Showing 1–50 of 186 results for author: Phillips, A

Lubrication-mediated rebounds off fluid baths

Authors: Katie A Phillips, Paul A Milewski

Abstract: We present herein the derivation of a lubrication-mediated (LM) quasi-potential model for droplet rebounds off deep liquid baths, assuming the presence of a persistent dynamic air layer which acts as a lubricating pressure transfer. We then present numerical simulations of the LM model for axisymmetric rebounds of solid spheres and compare quantitatively to current results in the literature, inclu… ▽ More We present herein the derivation of a lubrication-mediated (LM) quasi-potential model for droplet rebounds off deep liquid baths, assuming the presence of a persistent dynamic air layer which acts as a lubricating pressure transfer. We then present numerical simulations of the LM model for axisymmetric rebounds of solid spheres and compare quantitatively to current results in the literature, including experimental data in the low speed impact regime. In this regime the LM model has the advantage of being far more computationally tractable than DNS and is also able to provide detailed behaviour within the micro-metric thin lubrication region. The LM system has an interesting mathematical structure, with the lubrication layer providing a free boundary elliptic problem mediating the drop and bath free-boundary evolutionary equations. △ Less

Submitted 24 June, 2024; originally announced June 2024.

Modelling droplet rebounds off fluid baths

Authors: Katie A Phillips, Radu Cimpeanu, Paul A Milewski

Abstract: In order for a droplet to rebound rather than coalesce from a liquid surface, there must be a layer of air persisting throughout the impact. This air layer acts as a lubricant to the system and permits a pressure transfer between the two liquid bodies. Through considering the bath, air, and drop regions of fluid, we introduce a fully coupled reduced dynamic model of two-dimensional droplets reboun… ▽ More In order for a droplet to rebound rather than coalesce from a liquid surface, there must be a layer of air persisting throughout the impact. This air layer acts as a lubricant to the system and permits a pressure transfer between the two liquid bodies. Through considering the bath, air, and drop regions of fluid, we introduce a fully coupled reduced dynamic model of two-dimensional droplets rebounding off liquid baths, which incorporates a evolving lubricating air layer. Numerical solutions of the lubrication-mediated model are compared to widely used direct numerical simulation (DNS) of the Navier-Stokes equations. The reduced model captures rebounding dynamics well in the regime where it is most relevant: for low speed impacts of small droplets, where capillary forces are important. Numerically, the reduced model is efficient, allowing for the computation of multiple rebounds and of long time dynamics of droplets rebounding on a vibrating bath. Furthermore, the lubrication-mediated model is able to provide detailed information within the air layer such as pressure and lubrication-layer geometry, which is often omitted from reduced models, and typically obtained by higher resolution but higher cost DNS methods. △ Less

Submitted 24 June, 2024; originally announced June 2024.

Procedural Construction of Atomistic Polyurethane Block Copolymer Models for High Throughput Simulations

Authors: Dominic Robe, Adrian Menzel, Andrew W Phillips, Elnaz Hajizadeh

Abstract: In this work, methods are presented to automatically generate a fully atomistic LAMMPS models of arbitrary linear multiblock polyurethane copolymers. The routine detailed here receives as parameters the number of repeat units per hard block, the number of units in a soft block, and the number of soft blocks per chain, as well as chemical formulae of three monomers which will form the hard componen… ▽ More In this work, methods are presented to automatically generate a fully atomistic LAMMPS models of arbitrary linear multiblock polyurethane copolymers. The routine detailed here receives as parameters the number of repeat units per hard block, the number of units in a soft block, and the number of soft blocks per chain, as well as chemical formulae of three monomers which will form the hard component, soft component, and chain extender. A routine is detailed for converting the chemical structure of a free monomer to the urethane bonded repeat units in a polymer. The python package RadonPy is leveraged to assemble these units into blocks, and the blocks into copolymers. Care is taken in this work to ensure that plausible atomic charges are assigned to repeat units in different parts of the chain. The static structure factor is calculated for a variety of chemistries, and the results compared with wide angle x-ray scattering data from experiments with corresponding composition. The generated models reproduce the amorphous halo observed in the scattering data as well as some of the finer details. Structure factor calculations are decomposed into the partial structure factors to interrogate the structural properties of the two block types separately. Parametric surveys are carried out of the effects of various parameters, including temperature, soft block length, and block connectivity on the observed structure. The routine detailed here for constructing models is robust enough to be executed automatically in a high throughput workflow for material design and discovery. △ Less

Submitted 24 May, 2024; originally announced May 2024.

Comments: 6 figures, 18 pages

Particle Denoising Diffusion Sampler

Authors: Angus Phillips, Hai-Dang Dau, Michael John Hutchinson, Valentin De Bortoli, George Deligiannidis, Arnaud Doucet

Abstract: Denoising diffusion models have become ubiquitous for generative modeling. The core idea is to transport the data distribution to a Gaussian by using a diffusion. Approximate samples from the data distribution are then obtained by estimating the time-reversal of this diffusion using score matching ideas. We follow here a similar strategy to sample from unnormalized probability densities and comput… ▽ More Denoising diffusion models have become ubiquitous for generative modeling. The core idea is to transport the data distribution to a Gaussian by using a diffusion. Approximate samples from the data distribution are then obtained by estimating the time-reversal of this diffusion using score matching ideas. We follow here a similar strategy to sample from unnormalized probability densities and compute their normalizing constants. However, the time-reversed diffusion is here simulated by using an original iterative particle scheme relying on a novel score matching loss. Contrary to standard denoising diffusion models, the resulting Particle Denoising Diffusion Sampler (PDDS) provides asymptotically consistent estimates under mild assumptions. We demonstrate PDDS on multimodal and high dimensional sampling tasks. △ Less

Submitted 15 June, 2024; v1 submitted 9 February, 2024; originally announced February 2024.

Comments: To be published in ICML 2024. 37 pages, 20 figures, 3 tables, 5 algorithms

SleepNet: Attention-Enhanced Robust Sleep Prediction using Dynamic Social Networks

Authors: Maryam Khalid, Elizabeth B. Klerman, Andrew W. Mchill, Andrew J. K. Phillips, Akane Sano

Abstract: Sleep behavior significantly impacts health and acts as an indicator of physical and mental well-being. Monitoring and predicting sleep behavior with ubiquitous sensors may therefore assist in both sleep management and tracking of related health conditions. While sleep behavior depends on, and is reflected in the physiology of a person, it is also impacted by external factors such as digital media… ▽ More Sleep behavior significantly impacts health and acts as an indicator of physical and mental well-being. Monitoring and predicting sleep behavior with ubiquitous sensors may therefore assist in both sleep management and tracking of related health conditions. While sleep behavior depends on, and is reflected in the physiology of a person, it is also impacted by external factors such as digital media usage, social network contagion, and the surrounding weather. In this work, we propose SleepNet, a system that exploits social contagion in sleep behavior through graph networks and integrates it with physiological and phone data extracted from ubiquitous mobile and wearable devices for predicting next-day sleep labels about sleep duration. Our architecture overcomes the limitations of large-scale graphs containing connections irrelevant to sleep behavior by devising an attention mechanism. The extensive experimental evaluation highlights the improvement provided by incorporating social networks in the model. Additionally, we conduct robustness analysis to demonstrate the system's performance in real-life conditions. The outcomes affirm the stability of SleepNet against perturbations in input data. Further analyses emphasize the significance of network topology in prediction performance revealing that users with higher eigenvalue centrality are more vulnerable to data perturbations. △ Less

Submitted 26 January, 2024; v1 submitted 19 January, 2024; originally announced January 2024.

Comments: Accepted for publication in Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies (IMWUT), 8 (March 2024)

Which algorithm to select in sports timetabling?

Authors: David Van Bulck, Dries Goossens, Jan-Patrick Clarner, Angelos Dimitsas, George H. G. Fonseca, Carlos Lamas-Fernandez, Martin Mariusz Lester, Jaap Pedersen, Antony E. Phillips, Roberto Maria Rosati

Abstract: Any sports competition needs a timetable, specifying when and where teams meet each other. The recent International Timetabling Competition (ITC2021) on sports timetabling showed that, although it is possible to develop general algorithms, the performance of each algorithm varies considerably over the problem instances. This paper provides an instance space analysis for sports timetabling, resulti… ▽ More Any sports competition needs a timetable, specifying when and where teams meet each other. The recent International Timetabling Competition (ITC2021) on sports timetabling showed that, although it is possible to develop general algorithms, the performance of each algorithm varies considerably over the problem instances. This paper provides an instance space analysis for sports timetabling, resulting in powerful insights into the strengths and weaknesses of eight state-of-the-art algorithms. Based on machine learning techniques, we propose an algorithm selection system that predicts which algorithm is likely to perform best when given the characteristics of a sports timetabling problem instance. Furthermore, we identify which characteristics are important in making that prediction, providing insights in the performance of the algorithms, and suggestions to further improve them. Finally, we assess the empirical hardness of the instances. Our results are based on large computational experiments involving about 50 years of CPU time on more than 500 newly generated problem instances. △ Less

Submitted 5 July, 2024; v1 submitted 4 September, 2023; originally announced September 2023.

