Data-driven Constraints Research Articles

New Data-Driven Constraints on the Sign of Gluon Polarization in the Proton.

Recently, the possible existence of negative gluon helicity Δg has been observed to be compatible with existing empirical constraints, including from jet production in polarized proton-proton collisions at the Relativistic Heavy Ion Collider, and lattice QCD data on polarized gluon Ioffe time distributions. We perform a new global analysis of polarized parton distributions in the proton with new constraints from the high-x region of deep-inelastic scattering (DIS). A dramatic reduction in the quality of the fit for the negative Δg replicas compared to those with positive Δg suggests that the negative Δg solution cannot simultaneously account for high-x polarized DIS data along with lattice and polarized jet data.

Consistent Eccentricities for Gravitational-wave Astronomy: Resolving Discrepancies between Astrophysical Simulations and Waveform Models

Detecting imprints of orbital eccentricity in gravitational-wave (GW) signals promises to shed light on the formation mechanisms of binary black holes. To constrain the formation mechanisms, distributions of eccentricity derived from numerical simulations of astrophysical formation channels are compared to the estimates of eccentricity inferred from GW signals. We report that the definition of eccentricity typically used in astrophysical simulations is inconsistent with the one used while modeling GW signals, with the differences mainly arising due to the choice of reference frequency used in both cases. We also posit a prescription for calculating eccentricity from astrophysical simulations, by evolving ordinary differential equations obtained from post-Newtonian theory and using the dominant (ℓ = m = 2) mode’s frequency as the reference frequency; this ensures consistency in the definitions. On comparing the existing eccentricities of the binaries present in the Cluster Monte Carlo catalog of globular cluster simulations with the eccentricities calculated using the prescription presented here, we find a significant discrepancy at e ≳ 0.2; this discrepancy becomes worse with increasing eccentricity. We note the implications this discrepancy has for existing studies and recommend that care be taken when comparing data-driven constraints on eccentricity to expectations from astrophysical formation channels.

Data-driven constraints on cosmic-ray diffusion: Probing self-generated turbulence in the Milky Way

Data-driven constraints on cosmic-ray diffusion: Probing self-generated turbulence in the Milky Way

Data-Driven Minimax Optimization with Expectation Constraints

The realm of data-driven optimization has garnered substantial attention in recent years. However, there remains a paucity of research addressing challenges posed by intricate data-driven constraints. In their paper titled “Data-Driven Minimax Optimization with Expectation Constraints,” Shuoguang Yang, Xudong Li, and Guanghui Lan concentrate on the realm of data-driven minimax optimization while considering expectation constraints. To grapple with these complex models, the authors introduce a novel class of optimal primal-dual algorithms. Their work showcases the practical efficacy of these algorithms through the resolution of real-world problems, including data-driven robust pricing and the maximization of the area under the ROC curve (AUC) while adhering to fairness constraints.

Human-Centric Parcel Delivery at Deutsche Post with Operations Research and Machine Learning

Features such as estimated delivery time windows and live tracking of shipments play a key role in improving the customer experience in last-mile delivery. The building blocks for enabling these features are reliable knowledge about the expected order of deliveries in a tour and precise delivery time window predictions. For Deutsche Post’s parcel delivery service in Germany, we developed a courier-centric routing algorithm and a corresponding state-of-the-art machine learning model for delivery time window predictions. The routing algorithm combines operations research with statistics and machine learning to implicitly gather and use the tacit knowledge of our experienced couriers within the tour generation. This is achieved by deducing and selecting appropriate precedence constraints from historical delivery data. This novel combination of optimization with data-driven constraints enabled us to provide custom routes to the individual couriers. It proved to be a main driver allowing us to provide accurate delivery time window predictions and live tracking of shipments. Our solution is used by Deutsche Post to plan the daily routes of couriers to the approximately 13,000 parcel delivery districts in Germany as well as to provide live tracking and estimated delivery time windows for 1.6 million parcels each day. History: This paper has been accepted for the INFORMS Journal on Applied Analytics Special Issue—2022 Daniel H. Wagner Prize for Excellence in the Practice of Advanced Analytics and Operations Research.

