Search

By nature, transmissible human knowledge is enumerable: every sentence, movie, audio record can be encoded in a sufficiently long string of 0's and 1's. The works of G\"odel, Turing and others showed that there are inherent limits and properties associated with the fact that language technology is enumerable. G\"odel's numbering technique is universal for enumerable structures and shows strong limits of the language technology. Computability theory is a particular example: programs can be numbered and all sorts of limits can be studied from there. Computability is also at the heart of science since any experimental validation of a theory supposes that theoretical results have been computed, then checked against concrete experiments. It implies that limitations on what is computable ultimately are also limits of what we understand as "scientific theory", and more generally to all the transmissible knowledge. We argue that it is fruitful to look a epistemology from a computability perspective. We show that it allows to precisely define different kinds of knowledge acquisition techniques, and helps the study of how they are related to one another.

Transverse 2D phase space distribution of a 2.1 MeV, 5 mA H$^-$ beam is measured at the PIPIT test accelerator at Fermilab with an Allison scanner. The paper describes the design, calibration, and performance of the scanner as well as the main results of the beam measurements. Analyses of the recorded phase portraits are performed primarily in action-phase coordinates; the stability of the action under linear optics makes it easier to compare measurements taken with different beamline conditions, e.g. in various locations. The intensity of a single measured point (\pixel") is proportional to the phase density in the corresponding portion of the beam. When the Twiss parameters are calculated using only the high-phase density part of the beam, the pixel intensity in the beam core is found to be decreasing exponentially with action and to be phase-independent. Outside of the core, the intensities decrease with action at a significantly slower rate than in the core. This `tail' comprises 10-30% of the beam, with 0.1% of the total measured intensity extending beyond the action 10-20 times larger than the rms emittance. The transition from the core to the tail is accompanied by the appearance of a strong phase dependence, which is characterized in action-phase coordinates by two `branches' extending beyond the core. A set of selected measurements shows, in part, that there is no measurable emittance dilution along the beam line in the main portion of the beam; the beam parameters are practically constant over a 0.5 ms pulse; and scraping in various parts of the beam line is an effective way to decrease the transverse tails by removing the branches.

The warm front end of the PIP2IT accelerator, assem-bled and commissioned at Fermilab, consists of a 15 mA DC, 30 keV H- ion source, a 2 m long Low Energy Beam Transport (LEBT) line, and a 2.1 MeV, 162.5 MHz CW RFQ, followed by a 10 m long Medium Energy Beam Transport (MEBT) line. A part of the commissioning efforts involves operation with the average beam power emulating the operation of the proposed PIP-II accelera-tor, which will have a duty factor of 1.1% or above. The maximum achieved power is 5 kW (2.1 MeV x 5 mA x 25 ms x 20 Hz). This paper describes the difficulties encoun-tered and some of the solutions that were implemented., Comment: 4 pages, NAPAC2019

Short laser pulses are widely used for controlling molecular rotational degrees of freedom and inducing molecular alignment, orientation, unidirectional rotation and other types of coherent rotational motion. To follow the ultra-fast rotational dynamics in real time, several techniques for producing molecular movies have been proposed based on the Coulomb explosion of rotating molecules, or recovering molecular orientation from the angular distribution of high-harmonics. The present work offers and demonstrates a novel non-destructive optical method for direct visualization and recording of movies of coherent rotational dynamics in a molecular gas. The technique is based on imaging the time-dependent polarization dynamics of a probe light propagating through a gas of coherently rotating molecules. The probe pulse continues through a radial polarizer, and is then recorded by a camera. We illustrate the technique by implementing it with two examples of time-resolved rotational dynamics: alignment-antialignment cycles in a molecular gas excited by a single linearly polarized laser pulse, and unidirectional molecular rotation induced by a pulse with twisted polarization. This method may open new avenues in studies on fast chemical transformation phenomena and ultrafast molecular dynamics caused by strong laser fields of various complexities.