Comments: This is the peer-reviewed author-version of https://doi.org/10.1016/j.ejor.2024.06.005, published in the European Journal of Operational Research. Copyright 2024. This manuscript version is made available under the LCC-BY-NC-ND 4.0 license (https://creativecommons.org/licenses/by-nc-nd/4.0/)

Journal ref: European Journal of Operational Research, 2024, 318, 575-591

On (not) deriving the entropy of barocaloric phase transitions from crystallography and neutron spectroscopy

Authors: Anthony E. Phillips, Helen C. Walker

Abstract: We review well-known signatures of disorder in crystallographic and inelastic neutron scattering data. We show that these can arise from different types of disorder, corresponding to different values of the system entropy. Correlating the entropy of a material with its atomistic structure and dynamics is in general a difficult problem that requires correlating information between multiple experime… ▽ More We review well-known signatures of disorder in crystallographic and inelastic neutron scattering data. We show that these can arise from different types of disorder, corresponding to different values of the system entropy. Correlating the entropy of a material with its atomistic structure and dynamics is in general a difficult problem that requires correlating information between multiple experimental techniques including crystallography, spectroscopy, and calorimetry. These comments are illustrated with particular reference to barocalorics, but are relevant to a broad range of calorics and other disordered crystalline materials. △ Less

Submitted 14 August, 2023; originally announced August 2023.

Comments: 11 pages, 3 figures

Bayesian Decision Trees Inspired from Evolutionary Algorithms

Authors: Efthyvoulos Drousiotis, Alexander M. Phillips, Paul G. Spirakis, Simon Maskell

Abstract: Bayesian Decision Trees (DTs) are generally considered a more advanced and accurate model than a regular Decision Tree (DT) because they can handle complex and uncertain data. Existing work on Bayesian DTs uses Markov Chain Monte Carlo (MCMC) with an accept-reject mechanism and sample using naive proposals to proceed to the next iteration, which can be slow because of the burn-in time needed. We c… ▽ More Bayesian Decision Trees (DTs) are generally considered a more advanced and accurate model than a regular Decision Tree (DT) because they can handle complex and uncertain data. Existing work on Bayesian DTs uses Markov Chain Monte Carlo (MCMC) with an accept-reject mechanism and sample using naive proposals to proceed to the next iteration, which can be slow because of the burn-in time needed. We can reduce the burn-in period by proposing a more sophisticated way of sampling or by designing a different numerical Bayesian approach. In this paper, we propose a replacement of the MCMC with an inherently parallel algorithm, the Sequential Monte Carlo (SMC), and a more effective sampling strategy inspired by the Evolutionary Algorithms (EA). Experiments show that SMC combined with the EA can produce more accurate results compared to MCMC in 100 times fewer iterations. △ Less

Submitted 30 May, 2023; originally announced May 2023.

Comments: arXiv admin note: text overlap with arXiv:2301.09090

Seven Classes of Rotational Variables From a Study of 50,000 Spotted Stars with ASAS-SN, Gaia, and APOGEE

Authors: Anya Phillips, C. S. Kochanek, Tharindu Jayasinghe, Lyra Cao, Collin T. Christy, D. M. Rowan, Marc Pinsonneault

Abstract: We examine the properties of $\sim50,000$ rotational variables from the ASAS-SN survey using distances, stellar properties, and probes of binarity from $\textit{Gaia}$ DR3 and the SDSS APOGEE survey. They have high amplitudes and span a broader period range than previously studied $\textit{Kepler}$ rotators. We find they divide into three groups of main sequence stars (MS1, MS2s, MS2b) and four of… ▽ More We examine the properties of $\sim50,000$ rotational variables from the ASAS-SN survey using distances, stellar properties, and probes of binarity from $\textit{Gaia}$ DR3 and the SDSS APOGEE survey. They have high amplitudes and span a broader period range than previously studied $\textit{Kepler}$ rotators. We find they divide into three groups of main sequence stars (MS1, MS2s, MS2b) and four of giants (G1/3, G2, G4s, and G4b). MS1 stars are slowly rotating (10-30 days), likely single stars with a limited range of temperatures. MS2s stars are more rapidly rotating (days) single stars spanning the lower main sequence up to the Kraft break. There is a clear period gap (or minimum) between MS1 and MS2s, similar to that seen for lower temperatures in the $\textit{Kepler}$ samples. MS2b stars are tidally locked binaries with periods of days. G1/3 stars are heavily spotted, tidally locked RS CVn with periods of tens of days. G2 stars are less luminous, heavily spotted, tidally locked sub-subgiants with periods of $\sim10$ days. G4s stars have intermediate luminosities to G1/3 and G2, slow rotation periods (approaching 100 days) and are almost certainly all merger remnants. G4b stars have similar rotation periods and luminosities to G4s, but consist of sub-synchronously rotating binaries. We see no difference in indicators for the presence of very wide binary companions between any of these groups and control samples of photometric twin stars built for each group. △ Less

Submitted 16 May, 2023; originally announced May 2023.

Comments: 15 pages, 11 figures, 5 tables, submitted to MNRAS

Atomic Layer Deposited Protective Coating of Aluminum Oxide on Silver-based Telescope Mirror A Comparison Between a Pure Ozone and H2O Precursor

Authors: Søren A. Tornøe, Brandon Cheney, Brian Dupraw, Yoshimasa Okamura, Andrew C. Phillips, Takayuki Hagiwara, Tetsuya Nishiguchi, Nobuhiko P. Kobayashi

Abstract: Although silver-based telescope mirrors excel over other materials such as gold and aluminum in the visible-infrared spectral range, they require robust protective coatings to overcome their inherent low durability. Our research shows that a single-layer of aluminum oxide (AlOx) deposited through thermal atomic layer deposition (ALD) using trimethylaluminum (TMA) and water (H2O) at low temperature… ▽ More Although silver-based telescope mirrors excel over other materials such as gold and aluminum in the visible-infrared spectral range, they require robust protective coatings to overcome their inherent low durability. Our research shows that a single-layer of aluminum oxide (AlOx) deposited through thermal atomic layer deposition (ALD) using trimethylaluminum (TMA) and water (H2O) at low temperatures (~60°C) serves as an acceptable protective coating without adversely impacting the optical performance of the mirrors. While silver-based mirrors protected with a single-layer of AlOx perform decently in the field, in environmental tests under high-humidity at high-temperature conditions that accelerate underlying failure mechanisms, they degrade quickly, suggesting that there is room for improvement. This paper describes a study that compares the performance and endurance of two sets of silver-based mirrors protected by a single-layer of AlOx prepared by thermal ALD with two types of oxygen precursors: H2O and pure ozone (PO). The study shows that while the two types of samples, regardless of their oxygen precursors, initially have comparable spectral reflectance, the reflectance of the samples with AlOx protective coatings prepared with PO remain nearly constant 1.6 times longer than those with AlOx protective coatings prepared with H2O in the environmental test, suggesting promising characteristics of AlOx protective coatings prepared with PO. △ Less

Submitted 8 May, 2023; originally announced May 2023.

Comments: 18 pages, 5 figures, 1 table

Instructing nontraditional physics labs: Toward responsiveness to student epistemic framing

Authors: Meagan Sundstrom, Rebeckah K. Fussell, Anna McLean Phillips, Mark Akubo, Scott E. Allen, David Hammer, Rachel E. Scherr, N. G. Holmes

Abstract: Research on nontraditional laboratory (lab) activities in physics shows that students often expect to verify predetermined results, as takes place in traditional activities. This understanding of what is taking place, or epistemic framing, may impact their behaviors in the lab, either productively or unproductively. In this paper, we present an analysis of student epistemic framing in a nontraditi… ▽ More Research on nontraditional laboratory (lab) activities in physics shows that students often expect to verify predetermined results, as takes place in traditional activities. This understanding of what is taking place, or epistemic framing, may impact their behaviors in the lab, either productively or unproductively. In this paper, we present an analysis of student epistemic framing in a nontraditional lab to understand how instructional context, specifically instructor behaviors, may shape student framing. We present video data from a lab section taught by an experienced teaching assistant (TA), with 19 students working in seven groups. We argue that student framing in this lab is evidenced by whether or not students articulate experimental predictions and by the extent to which they take up opportunities to construct knowledge (epistemic agency). We show that the TA's attempts to shift student frames generally succeed with respect to experimental predictions but are less successful with respect to epistemic agency. In part, we suggest, the success of the TA's attempts reflects whether and how they are responsive to students' current framing. This work offers evidence that instructors can shift students' frames in nontraditional labs, while also illuminating the complexities of both student framing and the role of the instructor in shifting that framing in this context. △ Less

Submitted 28 April, 2023; originally announced April 2023.