Evaluate DC Meter Adoption for House-Level Storage Devices

Although batteries are increasingly adopted in individual households, utilities typically do not know the real behaviors of the customer-owned batteries. Therefore, it is hard for the utilities to evaluate the necessity of adding a DC meter on the DC side of the battery. Meanwhile, the customers do not know the benefits they can get, so they cannot make an adoption decision of DC meters. To solve these practical problems, this paper aims to provide a DC meter evaluation tool for utilities and customers to calculate their costs and revenues. Specifically, we formulate a bi-level optimization framework that considers the battery incentive design and physical law simultaneously. To reflect the reality, the optimization is also based on data-driven constraints based on big utility data and accurate performance. While the optimization problem is complex, we enforce convexity via various designs to provide the optimal solution for incentive planning. Through simulation, the battery incentive design model is tested to be valid under different market rates and case studies. The proposed optimization model provides a promising tool for utilities and customers to evaluate DC meter adoption decisions.

From inflation to black hole mergers and back again: Gravitational-wave data-driven constraints on inflationary scenarios with a first-principle model of primordial black holes across the QCD epoch

Recent population studies have searched for a subpopulation of primordial black holes (PBHs) in the gravitational-wave (GW) events so far detected by LIGO/Virgo/KAGRA (LVK), in most cases adopting a phenomenological PBH mass distribution. When deriving such population from first principles in the standard scenario, however, the equation of state of the Universe at the time of PBH formation may strongly affect the PBH abundance and mass distribution, which ultimately depend on the power spectrum of cosmological perturbations. Here we improve on previous population studies on several aspects: (i) we adopt state-of-the-art PBH formation models describing the collapse of cosmological perturbations across the QCD epoch; (ii) we perform the first Bayesian multi-population inference on GW data including PBHs and directly using power spectrum parameters instead of phenomenological distributions; (iii) we critically confront the PBH scenario with LVK phenomenological models describing the GWTC-3 catalog both in the neutron-star and in the BH mass ranges, also considering PBHs as subpopulation of the total events. Our results confirm that LVK observations prevent the majority of the dark matter to be in the form of stellar mass PBHs. We find that the best fit PBH model can comprise a small fraction of the total events, in particular it can naturally explain events in the mass gaps. If the lower mass-gap event GW190814 is interpreted as a PBH binary, we predict that LVK should detect up to a few subsolar mergers and one to $\approx 30$ lower mass gap events during the upcoming O4 and O5 runs. Finally, mapping back the best-fit power spectrum into an ultra slow-roll inflationary scenario, we show that the latter predicts detectable PBH mergers in the LVK band, a stochastic GW background detectable by current and future instruments, and may include the entirety of dark matter in asteroid-mass PBHs.

A novel estimator of the interaction matrix in Graphical Gaussian Model of omics data using the entropy of non-equilibrium systems.

Inferring the direct relationships between biomolecules from omics datasets is essential for the understanding of biological and disease mechanisms. Gaussian Graphical Model (GGM) provides a fairly simple and accurate representation of these interactions. However, estimation of the associated interaction matrix using data is challenging due to a high number of measured molecules and a low number of samples. In this article, we use the thermodynamic entropy of the non-equilibrium system of molecules and the data-driven constraints among their expressions to derive an analytic formula for the interaction matrix of Gaussian models. Through a data simulation, we show that our method returns an improved estimation of the interaction matrix. Also, using the developed method, we estimate the interaction matrix associated with plasma proteome and construct the corresponding GGM and show that known NAFLD-related proteins like ADIPOQ, APOC, APOE, DPP4, CAT, GC, HP, CETP, SERPINA1, COLA1, PIGR, IGHD, SAA1 and FCGBP are among the top 15% most interacting proteins of the dataset. The supplementary materials can be found in the following URL: http://dynamic-proteome.utmb.edu/PrecisionMatrixEstimater/PrecisionMatrixEstimater.aspx. Supplementary data are available at Bioinformatics online.

A Hybrid Enhanced Scatter Search—Composite I-Distance Indicator (eSS-CIDI) Optimization Approach for Determining Weights Within Composite Indicators

Considering the impact composite indicators can have on public opinion and policy development, the need for their frameworks to be methodologically sound and statistically verified is growing daily. One of the issues in the process of composite indicator construction which has generated much debate is how to choose the weighting scheme. To address this slippery step, we propose an optimization approach based on the enhanced Scatter Search (eSS) metaheuristic. In this paper, the eSS algorithm is applied to obtain a weighting scheme which will increase the stability of the composite indicator. The objective function is based on the relative contributions of indicators, while the problem constraints rely on the bootstrap Composite I-distance Indicator (CIDI) approach which is also proposed herein. The eSS-CIDI approach combines the exploration capability of eSS and the data-driven constraints devised from the bootstrap CIDI. This novel weighting approach was tested on two acknowledged composite indicators: the Academic Ranking of World Universities (ARWU) and the Networked Readiness Index (NRI). Results indicate that the composite indicators created using the eSS-CIDI weighting approach are more stable than the official ones.

