:
Marc Bury
Abstract:
OBDD-based graph algorithms deal with the characteristic function of the edge set E of a graph $G = (V,E)$ which is represented by an OBDD and solve optimization problems by mainly using functional operations. We present an OBDD-based algorithm which uses randomization for the first time. In particular, we give a maximal matching algorithm with $O(\log^3 \vert V \vert)$ functional operations in ex…
▽ More
OBDD-based graph algorithms deal with the characteristic function of the edge set E of a graph $G = (V,E)$ which is represented by an OBDD and solve optimization problems by mainly using functional operations. We present an OBDD-based algorithm which uses randomization for the first time. In particular, we give a maximal matching algorithm with $O(\log^3 \vert V \vert)$ functional operations in expectation. This algorithm may be of independent interest. The experimental evaluation shows that this algorithm outperforms known OBDD-based algorithms for the maximal matching problem.
In order to use randomization, we investigate the OBDD complexity of $2^n$ (almost) $k$-wise independent binary random variables. We give a OBDD construction of size $O(n)$ for $3$-wise independent random variables and show a lower bound of $2^{Ω(n)}$ on the OBDD size for $k \geq 4$. The best known lower bound was $Ω(2^n/n)$ for $k \approx \log n$ due to Kabanets. We also give a very simple construction of $2^n$ $(\varepsilon, k)$-wise independent binary random variables by constructing a random OBDD of width $O(n k^2/\varepsilon)$.
△ Less
Submitted 15 April, 2015;
originally announced April 2015.
-
Random Projections for k-Means: Maintaining Coresets Beyond Merge & Reduce
Authors:
Marc Bury,
Chris Schwiegelshohn
Abstract:
We give a new construction for a small space summary satisfying the coreset guarantee of a data set with respect to the $k$-means objective function. The number of points required in an offline construction is in $\tilde{O}(k ε^{-2}\min(d,kε^{-2}))$ which is minimal among all available constructions.
Aside from two constructions with exponential dependence on the dimension, all known coresets ar…
▽ More
We give a new construction for a small space summary satisfying the coreset guarantee of a data set with respect to the $k$-means objective function. The number of points required in an offline construction is in $\tilde{O}(k ε^{-2}\min(d,kε^{-2}))$ which is minimal among all available constructions.
Aside from two constructions with exponential dependence on the dimension, all known coresets are maintained in data streams via the merge and reduce framework, which incurs are large space dependency on $\log n$. Instead, our construction crucially relies on Johnson-Lindenstrauss type embeddings which combined with results from online algorithms give us a new technique for efficiently maintaining coresets in data streams without relying on merge and reduce. The final number of points stored by our algorithm in a data stream is in $\tilde{O}(k^2 ε^{-2} \log^2 n \min(d,kε^{-2}))$.
△ Less
Submitted 18 February, 2020; v1 submitted 7 April, 2015;
originally announced April 2015.
-
Numerical evaluation of multi-gluon amplitudes for High Energy Factorization
Authors:
M. Bury,
A. van Hameren
Abstract:
We present a program to evaluate tree-level multi-gluon amplitudes with up to two of them off-shell. Furthermore, it evaluates squared amplitudes summed over colors and helicities for up to six external gluons. It employs both analytic expressions, obtained via BCFW recursion, and numerical BCFW recursion. It has been validated numerically with the help of an independent program employing numerica…
▽ More
We present a program to evaluate tree-level multi-gluon amplitudes with up to two of them off-shell. Furthermore, it evaluates squared amplitudes summed over colors and helicities for up to six external gluons. It employs both analytic expressions, obtained via BCFW recursion, and numerical BCFW recursion. It has been validated numerically with the help of an independent program employing numerical Dyson-Schwinger recursion.
△ Less
Submitted 17 June, 2015; v1 submitted 30 March, 2015;
originally announced March 2015.