Comments: Submitted to Physical Review Physics Education Research

A Study of the Properties and Dynamics of the Disk of Satellites in a Milky-Way-like Galaxy System

Authors: Xinghai Zhao, Grant J. Mathews, Lara Arielle Phillips, Guobao Tang

Abstract: The dynamics of the satellite systems of Milky-Way-like galaxies offer a useful means by which to study the galaxy formation process in the cosmological context. It has been suggested that the currently observed anisotropic distribution of the satellites in such galaxy systems is inconsistent with the concordance $ΛCDM$ cosmology model on the galactic scale if the observed satellites are random sa… ▽ More The dynamics of the satellite systems of Milky-Way-like galaxies offer a useful means by which to study the galaxy formation process in the cosmological context. It has been suggested that the currently observed anisotropic distribution of the satellites in such galaxy systems is inconsistent with the concordance $ΛCDM$ cosmology model on the galactic scale if the observed satellites are random samples of the dark matter (DM) sub-halos that are nearly isotropically distributed around the central galaxy. In this study, we present original high-resolution zoom-in studies of central galaxies and satellite systems based upon initial conditions for the DM distribution from the Aquarius simulations but with substantial high-resolution baryon physics added. We find that the galaxy most like the Milky Way in this study does indeed contain a disk of satellites (DOS). Although one galaxy DOS system does not answer the question of how common such disks are, it does allow the opportunity to explore the properties and dynamics of the DOS system. Our investigation centers on the spatial arrangement (distances, angles, etc.) of satellites in this Milky-Way-like galaxy system with a specific emphasis on identifying and analyzing the disk-like structure along with its dynamical and morphological properties. Among the conclusions from this study, we find that the satellites and DM sub-halos in the galaxy simulations are anisotropically distributed. The dynamical properties of the satellites, however, indicate that the direction of the angular momentum vector of the whole satellite system is different from the normal direction of the fitted DOS and from the normal direction of the velocity dispersion of the system. Hence, the fitted DOS appears to be comprised of infalling sub-halos and is not a rotationally supported system. △ Less

Submitted 6 December, 2023; v1 submitted 11 February, 2023; originally announced February 2023.

Comments: 22 pages, 14 figures

Journal ref: Galaxies 2023, 11(6), 114

Deep Drilling in the Time Domain with DECam: Survey Characterization

Authors: Melissa L. Graham, Robert A. Knop, Thomas Kennedy, Peter E. Nugent, Eric Bellm, Márcio Catelan, Avi Patel, Hayden Smotherman, Monika Soraisam, Steven Stetzler, Lauren N. Aldoroty, Autumn Awbrey, Karina Baeza-Villagra, Pedro H. Bernardinelli, Federica Bianco, Dillon Brout, Riley Clarke, William I. Clarkson, Thomas Collett, James R. A. Davenport, Shenming Fu, John E. Gizis, Ari Heinze, Lei Hu, Saurabh W. Jha , et al. (19 additional authors not shown)

Abstract: This paper presents a new optical imaging survey of four deep drilling fields (DDFs), two Galactic and two extragalactic, with the Dark Energy Camera (DECam) on the 4 meter Blanco telescope at the Cerro Tololo Inter-American Observatory (CTIO). During the first year of observations in 2021, $>$4000 images covering 21 square degrees (7 DECam pointings), with $\sim$40 epochs (nights) per field and 5… ▽ More This paper presents a new optical imaging survey of four deep drilling fields (DDFs), two Galactic and two extragalactic, with the Dark Energy Camera (DECam) on the 4 meter Blanco telescope at the Cerro Tololo Inter-American Observatory (CTIO). During the first year of observations in 2021, $>$4000 images covering 21 square degrees (7 DECam pointings), with $\sim$40 epochs (nights) per field and 5 to 6 images per night per filter in $g$, $r$, $i$, and/or $z$, have become publicly available (the proprietary period for this program is waived). We describe the real-time difference-image pipeline and how alerts are distributed to brokers via the same distribution system as the Zwicky Transient Facility (ZTF). In this paper, we focus on the two extragalactic deep fields (COSMOS and ELAIS-S1), characterizing the detected sources and demonstrating that the survey design is effective for probing the discovery space of faint and fast variable and transient sources. We describe and make publicly available 4413 calibrated light curves based on difference-image detection photometry of transients and variables in the extragalactic fields. We also present preliminary scientific analysis regarding Solar System small bodies, stellar flares and variables, Galactic anomaly detection, fast-rising transients and variables, supernovae, and active galactic nuclei. △ Less

Submitted 16 November, 2022; originally announced November 2022.

Comments: 22 pages, 17 figures, 2 tables. Accepted to MNRAS

Unlocking the potential of two-point cells for energy-efficient and resilient training of deep nets

Authors: Ahsan Adeel, Adewale Adetomi, Khubaib Ahmed, Amir Hussain, Tughrul Arslan, W. A. Phillips

Abstract: Context-sensitive two-point layer 5 pyramidal cells (L5PCs) were discovered as long ago as 1999. However, the potential of this discovery to provide useful neural computation has yet to be demonstrated. Here we show for the first time how a transformative L5PCs-driven deep neural network (DNN), termed the multisensory cooperative computing (MCC) architecture, can effectively process large amounts… ▽ More Context-sensitive two-point layer 5 pyramidal cells (L5PCs) were discovered as long ago as 1999. However, the potential of this discovery to provide useful neural computation has yet to be demonstrated. Here we show for the first time how a transformative L5PCs-driven deep neural network (DNN), termed the multisensory cooperative computing (MCC) architecture, can effectively process large amounts of heterogeneous real-world audio-visual (AV) data, using far less energy compared to best available 'point' neuron-driven DNNs. A novel highly-distributed parallel implementation on a Xilinx UltraScale+ MPSoC device estimates energy savings up to 245759 $ \times $ 50000 $μ$J (i.e., 62% less than the baseline model in a semi-supervised learning setup) where a single synapse consumes $8e^{-5}μ$J. In a supervised learning setup, the energy-saving can potentially reach up to 1250x less (per feedforward transmission) than the baseline model. The significantly reduced neural activity in MCC leads to inherently fast learning and resilience against sudden neural damage. This remarkable performance in pilot experiments demonstrates the embodied neuromorphic intelligence of our proposed cooperative L5PC that receives input from diverse neighbouring neurons as context to amplify the transmission of most salient and relevant information for onward transmission, from overwhelmingly large multimodal information utilised at the early stages of on-chip training. Our proposed approach opens new cross-disciplinary avenues for future on-chip DNN training implementations and posits a radical shift in current neuromorphic computing paradigms. △ Less

Submitted 22 December, 2022; v1 submitted 24 October, 2022; originally announced November 2022.

Dynamics in the ordered and disordered phases of barocaloric adamantane

Authors: Bernet E. Meijer, Richard J. C. Dixey, Franz Demmel, Robin Perry, Helen C. Walker, Anthony E. Phillips

Abstract: High-entropy order-disorder phase transitions can be used for efficient and eco-friendly barocaloric solid-state cooling. Here the barocaloric effect is reported in an archetypal plastic crystal, adamantane. Adamantane has a colossal isothermally reversible entropy change of 106 J K-1 kg-1 . Extremely low hysteresis means that this can be accessed at pressure differences less than 200 bar. Configu… ▽ More High-entropy order-disorder phase transitions can be used for efficient and eco-friendly barocaloric solid-state cooling. Here the barocaloric effect is reported in an archetypal plastic crystal, adamantane. Adamantane has a colossal isothermally reversible entropy change of 106 J K-1 kg-1 . Extremely low hysteresis means that this can be accessed at pressure differences less than 200 bar. Configurational entropy can only account for about 40% of the total entropy change; the remainder is due to vibrational effects. Using neutron spectroscopy and supercell lattice dynamics calculations, it is found that this vibrational entropy change is mainly caused by softening in the high-entropy phase of acoustic modes that correspond to molecular rotations. We attribute this behaviour to the contrast between an 'interlocked' state in the low-entropy phase and sphere-like behaviour in the high-entropy phase. Although adamantane is a simple van der Waals solid with near-spherical molecules, this approach can be leveraged for the design of more complex barocaloric molecular crystals. Moreover, this study shows that supercell lattice dynamics calculations can accurately map the effect of orientational disorder on the phonon spectrum, paving the way for studying the vibrational entropy, thermal conductivity, and other thermodynamic effects in more complex materials. △ Less

Submitted 25 October, 2022; originally announced October 2022.