Basis Pursuit Denoise With Nonsmooth Constraints

Level-set optimization formulations with data-driven constraints minimize a regularization functional subject to matching observations to a given error level. These formulations are widely used, particularly for matrix completion and sparsity promotion in data interpolation and denoising. The misfit level is typically measured in the ℓ <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sub> norm, or other smooth metrics. In this paper, we present a new flexible algorithmic framework that targets nonsmooth level-set constraints, including ℓ <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1</sub> , ℓ <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">∞</sub> , and even ℓ <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">0</sub> norms. These constraints give greater flexibility for modeling deviations in observation and denoising, and have significant impact on the solution. Measuring error in the ℓ <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1</sub> and ℓ <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">0</sub> norms makes the result more robust to large outliers, while matching many observations exactly. We demonstrate the approach for basis pursuit denoise (BPDN) problems as well as for extensions of BPDN to matrix factorization, with applications to interpolation and denoising of 4D seismic data. The new methods are particularly promising for seismic applications, where the amplitude in the data varies significantly, and measurement noise in low-amplitude regions can wreak havoc for standard Gaussian error models.

Constraint nip-tomographic inversion of strong sparse seismic data

This work is a result of specific numerical experiments motivated by real cases of processing strong sparse seismic data, as an application of techniques based on the common-reflection-surface (CRS) stack technology aiming at estimating a smooth velocity depth distribution. The paper is primarily limited to numerical tests with a depth velocity model that attends closely the paraxial theory validated by the seismic ray hypotheses. A complete modeling of a seismic survey was performed, and the common-shot sections were submitted to random muting of traces, to noise addition, and afterwards followed by reconstruction of the section by trace interpolation. The interpolation was controlled by the 2D spectral non-aliasing condition, where the t − x spectral amplitude content was limited to the two main Fourier quadrants f − k. It was admitted that most information was based on primary compressional (P) wave content; therefore, multiples and the P − S conversion were considered as noise. The trace interpolation used the stack attributes of the original gather (conventional stack) with sparse data to construct supergather sections (for the supergather stack). The velocity distribution in depth uses the principle of interpreting the inversion data as normal incidence point (NPI) information. The applied inversion algorithm is NIP-tomographic, classified as curve fitting, non-linear, multi-parametric, that uses the wave front kinematic and dynamic CRS attributes as data-driven constraints to estimate a consistent depth velocity distribution. As a general conclusion, we emphasized also interpolation, inclusive of sparse data, as a step for spectral analysis, consequently in filtering, stacking, and tomography to obtain a velocity distribution for further use in the estimation for velocity distribution, imaging, geological interpretation and sedimentary basin modeling.

Learning with Weak Supervision from Physics and Data‐Driven Constraints

In many applications of machine learning, labeled data is scarce and obtaining additional labels is expensive. We introduce a new approach to supervising learning algorithms without labels by enforcing a small number of domain‐specific constraints over the algorithms' outputs. The constraints can be provided explicitly based on prior knowledge — for example, we may require that objects detected in videos satisfy the laws of physics — or implicitly extracted from data using a novel framework inspired by adversarial training. We demonstrate the effectiveness of constraint‐based learning on a variety of tasks — including tracking, object detection, and human pose estimation — and we find that algorithms supervised with constraints achieve high accuracies with only a small number of labels, or with no labels at all in some cases.

Electron neutrino appearance in the NOvA experiment

NOvA is an off-axis, two-detector experiment studying neutrino oscillations with the νµ beam from Fermilab. This paper describes the νe appearance analysis, including data-driven constraints from Near Detector measurements. The data set corresponds to an exposure equivalent to 6.05 × 1020 protons-on-target in the Far Detector. We observed 33 νe candidates with a predicted background of 8.2 ± 0.8 (syst.), for a significance of appearance higher than 8σ. Preliminary results for the allowed values of δCP and θ23 in both hierarchies are presented.