The central part of PIP-II program of upgrades pro-posed for the Fermilab injection complex is an 800 MeV, 2 mA, CW-compatible SRF linac. Acceleration in super-conducting cavities begins from a low energy of 2.1 MeV, so that the first cryomodule, Half Wave Resonator (HWR) borders the warm Medium Beam Transport (MEBT) line. To minimize the amount of gas that may enter the SRF linac in a case if a vacuum failure occurs in the warm front end, a vacuum protection system is envisioned to be used in the PIP-II MEBT. It features a fast closing valve with two sensors and a differential pumping insert. The system prototype was installed in the PIP-II Injector Test (PIP2IT) accelerator and successfully tested in several modes modelling the vacuum failures. The report presents the design of the vacuum protection system and results of its tests., Comment: 3 pages, NAPAC2019

How can we build recommender systems to take into account fairness? Real-world recommender systems are often composed of multiple models, built by multiple teams. However, most research on fairness focuses on improving fairness in a single model. Further, recent research on classification fairness has shown that combining multiple "fair" classifiers can still result in an "unfair" classification system. This presents a significant challenge: how do we understand and improve fairness in recommender systems composed of multiple components? In this paper, we study the compositionality of recommender fairness. We consider two recently proposed fairness ranking metrics: equality of exposure and pairwise ranking accuracy. While we show that fairness in recommendation is not guaranteed to compose, we provide theory for a set of conditions under which fairness of individual models does compose. We then present an analytical framework for both understanding whether a real system's signals can achieve compositional fairness, and improving which component would have the greatest impact on the fairness of the overall system. In addition to the theoretical results, we find on multiple datasets -- including a large-scale real-world recommender system -- that the overall system's end-to-end fairness is largely achievable by improving fairness in individual components., Comment: WSDM 2021

As recent literature has demonstrated how classifiers often carry unintended biases toward some subgroups, deploying machine learned models to users demands careful consideration of the social consequences. How should we address this problem in a real-world system? How should we balance core performance and fairness metrics? In this paper, we introduce a MinDiff framework for regularizing classifiers toward different fairness metrics and analyze a technique with kernel-based statistical dependency tests. We run a thorough study on an academic dataset to compare the Pareto frontier achieved by different regularization approaches, and apply our kernel-based method to two large-scale industrial systems demonstrating real-world improvements.

We consider the recommendations of the World Wide Web Consortium (W3C) about the Resource Description Framework (RDF) and the associated query language SPARQL. We propose a new formal framework based on category theory which provides clear and concise formal definitions of the main basic features of RDF and SPARQL. We propose to define the notions of RDF graphs as well as SPARQL basic graph patterns as objects of some nested categories. This allows one to clarify, in particular, the role of blank nodes. Furthermore, we consider basic SPARQL CONSTRUCT and SELECT queries and formalize their operational semantics following a novel algebraic graph transformation approach called POIM.

We present a minimal model to study liquid phase separation in a fixed pH ensemble. The model describes a mixture composed of macromolecules that exist in three different charge states and have a tendency to phase separate. We introduce the pH dependence of phase separation by means of a set of reactions describing the protonation and deprotonation of macromolecules, as well as the self-ionisation of water. We use conservation laws to identify the conjugate thermodynamic variables at chemical equilibrium. Using this thermodynamic conjugate variables we perform a Legendre transform which defines the corresponding free energy at fixed pH. We first study the possible phase diagram topologies at the isoelectric point of the macromolecules. We then show how the phase behavior depends on pH by moving away from the isoelectric point. We find that phase diagrams as a function of pH strongly depend on whether oppositely charged macromolecules or neutral macromolecules have a stronger tendency to phase separate. We predict the existence of reentrant behavior as a function of pH. In addition, our model also predicts that the region of phase separation is typically broader at the isoelectric point. This model could account for both, the protein separation observed in yeast cells for pH values close to the isoelectric point of many cytosolic proteins and also for the in vitro experiments of single proteins exhibiting phase separation as a function of pH.

We explore how a magnetic field breaks the symmetry of an initially spherical giant liposome filled with a magnetic colloid. The condition of rotational symmetry along the field axis leads either to a prolate or to an oblate ellipsoid. We demonstrate that an electrostatic interaction between the nanoparticles and the membrane triggers the shape transition. PACS numbers: 75.