Comments: 14 pages, 6 figures

Spectral Diffusion Processes

Authors: Angus Phillips, Thomas Seror, Michael Hutchinson, Valentin De Bortoli, Arnaud Doucet, Emile Mathieu

Abstract: Score-based generative modelling (SGM) has proven to be a very effective method for modelling densities on finite-dimensional spaces. In this work we propose to extend this methodology to learn generative models over functional spaces. To do so, we represent functional data in spectral space to dissociate the stochastic part of the processes from their space-time part. Using dimensionality reducti… ▽ More Score-based generative modelling (SGM) has proven to be a very effective method for modelling densities on finite-dimensional spaces. In this work we propose to extend this methodology to learn generative models over functional spaces. To do so, we represent functional data in spectral space to dissociate the stochastic part of the processes from their space-time part. Using dimensionality reduction techniques we then sample from their stochastic component using finite dimensional SGM. We demonstrate our method's effectiveness for modelling various multimodal datasets. △ Less

Submitted 28 November, 2022; v1 submitted 28 September, 2022; originally announced September 2022.

Comments: 17 pages, 11 figures, Score-based Method Workshop at 36th Conference on Neural Information Processing Systems (NeurIPS 2022)

A comparison of partial information decompositions using data from real and simulated layer 5b pyramidal cells

Authors: Jim W. Kay, Jan M. Schulz, W. A. Phillips

Abstract: Partial information decomposition allows the joint mutual information between an output and a set of inputs to be divided into components that are synergistic or shared or unique to each input. We consider five different decompositions and compare their results on data from layer 5b pyramidal cells in two different studies. The first study was of the amplification of somatic action potential outpu… ▽ More Partial information decomposition allows the joint mutual information between an output and a set of inputs to be divided into components that are synergistic or shared or unique to each input. We consider five different decompositions and compare their results on data from layer 5b pyramidal cells in two different studies. The first study was of the amplification of somatic action potential output by apical dendritic input and its regulation by dendritic inhibition. We find that two of the decompositions produce much larger estimates of synergy and shared information than the others, as well as large levels of unique misinformation. When within-neuron differences in the components are examined, the five methods produce more similar results for all but the shared information component, for which two methods produce a different statistical conclusion from the others. There are some differences in the expression of unique information asymmetry among the methods. It is significantly larger, on average, under dendritic inhibition. Three of the methods support a previous conclusion that apical amplification is reduced by dendritic inhibition. The second study used a detailed compartmental model to produce action potentials for many combinations of the numbers of basal and apical synaptic inputs. Two analyses of decompositions are conducted on subsets of the data. In the first, the decompositions reveal a bifurcation in unique information asymmetry. For three of the methods this suggests that apical drive switches to basal drive as the strength of the basal input increases, while the other two show changing mixtures of information and misinformation. Decompositions produced using the second set of subsets show that all five decompositions provide support for properties of cooperative context-sensitivity - to varying extents. △ Less

Submitted 13 June, 2022; originally announced June 2022.

Comments: 27 pages, 11 figures

Journal ref: Published in Entropy, 24th July, 2022, 24(8), 1021

Sex and Gender in the Computer Graphics Research Literature

Authors: Ana Dodik, Silvia Sellán, Theodore Kim, Amanda Phillips

Abstract: We survey the treatment of sex and gender in the Computer Graphics research literature from an algorithmic fairness perspective. The established practices on the use of gender and sex in our community are scientifically incorrect and constitute a form of algorithmic bias with potential harmful effects. We propose ways of addressing these as technical limitations. We survey the treatment of sex and gender in the Computer Graphics research literature from an algorithmic fairness perspective. The established practices on the use of gender and sex in our community are scientifically incorrect and constitute a form of algorithmic bias with potential harmful effects. We propose ways of addressing these as technical limitations. △ Less

Submitted 1 June, 2022; originally announced June 2022.

A novel machine learning enabled hybrid optimization framework for efficient and transferable coarse-graining of a model polymer

Authors: Zakiya Shireen, Hansani Weeratunge, Adrian Menzel, Andrew W Phillips, Ronald G Larson, Kate Smith-Miles, Elnaz Hajizadeh

Abstract: This work presents a novel framework governing the development of an efficient, accurate, and transferable coarse-grained (CG) model of a polyether material. The proposed framework combines the two fundamentally different classical optimization approaches for the development of coarse-grained model parameters; namely bottom-up and top-down approaches. This is achieved through integrating the optim… ▽ More This work presents a novel framework governing the development of an efficient, accurate, and transferable coarse-grained (CG) model of a polyether material. The proposed framework combines the two fundamentally different classical optimization approaches for the development of coarse-grained model parameters; namely bottom-up and top-down approaches. This is achieved through integrating the optimization algorithms into a machine learning (ML) model, trained using molecular dynamics (MD) simulation data. In the bottom-up approach, bonded interactions of the CG model are optimized using deep neural networks (DNN), where atomistic bonded distributions are matched. The atomistic distributions emulate the local chain structure. In the top-down approach, optimization of nonbonded potentials is accomplished by reproducing the temperature-dependent experimental density. We demonstrate that CG model parameters achieved through our machine-learning enabled hybrid optimization framework fulfills the thermodynamic consistency and transferability issues associated with the classical approaches to coarse-graining model polymers. We demonstrate the efficiency, accuracy, and transferability of the developed CG model, using our novel framework through accurate predictions of chain size as well as chain dynamics, including the limiting behavior of the glass transition temperature, diffusion, and stress relaxation spectrum, where none were included in the potential parameterization process. The accuracy of the predicted properties are evaluated in the context of molecular theories and available experimental data. △ Less

Submitted 28 April, 2022; originally announced April 2022.

Position fixing with cold atom gravity gradiometers

Authors: Alexander M. Phillips, Michael J. Wright, Isabelle Riou, Stephen Maddox, Simon Maskell, Jason F. Ralph

Abstract: This paper proposes a position fixing method for autonomous navigation using partial gravity gradient solutions from cold atom interferometers. Cold atom quantum sensors can provide ultra-precise measurements of inertial quantities, such as acceleration and rotation rates. However, we investigate the use of pairs of cold atom interferometers to measure the local gravity gradient and to provide pos… ▽ More This paper proposes a position fixing method for autonomous navigation using partial gravity gradient solutions from cold atom interferometers. Cold atom quantum sensors can provide ultra-precise measurements of inertial quantities, such as acceleration and rotation rates. However, we investigate the use of pairs of cold atom interferometers to measure the local gravity gradient and to provide position information by referencing these measurements against a suitable database. Simulating the motion of a vehicle, we use partial gravity gradient measurements to reduce the positional drift associated with inertial navigation systems. Using standard open source global gravity databases, we show stable navigation solutions for trajectories of over 1000km. △ Less

Submitted 11 April, 2022; originally announced April 2022.

Comments: 10 pages, 6 figures

Journal ref: AVS Quantum Science, Volume 4, Issue 2, 10.1116/5.0095677, 2022

A machine learning accelerated inverse design of underwater acoustic polyurethane coatings with cylindrical voids

Authors: Hansani Weeratunge, Zakiya Shireen, Sagar Iyer, Richard Sandberg, Saman Halgamuge, Adrian Menzel, Andrew Phillips, Elnaz Hajizadeh

Abstract: Here, we report the development of a detailed "Materials Informatics" framework for the design of acoustic coatings for underwater sound attenuation through integrating Machine Learning (ML) and statistical optimization algorithms with a Finite Element Model (FEM). The finite element models were developed to simulate the realistic performance of the acoustic coatings based on polyurethane (PU) ela… ▽ More Here, we report the development of a detailed "Materials Informatics" framework for the design of acoustic coatings for underwater sound attenuation through integrating Machine Learning (ML) and statistical optimization algorithms with a Finite Element Model (FEM). The finite element models were developed to simulate the realistic performance of the acoustic coatings based on polyurethane (PU) elastomers with embedded cylindrical voids. The FEM results revealed that the frequency-dependent viscoelastic behavior of the polyurethane matrix has a significant impact on the magnitude and frequency of the absorption peak associated with the cylinders at low frequencies, which has been commonly ignored in previous studies on similar systems. The data generated from the FEM was used to train a Deep Neural Network (DNN) to accelerate the design process, and subsequently, was integrated with a Genetic Algorithm (GA) to determine the optimal geometric parameters of the cylinders to achieve maximized, broadband, low-frequency waterborne sound attenuation. A significant, broadband, low-frequency attenuation is achieved by optimally configuring the layers of cylindrical voids and using attenuation mechanisms, including Fabry-Pérot resonance and Bragg scattering of the layers of voids. Integration of the machine learning technique into the optimization algorithm further accelerated the exploration of the high dimensional design space for the targeted performance. The developed DNN exhibited significantly increased speed (by a factor of $4.5\times 10^3$ ) in predicting the absorption coefficient compared to the conventional FEM(s). Therefore, the acceleration brought by the materials informatics framework brings a paradigm shift to the design and development of acoustic coatings compared to the conventional trial-and-error practices. △ Less

Submitted 1 March, 2022; originally announced March 2022.