Determining the population properties of spinning black holes

There are at least two formation scenarios consistent with the first gravitational-wave observations of binary black hole mergers. In field models, black hole binaries are formed from stellar binaries that may undergo common envelope evolution. In dynamic models, black hole binaries are formed through capture events in globular clusters. Both classes of models are subject to significant theoretical uncertainties. Nonetheless, the conventional wisdom holds that the distribution of spin orientations of dynamically merging black holes is nearly isotropic while field-model black holes prefer to spin in alignment with the orbital angular momentum. We present a framework in which observations of black hole mergers can be used to measure ensemble properties of black hole spin such as the typical black hole spin misalignment. We show how to obtain constraints on population hyperparameters using minimal assumptions so that the results are not strongly dependent on the uncertain physics of formation models. These data-driven constraints will facilitate tests of theoretical models and help determine the formation history of binary black holes using information encoded in their observed spins. We demonstrate that the ensemble properties of binary detections can be used to search for and characterize the properties of two distinct populations of black hole mergers.

Constraining multiplicative bias in CFHTLenS weak lensing shear data

Several recent cosmological analyses have found tension between constraints derived from the Canada-France-Hawaii Telescope Lensing Survey (CFHTLenS) data and those derived from other data sets, such as the Planck cosmic microwave background (CMB) temperature anisotropies. Similarly, a direct cross-correlation of the CFHTLenS data with Planck CMB lensing data yielded an anomalously low amplitude compared to expectations based on Planck or WMAP-derived cosmological parameters (Liu & Hill 2015). One potential explanation for these results is a multiplicative bias afflicting the CFHTLenS galaxy shape measurements, from which shears are inferred. Simulations are used in the CFHTLenS pipeline to calibrate such biases, but no data-driven constraints have been presented to date. In this paper, we cross-correlate CFHTLenS galaxy density maps with CFHTLenS shear maps and Planck CMB lensing maps to calibrate an additional multiplicative shear bias (m) in CFHTLenS (beyond the multiplicative correction that has already been applied to the CFHTLenS galaxy shears), following methods suggested by Vallinotto (2012) and Das et al. (2013). We analyze three magnitude-limited galaxy samples, finding 2-4sigma evidence for m<1 using the deepest sample (i<24), while the others are consistent with m=1 (no bias). This matches the expectation that the shapes of faint galaxies are the most prone to measurement biases. Our results for m are essentially independent of the assumed cosmology, and only weakly sensitive to assumptions about the galaxy bias. We consider three galaxy bias models, finding in all cases that the best-fit multiplicative shear bias is less than unity. A value of m~0.9 would suffice to reconcile the amplitude of density fluctuations inferred from the CFHTLenS shear two-point statistics with that inferred from Planck CMB temperature data. This scenario is consistent with our results.

Dosimetric and Toxicity Analyses of Reirradiation for Recurrent Pediatric Brain Tumors

Background: Reirradiation is increasingly used to treat recurrent pediatric brain tumors, but data-driven dose-volume constraints for repeat radiation are lacking. Methods: Records of 12 pediatric patients treated with reirradiation for recurrent brain tumors between July 2009 and May 2013 at MD Anderson Cancer Center were retrospectively reviewed for toxicity and outcomes. To determine dosimetric parameters, Digital Imaging and Communications in Medicine (DICOM) datasets of the initial and repeat radiation plans were deformed and merged to determine the maximum dose to 0.03 cc of the optic chiasm, optic nerves, spinal cord, brainstem, cochleae, pituitary, and normal brain and to 1 cc of the brainstem on the individual and composite plans. Results: Median follow-up was 3.5 years. Median age at initial radiation was 4.5 years and 6.7 years at repeat radiation. Patients had medulloblastoma (n = 4), primitive neuroectodermal tumor (PNET) (n = 2), anaplastic ependymoma (n = 2), or other tumors (n = 4). All patients initially received proton radiotherapy to a median dose of 55.8 cobalt grey equivalent (CGE). At recurrence, patients were treated with intensity-modulated radiation therapy (IMRT) (n = 6), proton (n = 5), or both (n = 1) to a median total dose of 42.5 CGE. All patients completed the planned second course of radiation. At last follow-up, four patients were alive with disease, five were dead, and three had no evidence of disease. No patient developed radiation necrosis. Two patients developed optic pathway defects, likely related to tumor progression. Four patients developed secondary hypothyroidism (median composite maximum dose of 36 CGE to pituitary), and one patient developed growth hormone deficiency (composite maximum dose of 39 CGE to pituitary). Conclusion: Repeat radiation for recurrent brain tumors may be performed in the pediatric population with acceptable shortand long-term toxicity. Establishment of dose-volume guidelines will facilitate treatment planning for these challenging cases.