(Bolukbasi et al., 2016) demonstrated that pretrained word embeddings can inherit gender bias from the data they were trained on. We investigate how this bias affects downstream classification tasks, using the case study of occupation classification (De-Arteaga et al.,2019). We show that traditional techniques for debiasing embeddings can actually worsen the bias of the downstream classifier by providing a less noisy channel for communicating gender information. With a relatively minor adjustment, however, we show how these same techniques can be used to simultaneously reduce bias and maintain high classification accuracy.

We discuss the physical mechanisms that promote or suppress the nucleation of a fluid-filled lumen inside a cell assembly or a tissue. We discuss lumen formation in a continuum theory of tissue material properties in which the tissue is described as a two-fluid system to account for its permeation by the interstitial fluid, and we include fluid pumping as well as active electric effects. Considering a spherical geometry and a polarized tissue, our work shows that fluid pumping and tissue flexoelectricity play a crucial role in lumen formation. We furthermore explore the large variety of long-time states that are accessible for the cell aggregate and its lumen. Our work reveals a role of the coupling of mechanical, electrical and hydraulic phenomena in tissue lumen formation., Comment: Published version

We calculate the Casimir stresses in a thin layer of active fluid with nematic order. By using a stochastic hydrodynamic approach for an active fluid layer of finite thickness $L$, we generalize the Casimir stress for nematic liquid crystals in thermal equilibrium to active systems. We show that the active Casimir stress differs significantly from its equilibrium counterpart. For contractile activity, the active Casimir stress, although attractive like its equilibrium counterpart, diverges logarithmically as $L$ approaches a threshold of the spontaneous flow instability from below. In contrast, for small extensile activity, it is repulsive, has no divergence at any $L$ and has a scaling with $L$ different from its equilibrium counterpart.

In many application settings, the data have missing entries which make analysis challenging. An abundant literature addresses missing values in an inferential framework: estimating parameters and their variance from incomplete tables. Here, we consider supervised-learning settings: predicting a target when missing values appear in both training and testing data. We show the consistency of two approaches in prediction. A striking result is that the widely-used method of imputing with a constant, such as the mean prior to learning is consistent when missing values are not informative. This contrasts with inferential settings where mean imputation is pointed at for distorting the distribution of the data. That such a simple approach can be consistent is important in practice. We also show that a predictor suited for complete observations can predict optimally on incomplete data, through multiple imputation. Finally, to compare imputation with learning directly with a model that accounts for missing values, we analyze further decision trees. These can naturally tackle empirical risk minimization with missing values, due to their ability to handle the half-discrete nature of incomplete variables. After comparing theoretically and empirically different missing values strategies in trees, we recommend using the "missing incorporated in attribute" method as it can handle both non-informative and informative missing values.

We study the dynamics of a thick polar epithelium subjected to the action of both an electric and a flow field in a planar geometry. We develop a generalized continuum hydrodynamic description and describe the tissue as a two component fluid system. The cells and the interstitial fluid are the two components and we keep all terms allowed by symmetry. In particular we keep track of the cell pumping activity for both solvent flow and electric current and discuss the corresponding orders of magnitude. We study the growth dynamics of tissue slabs, their steady states and obtain the dependence of the cell velocity, net cell division rate, and cell stress on the flow strength and the applied electric field. We find that finite thickness tissue slabs exist only in a restricted region of phase space and that relatively modest electric fields or imposed external flows can induce either proliferation or death., Comment: 18 pages, 12 figures

We present a novel algorithm for Fast Registration Of image Groups (FROG), applied to large 3D image groups. Our approach extracts 3D SURF keypoints from images, computes matched pairs of keypoints and registers the group by minimizing pair distances in a hubless way i.e. without computing any central mean image. Using keypoints significantly reduces the problem complexity compared to voxel-based approaches, and enables us to provide an in-core global optimization, similar to the Bundle Adjustment for 3D reconstruction. As we aim to register images of different patients, the matching step yields many outliers. Then we propose a new EM-weighting algorithm which efficiently discards outliers. Global optimization is carried out with a fast gradient descent algorithm. This allows our approach to robustly register large datasets. The result is a set of diffeomorphic half transforms which link the volumes together and can be subsequently exploited for computational anatomy and landmark detection. We show experimental results on whole-body CT scans, with groups of up to 103 volumes. On a benchmark based on anatomical landmarks, our algorithm compares favorably with the star-groupwise voxel-based ANTs and NiftyReg approaches while being much faster. We also discuss the limitations of our approach for lower resolution images such as brain MRI.