Physicality, Modeling and Making in a Computational Physics Class

Authors: Anna M. Phillips, Ezra J. Gouvea, Brian E. Gravel, Pierre-Hugues Beauchemin, Timothy J. Atherton

Abstract: Computation is intertwined with essentially all aspects of physics research and is invaluable for physicists' careers. Despite its disciplinary importance, integration of computation into physics education remains a challenge and, moreover, has tended to be constructed narrowly as a route to solving physics problems. Here, we broaden Physics Education Research's conception of computation by constr… ▽ More Computation is intertwined with essentially all aspects of physics research and is invaluable for physicists' careers. Despite its disciplinary importance, integration of computation into physics education remains a challenge and, moreover, has tended to be constructed narrowly as a route to solving physics problems. Here, we broaden Physics Education Research's conception of computation by constructing an epistemic \emph{metamodel} -- a model of modeling -- incorporating insights on computational modeling from the philosophy of science and prior work. The metamodel is formulated in terms of practices, things physicists do, and how these inform one another. We operationalize this metamodel in an educational environment that incorporates making, the creation of shared physical and digital artifacts, intended to promote students' agency, creativity and self-expression alongside doing physics. We present a content analysis of student work from initial implementations of this approach to illustrate the very complex epistemic maneuvers students make as they engaged in computational modeling. We demonstrate how our metamodel can be used to understand student practices, and conclude with implications of the metamodel for instruction and future research. △ Less

Submitted 7 December, 2022; v1 submitted 8 March, 2022; originally announced March 2022.

A global bifurcation organizing rhythmic activity in a coupled network

Authors: Georgi S. Medvedev, Matthew S. Mizuhara, Andrew Phillips

Abstract: We study a system of coupled phase oscillators near a saddle-node on an invariant circle bifurcation and driven by random intrinsic frequencies. Under the variation of control parameters, the system undergoes a phase transition changing the qualitative properties of collective dynamics. Using the Ott-Antonsen reduction and geometric techniques for ordinary differential equations, we identify a het… ▽ More We study a system of coupled phase oscillators near a saddle-node on an invariant circle bifurcation and driven by random intrinsic frequencies. Under the variation of control parameters, the system undergoes a phase transition changing the qualitative properties of collective dynamics. Using the Ott-Antonsen reduction and geometric techniques for ordinary differential equations, we identify a heteroclinic bifurcation in a family of vector fields on a cylinder, which explains the change in collective dynamics. Specifically, we show that the heteroclinic bifurcation separates two topologically distinct families of limit cycles: contractible limit cycles before the bifurcation from noncontractibile ones after the bifurcation. Both families are stable for the model at hand. △ Less

Submitted 17 August, 2022; v1 submitted 2 March, 2022; originally announced March 2022.

Comments: 22 pages, 12 figures

Journal ref: Chaos 32, 083116 (2022)

Pressure dependence of atomic dynamics in barocaloric ammonium sulfate: II. Vibrations

Authors: Shurong Yuan, Bernet E. Meijer, Guanqun Cai, Richard J. C. Dixey, Anthony E. Phillips, Helen C. Walker

Abstract: Ammonium sulfate is a giant inverse barocaloric material that is cheaply and commercially available. Exploiting its potential for cooling applications requires an understanding of the mechanism driving the entropy change. Here we report an investigation by inelastic neutron scattering and density functional theory of the phonons under working conditions of temperature and pressure. We find excelle… ▽ More Ammonium sulfate is a giant inverse barocaloric material that is cheaply and commercially available. Exploiting its potential for cooling applications requires an understanding of the mechanism driving the entropy change. Here we report an investigation by inelastic neutron scattering and density functional theory of the phonons under working conditions of temperature and pressure. We find excellent agreement between the experimental and calculated results. The ammonium librational modes that are crucial to the entropy change are identifiable by their negative Grüneisen parameter. Our results connect the differences in structure across the phase transition to those in the atomic dynamics, suggesting a route towards designing new caloric materials. △ Less

Submitted 5 August, 2022; v1 submitted 17 February, 2022; originally announced February 2022.

Comments: This manuscript has now been incorporated into 2201.02548. Please check there for the latest version

Pressure dependence of atomic dynamics in barocaloric ammonium sulfate: I. Rotations

Authors: Bernet E. Meijer, Guanqun Cai, Franz Demmel, Helen C. Walker, Anthony E. Phillips

Abstract: Solid-state cooling using barocaloric materials is a promising avenue for eco-friendly, inexpensive and highly efficient cooling. To design barocaloric compounds ready for deployment, it is essential to understand their thermodynamic behaviour under working conditions. To this end, we have studied the rotational dynamics in the molecular-ionic crystal ammonium sulfate under pressure, providing det… ▽ More Solid-state cooling using barocaloric materials is a promising avenue for eco-friendly, inexpensive and highly efficient cooling. To design barocaloric compounds ready for deployment, it is essential to understand their thermodynamic behaviour under working conditions. To this end, we have studied the rotational dynamics in the molecular-ionic crystal ammonium sulfate under pressure, providing detailed insight into the origin of its large barocaloric effect. Using quasielastic neutron scattering experiments, we show that rotation of the ammonium cations is facilitated by pressure in the low-entropy phase, with the rotational "hopping" motion increasing in frequency as the pressure-induced phase transition is approached. We explain this unusual behaviour in terms of the competing hydrogen-bond networks represented by the two phases. This work includes the first results of a recently developed low-background, high-pressure gas cell for neutron scattering, showcasing its power in obtaining high-precision measurements of molecular dynamics under pressure. △ Less

Submitted 27 January, 2022; originally announced January 2022.

Comments: 8 pages, 5 figures

Origin of the large entropy change in the molecular caloric and ferroelectric ammonium sulfate

Authors: Bernet E. Meijer, Shurong Yuan, Guanqun Cai, Richard J. Dixey, Franz Demmel, Martin T. Dove, Jiaxun Liu, Helen Y. Playford, Helen C. Walker, Anthony E. Phillips

Abstract: The deceptively simple inorganic salt ammonium sulfate undergoes a ferroelectric phase transition associated with a very large entropy change and both electrocaloric and barocaloric functionality. While the structural origins of the electrical polarisation are now well established, those of the entropy change have been controversial for over fifty years. This question is resolved here using a comb… ▽ More The deceptively simple inorganic salt ammonium sulfate undergoes a ferroelectric phase transition associated with a very large entropy change and both electrocaloric and barocaloric functionality. While the structural origins of the electrical polarisation are now well established, those of the entropy change have been controversial for over fifty years. This question is resolved here using a combination of DFT phonon calculations with inelastic neutron scattering under variable temperature and pressure, supported by complementary total and quasielastic neutron scattering experiments. A simple model of the entropy in which each molecular ion is disordered across the mirror plane in the high symmetry phase, although widely used in the literature, proves to be untenable. Instead, the entropy arises from low-frequency librations of ammonium ions in this phase, with harmonic terms that are very small or even negative. These results suggest that, in the search for molecular materials with functionality derived from large entropy changes, vibrational entropy arising from broad energy minima is likely to be just as important as configurational entropy arising from crystallographic disorder. △ Less

Submitted 5 August, 2022; v1 submitted 7 January, 2022; originally announced January 2022.