Understanding the Links between Labour and Economic Development

The World Development Report of 2013 places labour in the spotlight of development research and policymaking today. Yet, there are few systematic analyses of the multifaceted nature of the link between labour and economic development. This Special Issue identifies some analytical and data-driven constraints to advances in our understanding of the role of labour in economic development, highlights some new paradigms, and offers new interpretations of phenomena in the interrelated areas of labour informality, self-employment, internal (rural–urban) and international migration, and labour force discouragement. It re-emphasizes the postulate that the mobility of labour from low-productivity towards higher-productivity jobs both geographically and across sectors and enterprises is a crucial ingredient in ensuring sustainable growth and poverty alleviation. Hence, government (and international) policy effort should focus on dismantling institutional constraints to this successful transition.

Maximum likelihood estimation in constrained parameter spaces for mixtures of factor analyzers

Mixtures of factor analyzers are becoming more and more popular in the area of model based clustering of multivariate data. According to the likelihood approach in data modeling, it is well known that the unconstrained likelihood function may present spurious maxima and singularities. To reduce such drawbacks, in this paper we introduce a procedure for parameter estimation of mixtures of factor analyzers, which maximizes the likelihood function under the mild requirement that the eigenvalues of the covariance matrices lie into some interval [a,b]. Moreover, we give a recipe on how to select appropriate bounds for the constrained EM algorithm, directly from the handled data. We then analyze and measure its performance, compared with the usual non-constrained approach, and also with other constrained models in the literature. Results show that the data-driven constraints improve the estimation and the subsequent classification, at the same time.

WMAP7 constraints on oscillations in the primordial power spectrum

We use the WMAP 7 data to place constraints on oscillations supplementing an almost scale-invariant primordial power spectrum. Such oscillations are predicted by a variety of models, some of which amount to assuming there is some non-trivial choice of the vacuum state at the onset of inflation. In this paper we will explore data-driven constraints on two distinct models of initial state modifications. In both models the frequency, phase and amplitude are degrees of freedom of the theory for which the theoretical bounds are rather weak: both the amplitude and frequency have allowed values ranging over several orders of magnitude. This requires many computationally expensive evaluations of the model CMB spectra and their goodness-of-fit, even in a Markov Chain Monte Carlo (MCMC), normally the most efficient fitting method for such a problem. To search more efficiently we first run a densely spaced grid, with only 3 varying parameters; the frequency, the amplitude and the baryon density. We obtain the optimal frequency and run an MCMC at the best fit frequency, randomly varying all other relevant parameters. To reduce the computational time of each power spectrum computation, we adjust both comoving momentum integration and spline interpolation (in l) as a function of frequency and amplitude of the primordial power spectrum. Applying this to the WMAP 7 data allows us to improve existing constraints on the presence of oscillations. We confirm earlier findings that certain frequencies can improve the fitting over a model without oscillations. For those frequencies we compute the posterior probability, allowing us to put some constraints on the primordial parameter space of both models.

Tracking Nonstationary Visual Appearances by Data-Driven Adaptation

Without any prior about the target, the appearance is usually the only cue available in visual tracking. However, in general, the appearances are often nonstationary which may ruin the predefined visual measurements and often lead to tracking failure in practice. Thus, a natural solution is to adapt the observation model to the nonstationary appearances. However, this idea is threatened by the risk of adaptation drift that originates in its ill-posed nature, unless good data-driven constraints are imposed. Different from most existing adaptation schemes, we enforce three novel constraints for the optimal adaptation: 1) negative data, 2) bottom-up pair-wise data constraints, and 3) adaptation dynamics. Substantializing the general adaptation problem as a subspace adaptation problem, this paper presents a closed-form solution as well as a practical iterative algorithm for subspace tracking. Extensive experiments have demonstrated that the proposed approach can largely alleviate adaptation drift and achieve better tracking results for a large variety of nonstationary scenes.