The Proton Improvement Plan, Stage Two (PIP-II) is a program of upgrades proposed for the Fermilab injection complex, which central part is an 800 MeV, 2 mA CW SRF linac. A prototype of the PIP-II linac front end called PIP-II Injector Test (PIP2IT) is being built at Fermilab. As of now, a 15 mA DC, 30-keV H- ion source, a 2 m-long Low Energy Beam Transport (LEBT), a 2.1 MeV CW RFQ, followed by a 10 m Medium Energy Beam Transport (MEBT) have been assembled and commissioned. The MEBT bunch-by-bunch chopping system and the requirement of a low uncontrolled beam loss put stringent limitations on the beam envelope and its variation. Measurements of transverse and longitudinal beam dynamics in the MEBT were performed in the range of 1-10 mA of the RFQ beam current. Almost all measurements are made with 10 {\mu}s beam pulses in order to avoid damage to the beam line. This report presents measurements of the transverse optics with differential trajectories, reconstruction of the beam envelope with scrapers and an Allison emittance scanner, as well as bunch length measurements with a Fast Faraday Cup., Comment: 6 pp

Simplest extensions of single particle dynamics in momentum conserving active fluid - that of an active suspension of two colloidal particles or a single particle confined by a wall - exhibit strong departures from Boltzmann behavior, resulting in either a breakdown of an effective temperature description or a steady state with nonzero entropy production rate. This is a consequence of hydrodynamic interactions that introduce multiplicative noise in the stochastic description of the particle positions. This results in fluctuation induced interactions that depend on distance as a power law. We find that the dynamics of activated colloids in a passive fluid, with stochastic forcing localized on the particle, is different from that of passive colloids in an active fluctuating fluid.

The Warm Front End (WFE) of the Proton Improvement Plan II Injector Test at Fermilab has been constructed to its full length. It includes a 15-mA DC, 30-keV H- ion source, a 2 m-long Low Energy Beam Transport (LEBT) with a switching dipole magnet, a 2.1 MeV CW RFQ, followed by a Medium Energy Beam Transport (MEBT) with various diagnostics and a dump. This report presents the commissioning status, focusing on beam measurements in the MEBT. In particular, a beam with the parameters required for injection into the Booster (5 mA, 0.55 ms macro-pulse at 20 Hz) was transported through the WFE., Comment: 4 pp

We theoretically explore fluidization of epithelial tissues by active T1 neighbor exchanges. We show that the geometry of cell-cell junctions encodes important information about the local features of the energy landscape, which we support by an elastic theory of T1 transformations. Using a 3D vertex model, we show that the degree of active noise driving forced cell rearrangements governs the stress-relaxation time-scale of the tissue. We study tissue response to in-plane shear at different time scales. At short time, the tissue behaves as a solid, whereas its long-time fluid behavior can be associated with an effective viscosity which scales with the rate of active T1 transformations. Furthermore, we develop a coarse-grained theory, where we treat the tissue as an active fluid and confirm the results of the vertex model. The impact of cell rearrangements on tissue shape is illustrated by studying axial compression of an epithelial tube.

In a Low Energy Beam Transport line (LEBT), the emittance growth due to the beam's space charge is typically suppressed by way of neutralization from either electrons or ions, which originate from ionization of the background gas. In cases where the beam is chopped, the neutralization pattern usually changes throughout the beginning of the pulse, causing the Twiss parameters to differ significantly from their steady state values, which, in turn, may result in beam losses downstream. For a modest beam perveance, there is an alternative solution, in which the beam is kept un-neutralized in the portion of the LEBT that contains the chopper. The emittance can be nearly preserved if the transition to the un-neutralized section occurs where the beam exhibits low transverse tails. This report introduces the rationale for the proposed scheme and formulates the physical arguments for it as well as its limitations. An experimental realization of the scheme was carried out at Fermilab's PIP2IT where low beam emittance dilution was demonstrated for a 5 mA, 30 keV $H^{-}$ beam., Comment: 10 pp