Comments: 18 pages, 6 figures

Supernova Siblings and their Parent Galaxies in the Zwicky Transient Facility Bright Transient Surve

Authors: M. L. Graham, C. Fremling, D. A. Perley, R. Biswas, C. A. Phillips, J. Sollerman, P. E. Nugent, S. Nance, S. Dhawan, J. Nordin, A. Goobar, A. Miller, J. D. Neill, X. J. Hall, M. J. Hankins, D. A. Duev, M. M. Kasliwal, M. Rigault, E. C. Bellm, D. Hale, P. Mróz, S. R. Kulkarni

Abstract: Supernova (SN) siblings -- two or more SNe in the same parent galaxy -- are useful tools for exploring progenitor stellar populations as well as properties of the host galaxies such as distance, star formation rate, dust extinction, and metallicity. Since the average SN rate for a Milky Way-type galaxy is just one per century, a large imaging survey is required to discover an appreciable sample of… ▽ More Supernova (SN) siblings -- two or more SNe in the same parent galaxy -- are useful tools for exploring progenitor stellar populations as well as properties of the host galaxies such as distance, star formation rate, dust extinction, and metallicity. Since the average SN rate for a Milky Way-type galaxy is just one per century, a large imaging survey is required to discover an appreciable sample of SN siblings. From the wide-field Zwicky Transient Facility (ZTF) Bright Transient Survey (BTS; which aims for spectroscopic completeness for all transients which peak brighter than $r{<}$18.5 mag) we present 10 SN siblings in 5 parent galaxies. For each of these families we analyze the SN's location within the host and its underlying stellar population, finding agreement with expectations that SNe from more massive progenitors are found nearer to their host core and in regions of more active star formation. We also present an analysis of the relative rates of core collapse and thermonuclear SN siblings, finding a significantly lower ratio than past SN sibling samples due to the unbiased nature of the ZTF. △ Less

Submitted 29 December, 2021; originally announced December 2021.

Comments: 14 pages, 4 tables, 7 figures, accepted to MNRAS

Connected Components for Infinite Graph Streams: Theory and Practice

Authors: Jonathan W. Berry, Cynthia A Phillips, Alexandra M. Porter

Abstract: Motivated by the properties of unending real-world cybersecurity streams, we present a new graph streaming model: XStream. We maintain a streaming graph and its connected components at single-edge granularity. In cybersecurity graph applications, input streams typically consist of edge insertions; individual deletions are not explicit. Analysts maintain as much history as possible and will trigger… ▽ More Motivated by the properties of unending real-world cybersecurity streams, we present a new graph streaming model: XStream. We maintain a streaming graph and its connected components at single-edge granularity. In cybersecurity graph applications, input streams typically consist of edge insertions; individual deletions are not explicit. Analysts maintain as much history as possible and will trigger customized bulk deletions when necessary Despite a variety of dynamic graph processing systems and some canonical literature on theoretical sliding-window graph streaming, XStream is the first model explicitly designed to accommodate this usage model. Users can provide Boolean predicates to define bulk deletions. Edge arrivals are expected to occur continuously and must always be handled. XStream is implemented via a ring of finite-memory processors. We give algorithms to maintain connected components on the input stream, answer queries about connectivity, and to perform bulk deletion. The system requires bandwidth for internal messages that is some constant factor greater than the stream arrival rate. We prove a relationship among four quantities: the proportion of query downtime allowed, the proportion of edges that survive an aging event, the proportion of duplicated edges, and the bandwidth expansion factor. In addition to presenting the theory behind XStream, we present computational results for a single-threaded prototype implementation. Stream ingestion rates are bounded by computer architecture. We determine this bound for XStream inter-process message-passing rates in Intel TBB applications on Intel Sky Lake processors: between one and five million graph edges per second. Our single-threaded prototype runs our full protocols through multiple aging events at between one half and one a million edges per second, and we give ideas for speeding this up by orders of magnitude. △ Less

Submitted 30 November, 2021; originally announced December 2021.

Comments: 23 pages, 12 figures

Report number: Sandia National Laboratories Technical Report SAND2021-10819 O ACM Class: B.4; E.1; F.2

Adaptive Transfer Learning: a simple but effective transfer learning

Authors: Jung H Lee, Henry J Kvinge, Scott Howland, Zachary New, John Buckheit, Lauren A. Phillips, Elliott Skomski, Jessica Hibler, Courtney D. Corley, Nathan O. Hodas

Abstract: Transfer learning (TL) leverages previously obtained knowledge to learn new tasks efficiently and has been used to train deep learning (DL) models with limited amount of data. When TL is applied to DL, pretrained (teacher) models are fine-tuned to build domain specific (student) models. This fine-tuning relies on the fact that DL model can be decomposed to classifiers and feature extractors, and a… ▽ More Transfer learning (TL) leverages previously obtained knowledge to learn new tasks efficiently and has been used to train deep learning (DL) models with limited amount of data. When TL is applied to DL, pretrained (teacher) models are fine-tuned to build domain specific (student) models. This fine-tuning relies on the fact that DL model can be decomposed to classifiers and feature extractors, and a line of studies showed that the same feature extractors can be used to train classifiers on multiple tasks. Furthermore, recent studies proposed multiple algorithms that can fine-tune teacher models' feature extractors to train student models more efficiently. We note that regardless of the fine-tuning of feature extractors, the classifiers of student models are trained with final outputs of feature extractors (i.e., the outputs of penultimate layers). However, a recent study suggested that feature maps in ResNets across layers could be functionally equivalent, raising the possibility that feature maps inside the feature extractors can also be used to train student models' classifiers. Inspired by this study, we tested if feature maps in the hidden layers of the teacher models can be used to improve the student models' accuracy (i.e., TL's efficiency). Specifically, we developed 'adaptive transfer learning (ATL)', which can choose an optimal set of feature maps for TL, and tested it in the few-shot learning setting. Our empirical evaluations suggest that ATL can help DL models learn more efficiently, especially when available examples are limited. △ Less

Submitted 21 November, 2021; originally announced November 2021.

Comments: 10 pages, 7 figures

Gaps, Ambiguity, and Establishing Complexity-Class Containments via Iterative Constant-Setting

Authors: Lane A. Hemaspaandra, Mandar Juvekar, Arian Nadjimzadah, Patrick A. Phillips

Abstract: Cai and Hemachandra used iterative constant-setting to prove that Few $\subseteq$ $\oplus$P (and thus that FewP $\subseteq$ $\oplus$P). In this paper, we note that there is a tension between the nondeterministic ambiguity of the class one is seeking to capture, and the density (or, to be more precise, the needed "nongappy"-ness) of the easy-to-find "targets" used in iterative constant-setting. In… ▽ More Cai and Hemachandra used iterative constant-setting to prove that Few $\subseteq$ $\oplus$P (and thus that FewP $\subseteq$ $\oplus$P). In this paper, we note that there is a tension between the nondeterministic ambiguity of the class one is seeking to capture, and the density (or, to be more precise, the needed "nongappy"-ness) of the easy-to-find "targets" used in iterative constant-setting. In particular, we show that even less restrictive gap-size upper bounds regarding the targets allow one to capture ambiguity-limited classes. Through a flexible, metatheorem-based approach, we do so for a wide range of classes including the logarithmic-ambiguity version of Valiant's unambiguous nondeterminism class UP. Our work lowers the bar for what advances regarding the existence of infinite, P-printable sets of primes would suffice to show that restricted counting classes based on the primes have the power to accept superconstant-ambiguity analogues of UP. As an application of our work, we prove that the Lenstra-Pomerance-Wagstaff Conjecture implies that all (O(1) + loglogn)-ambiguity NP sets are in the restricted counting class $\rm RC_{PRIMES}$. △ Less

Submitted 9 February, 2024; v1 submitted 29 September, 2021; originally announced September 2021.

ACM Class: F.1.3

The $ω^3$ scaling of the vibrational density of states in quasi-2D nanoconfined solids

Authors: Yuanxi Yu, Chenxing Yang, Matteo Baggioli, Anthony E. Phillips, Alessio Zaccone, Lei Zhang, Ryoichi Kajimoto, Mitsutaka Nakamura, Dehong Yu, Liang Hong

Abstract: Atomic vibrations play a vital role in the functions of various physical, chemical, and biological systems. The vibrational properties and the specific heat of crystalline bulk materials are well described by Debye theory, which successfully predicts the quadratic $ω^{2}$ low-frequency scaling of the vibrational density of states (VDOS) in bulk ordered solids from few fundamental assumptions. Howe… ▽ More Atomic vibrations play a vital role in the functions of various physical, chemical, and biological systems. The vibrational properties and the specific heat of crystalline bulk materials are well described by Debye theory, which successfully predicts the quadratic $ω^{2}$ low-frequency scaling of the vibrational density of states (VDOS) in bulk ordered solids from few fundamental assumptions. However, the analogous framework for nanoconfined materials with fewer degrees of freedom has been far less well explored. Using inelastic neutron scattering, we characterize the VDOS of amorphous ice confined to a thickness of $\approx 1$ nm inside graphene oxide membranes and we observe a crossover from the Debye $ω^2$ scaling to an anomalous $ω^3$ behaviour upon reducing the confinement size $L$. Additionally, using molecular dynamics simulations, we confirm the experimental findings and also prove that such a scaling of the VDOS appears in both crystalline and amorphous solids under slab-confinement. We theoretically demonstrate that this low-frequency $ω^3$ law results from the geometric constraints on the momentum phase space induced by confinement along one spatial direction. Finally, we predict that the Debye scaling reappears at a characteristic frequency $ω_\times= v L/2π$, with $v$ the speed of sound of the material, and we confirm this quantitative estimate with simulations. This new physical phenomenon, revealed by combining theoretical, experimental and simulations results, is relevant to a myriad of systems both in synthetic and biological contexts and it could impact various technological applications for systems under confinement such as nano-devices or thin films. △ Less

Submitted 29 April, 2022; v1 submitted 17 August, 2021; originally announced August 2021.

Comments: v2: revised manuscript, matching the published version

Journal ref: Nat Commun 13, 3649 (2022)

Atomic structure of the continuous random network of amorphous C[(C6H4)2]2, PAF-1

Authors: Guanqun Cai, He Lin, Ziqiang Zhao, Jiaxun Liu, Anthony E Phillips, Thomas F Headen, Tristan G A Youngs, Yang Hai, Haolai Tian, Chunyong He, Yubin Ke, Juzhou Tao, Teng Ben, Martin T Dove

Abstract: We demonstrate that the amorphous material PAF-1, C[(C6H4)2]2, forms a continuous random network in which tetrahedral carbon sites are connected by 4,4'-biphenyl linkers. Experimental neutron total scattering measurements on deuterated, hydrogenous, and null-scattering samples agree with molecular dynamics simulations based on this model. From the MD model, we are able for the first time to interr… ▽ More We demonstrate that the amorphous material PAF-1, C[(C6H4)2]2, forms a continuous random network in which tetrahedral carbon sites are connected by 4,4'-biphenyl linkers. Experimental neutron total scattering measurements on deuterated, hydrogenous, and null-scattering samples agree with molecular dynamics simulations based on this model. From the MD model, we are able for the first time to interrogate the atomistic structure. The small-angle scattering is consistent with Porod scattering from particle surfaces, of the form Q^{-4}, where Q is the scattering vector. We measure a distinct peak in the scattering at Q = 0.45 Å^{-1}, corresponding to the first sharp diffraction peak in amorphous silica, which indicates the structural analogy between these two amorphous tetrahedral networks. △ Less

Submitted 13 August, 2021; originally announced August 2021.

Comments: 17 pages

Soft mode anisotropy in negative thermal expansion material ReO$_3$

Authors: Tobias A. Bird, Mark G. L. Wilkinson, David A. Keen, Ronald I. Smith, Nicholas C. Bristowe, Martin T. Dove, Anthony E. Phillips, Mark S. Senn

Abstract: We use a symmetry-motivated approach to analyse neutron pair distribution function data to investigate the mechanism of negative thermal expansion (NTE) in ReO$_3$. This analysis shows that the local structure of ReO$_3$ is dominated by an in-phase octahedral tilting mode and that the octahedral units are far less flexible to scissoring type deformations than the octahedra in the related compound… ▽ More We use a symmetry-motivated approach to analyse neutron pair distribution function data to investigate the mechanism of negative thermal expansion (NTE) in ReO$_3$. This analysis shows that the local structure of ReO$_3$ is dominated by an in-phase octahedral tilting mode and that the octahedral units are far less flexible to scissoring type deformations than the octahedra in the related compound ScF$_3$. These results support the idea that structural flexibility is an important factor in NTE materials, allowing the phonon modes that drive a volume contraction of the lattice to occupy a greater volume in reciprocal space. The lack of flexibility in ReO$_3$ restricts the NTE-driving phonons to a smaller region of reciprocal space, limiting the magnitude and temperature range of NTE. In addition, we investigate the thermal expansion properties of the material at high temperature and do not find the reported second NTE region. Finally, we show that the local fluctuations, even at elevated temperatures, respect the symmetry and order parameter direction of the observed $P4/mbm$ high pressure phase of ReO$_3$. The result indicates that the motions associated with rigid unit modes are highly anisotropic in these systems. △ Less

Submitted 6 August, 2021; originally announced August 2021.

Comments: 9 pages, 6 figures

One Representation to Rule Them All: Identifying Out-of-Support Examples in Few-shot Learning with Generic Representations

Authors: Henry Kvinge, Scott Howland, Nico Courts, Lauren A. Phillips, John Buckheit, Zachary New, Elliott Skomski, Jung H. Lee, Sandeep Tiwari, Jessica Hibler, Courtney D. Corley, Nathan O. Hodas

Abstract: The field of few-shot learning has made remarkable strides in developing powerful models that can operate in the small data regime. Nearly all of these methods assume every unlabeled instance encountered will belong to a handful of known classes for which one has examples. This can be problematic for real-world use cases where one routinely finds 'none-of-the-above' examples. In this paper we desc… ▽ More The field of few-shot learning has made remarkable strides in developing powerful models that can operate in the small data regime. Nearly all of these methods assume every unlabeled instance encountered will belong to a handful of known classes for which one has examples. This can be problematic for real-world use cases where one routinely finds 'none-of-the-above' examples. In this paper we describe this challenge of identifying what we term 'out-of-support' (OOS) examples. We describe how this problem is subtly different from out-of-distribution detection and describe a new method of identifying OOS examples within the Prototypical Networks framework using a fixed point which we call the generic representation. We show that our method outperforms other existing approaches in the literature as well as other approaches that we propose in this paper. Finally, we investigate how the use of such a generic point affects the geometry of a model's feature space. △ Less

Submitted 2 June, 2021; originally announced June 2021.

Comments: 15 pages

Controller Synthesis for Multi-Agent Systems with Intermittent Communication and Metric Temporal Logic Specifications

Authors: Zhe Xu, Federico M. Zegers, Bo Wu, Alexander J. Phillips, Warren Dixon, Ufuk Topcu

Abstract: This paper investigates the controller synthesis problem for a multi-agent system (MAS) with intermittent communication. We adopt a relay-explorer scheme, where a mobile relay agent with absolute position sensors switches among a set of explorers with relative position sensors to provide intermittent state information. We model the MAS as a switched system where the explorers' dynamics can be eith… ▽ More This paper investigates the controller synthesis problem for a multi-agent system (MAS) with intermittent communication. We adopt a relay-explorer scheme, where a mobile relay agent with absolute position sensors switches among a set of explorers with relative position sensors to provide intermittent state information. We model the MAS as a switched system where the explorers' dynamics can be either fully-actuated or underactuated. The objective of the explorers is to reach approximate consensus to a predetermined goal region. To guarantee the stability of the switched system and the approximate consensus of the explorers, we derive maximum dwell-time conditions to constrain the length of time each explorer goes without state feedback (from the relay agent). Furthermore, the relay agent needs to satisfy practical constraints such as charging its battery and staying in specific regions of interest. Both the maximum dwell-time conditions and these practical constraints can be expressed by metric temporal logic (MTL) specifications. We iteratively compute the optimal control inputs for the relay agent to satisfy the MTL specifications, while guaranteeing stability and approximate consensus of the explorers. We implement the proposed method on a case study with the CoppeliaSim robot simulator. △ Less

Submitted 5 February, 2021; originally announced April 2021.