The Proton Improvement Plan II (PIP-II) at Fermilab is a program of upgrades to the injection complex. At its core is the design and construction of a CW compatible, pulsed H- SRF linac. To validate the concept of the front-end of such machine, a test accelerator known as PIP-II Injector Test (PIP2IT) is under construction. It includes a 10 mA DC, 30 keV H- ion source, a 2 m-long Low Energy Beam Transport (LEBT), a 2.1 MeV CWRFQ, followed by a Medium Energy Beam Transport (MEBT) that feeds the first of 2 cryomodules increasing the beam energy to about 25 MeV, and a High Energy Beam Transport section (HEBT) that takes the beam to a dump. The ion source, LEBT, RFQ, and initial version of the MEBT have been built, installed, and commissioned. This report presents the overall status of the warm front end., Comment: 4 pp. 8th International Particle Accelerator Conference (IPAC 2017) 14-19 May 2017. Copenhagen, Denmark

A 2.1 MeV, 10 mA CW RFQ has been installed and commissioned at Fermilab's test accelerator known as PIP-II Injector Test. This report describes the measurements of the beam properties after acceleration in the RFQ, including the energy and emittance., Comment: 4 pp. Submitted to IPAC 2017

The PIP2IT test accelerator is under construction at Fermilab. Its ion source and Low Energy Beam Transport (LEBT) in its initial (straight) configuration have been commissioned to full specification parameters. This paper introduces the LEBT design and summarizes the outcome of the commissioning activities., Comment: 11 p

In a Low Energy Beam Transport line (LEBT), the emittance growth due to the beam's own space charge is typically suppressed by way of neutralization from either electrons or ions, which originate from ionization of the background gas. In cases where the beam is chopped, the neutralization pattern changes throughout the beginning of the pulse, causing the Twiss parameters to differ significantly from their steady state values, which, in turn, may result in beam losses downstream. For a modest beam perveance, there is an alternative solution, in which the beam is kept un-neutralized in the portion of the LEBT that contains the chopper. The emittance can be nearly preserved if the transition to the un-neutralized section occurs where the beam exhibits low transverse tails. This report discusses the experimental realization of such a scheme at Fermilab's PXIE, where low beam emittance dilution was demonstrated., Comment: 3 pp

Typical front-ends of modern light-ion high-intensity accelerators typically consist of an ion source, a Low Energy Beam Transport (LEBT), a Radiofrequency Quadrupole (RFQ) and a Medium Energy Beam Transport (MEBT), which is followed by the main linac accelerating structures. Over the years, many LEBTs have been designed, constructed and operated very successfully. In this paper, we present the guiding principles and compromises that lead to the design choices of the PXIE LEBT, including the rationale for a beam line that allows un-neutralized transport over a significant portion of the LEBT whether the beam is pulsed or DC., Comment: 3 pp

We investigate the effect of stress fluctuations on the stochastic dynamics of an inclusion embedded in a viscous gel. We show that, in non-equilibrium systems, stress fluctuations give rise to an effective attraction towards the boundaries of the confining domain, which is reminiscent of an active Casimir effect. We apply this generic result to the dynamics of deformations of the cell nucleus and we demonstrate the appearance of a fluctuation maximum at a critical level of activity, in agreement with recent experiments [E. Makhija, D. S. Jokhun, and G. V. Shivashankar, Proc. Natl. Acad. Sci. U.S.A. 113, E32 (2016)]., Comment: 12 pages, 5 figures

We study stochastic dynamics of a point and extended inclusion within a one dimensional confined active viscoelastic gel. We show that the dynamics of a point inclusion can be described by a Langevin equation with a confining potential and multiplicative noise. Using a systematic adiabatic elimination over the fast variables, we arrive at an overdamped equation with a proper definition of the multiplicative noise. To highlight various features and to appeal to different biological contexts, we treat the inclusion in turn as a rigid extended element, an elastic element and a viscoelastic (Kelvin-Voigt) element. The dynamics for the shape and position of the extended inclusion can be described by coupled Langevin equations. Deriving exact expressions for the corresponding steady state probability distributions, we find that the active noise induces an attraction to the edges of the confining domain. In the presence of a competing centering force, we find that the shape of the probability distribution exhibits a sharp transition upon varying the amplitude of the active noise. Our results could help understanding the positioning and deformability of biological inclusions, eg. organelles in cells, or nucleus and cells within tissues., Comment: 16 pages, 9 figures