Comments: arXiv admin note: substantial text overlap with arXiv:1909.09939

Critical processing temperature for high performance protected silver thin film mirrors

Authors: David M. Fryauf, Andrew C. Phillips, Nobuhiko P. Kobayashi

Abstract: Silver (Ag) mirrors for astronomical telescopes consist of multiple metallic and dielectric thin films. Furthermore, the topmost surface of such Ag mirrors needs to be covered by a protection coating. While the protection coating is often deposited at room temperature and the entire mirrors are also handled at room temperature, various thin film deposition techniques offer protection coatings with… ▽ More Silver (Ag) mirrors for astronomical telescopes consist of multiple metallic and dielectric thin films. Furthermore, the topmost surface of such Ag mirrors needs to be covered by a protection coating. While the protection coating is often deposited at room temperature and the entire mirrors are also handled at room temperature, various thin film deposition techniques offer protection coatings with improved characteristics when carried out at elevated temperatures. Thus, in this work, high-performance Ag mirrors were designed and fabricated with a new benchmark. The resulting Ag mirrors were annealed (i.e., post-fabrication annealing) at various temperatures to investigate the viability of introducing thermal processes during and/or after fabrication in improving overall optical performance and durability of protected silver mirrors. In our experiments, Ag mirror samples were deposited by electron-beam evaporation and subsequently annealed at various temperatures in the range from 60 °C to 300 °C, and then the mirror samples underwent an environmental stress test at 80 °C and 80% humidity for 10 days. While all the mirror samples annealed below 200 °C showed negligible corrosion after undergoing the stress testing, those annealed below 160 °C presented spectral reflectivity comparable to or higher than that of as-deposited reference samples. In contrast, the mirror samples annealed above 200 °C exhibited significant degradation after the stress testing. The comprehensive analysis indicated that delamination and voids caused by the growth of Ag grains during the annealing are the primary mechanisms of the degradation. △ Less

Submitted 16 April, 2021; originally announced April 2021.

Comments: 18 pages, 7 figures

Journal ref: J. Astron. Telesc. Instrum. Syst. 7(3), 034002 (2021)

What happened, and who cared? Evidencing research impact retrospectively

Authors: Chris D. White, Anthony Phillips, Beltran Sajonia-Coburgo-Gotha

Abstract: Higher Education Institutions in the UK and elsewhere are under increasing pressure to measure the impact of their research, which can include how the research has increased scientific engagement amongst the general public. For various reasons, the need for evidence can arise months, or even years, after a particular research discovery has been made. Furthermore, the right kind of evidence is need… ▽ More Higher Education Institutions in the UK and elsewhere are under increasing pressure to measure the impact of their research, which can include how the research has increased scientific engagement amongst the general public. For various reasons, the need for evidence can arise months, or even years, after a particular research discovery has been made. Furthermore, the right kind of evidence is needed to indicate genuine behavioural change amongst a given target audience, which can be difficult to obtain after time has passed. In this article, we present a number of strategies for retrospective evidencing of research engagement, and illustrate their use on example discoveries from up to five years ago. △ Less

Submitted 11 March, 2021; originally announced March 2021.

Comments: 17 pages, 12 figures

Report number: QMUL-PH-21-14

Not engaging with problems in the lab: Students' navigation of conflicting data and models

Authors: Anna McLean Phillips, Meagan Sundstrom, David G. Wu, N. G. Holmes

Abstract: With the adoption of instructional laboratories (labs) that require students to make their own decisions, there is a need to better understand students' activities as they make sense of their data and decide how to proceed. In particular, understanding when students do not engage productively with unexpected data may provide insights into how to better support students in more open-ended labs. We… ▽ More With the adoption of instructional laboratories (labs) that require students to make their own decisions, there is a need to better understand students' activities as they make sense of their data and decide how to proceed. In particular, understanding when students do not engage productively with unexpected data may provide insights into how to better support students in more open-ended labs. We examine video and audio data from groups within a lab session where students were expected to find data inconsistent with the predictions of two models. In prior work, we examined the actions of the four groups that productively grapple with this designed problem. Here, we analyze the engagement of the three groups that do not. We conducted three phases of analysis: 1) documenting large scale behaviors and time spent in on-topic discussion, 2) analyzing interactions with the teaching assistant, and 3) identifying students' framing--their expectations for what is taking place--when they were discussing their data. Our Phase 1 and 2 analysis show only minor differences between the groups that engaged with the problem and those that did not. Our Phase 3 analysis demonstrated that the groups that did not engage with the problem framed the lab activity as about confirming a known result or as a series of hoops to jump through to fulfill assignment requirements. Implications for instruction include supporting teaching assistants to attend to students' framing and agency within laboratory classrooms. △ Less

Submitted 2 March, 2021; originally announced March 2021.

Journal ref: Phys. Rev. Phys. Educ. Res. 17, 020112 (2021)

The Planet as Exoplanet Analog Spectrograph (PEAS): Design and First-Light

Authors: Emily C. Martin, Andrew J. Skemer, Matthew V. Radovan, Steven L. Allen, David Black, William T. S. Deich, Jonathan J. Fortney, Gabriel Kruglikov, Nicholas MacDonald, David Marques, Evan C. Morris, Andrew C. Phillips, Dale Sandford, Julissa Villalobos Valencia, Jason J. Wang, Pavl Zachary

Abstract: Exoplanets are abundant in our galaxy and yet characterizing them remains a technical challenge. Solar System planets provide an opportunity to test the practical limitations of exoplanet observations with high signal-to-noise data that we cannot access for exoplanets. However, data on Solar System planets differ from exoplanets in that Solar System planets are spatially resolved while exoplanets… ▽ More Exoplanets are abundant in our galaxy and yet characterizing them remains a technical challenge. Solar System planets provide an opportunity to test the practical limitations of exoplanet observations with high signal-to-noise data that we cannot access for exoplanets. However, data on Solar System planets differ from exoplanets in that Solar System planets are spatially resolved while exoplanets are unresolved point-sources. We present a novel instrument designed to observe Solar System planets as though they are exoplanets, the Planet as Exoplanet Analog Spectrograph (PEAS). PEAS consists of a dedicated 0.5-m telescope and off-the-shelf optics, located at Lick Observatory. PEAS uses an integrating sphere to disk-integrate light from the Solar System planets, producing spatially mixed light more similar to the spectra we can obtain from exoplanets. This paper describes the general system design and early results of the PEAS instrument. △ Less

Submitted 15 December, 2020; originally announced December 2020.

Comments: SPIE Astronomical Telescopes & Instrumentation 2020 Proceedings (11447-153), 9 pages, 9 figures

Universal $L^{-3}$ finite-size effects in the viscoelasticity of confined amorphous systems

Authors: A. E. Phillips, M. Baggioli, T. W. Sirk, K. Trachenko, A. Zaccone

Abstract: We present a theory of viscoelasticity of amorphous media, which takes into account the effects of confinement along one of three spatial dimensions. The framework is based on the nonaffine extension of lattice dynamics to amorphous systems, or nonaffine response theory. The size effects due to the confinement are taken into account via the nonaffine part of the shear storage modulus $G'$. The non… ▽ More We present a theory of viscoelasticity of amorphous media, which takes into account the effects of confinement along one of three spatial dimensions. The framework is based on the nonaffine extension of lattice dynamics to amorphous systems, or nonaffine response theory. The size effects due to the confinement are taken into account via the nonaffine part of the shear storage modulus $G'$. The nonaffine contribution is written as a sum over modes in $k$-space. With a rigorous argument based on the analysis of the $k$-space integral over modes, it is shown that the confinement size $L$ in one spatial dimension, e.g. the $z$ axis, leads to a infrared cut-off for the modes contributing to the nonaffine (softening) correction to the modulus that scales as $L^{-3}$. Corrections for finite sample size $D$ in the two perpendicular dimensions scale as $\sim (L/D)^4$, and are negligible for $L \ll D$. For liquids it is predicted that $G'\sim L^{-3}$ in agreement with a previous more approximate analysis, whereas for amorphous materials $G' \sim G'_{bulk} + βL^{-3}$. For the case of liquids, four different experimental systems are shown to be very well described by the $L^{-3}$ law. △ Less

Submitted 23 February, 2021; v1 submitted 9 December, 2020; originally announced December 2020.

Comments: Physical Review Materials, accepted, in press

Journal ref: Phys. Rev. Materials 5, 035602 (2021)

3-3-1 Self Interacting Dark Matter and the Galaxy Core-Cusp problem

Authors: Nguyen Quynh Lan, Grant J. Mathews, Lara Arielle Phillips, Miguel A. Correa, In-Saeng Suh, Jared. W. Coughlin

Abstract: The core-cusp problem remains as a challenging discrepancy between observations and simulations in the standard $Λ$CDM model for the formation of galaxies. The problem is that $Λ$CDM simulations predict a steep power-law mass density profile at the center of galactic dark matter halos. However, observations of dwarf galaxies in the Local Group reveal a density profile consistent with a nearly flat… ▽ More The core-cusp problem remains as a challenging discrepancy between observations and simulations in the standard $Λ$CDM model for the formation of galaxies. The problem is that $Λ$CDM simulations predict a steep power-law mass density profile at the center of galactic dark matter halos. However, observations of dwarf galaxies in the Local Group reveal a density profile consistent with a nearly flat distribution of dark matter near the center. A number of solutions to this dilemma have been proposed. Here, we summarize investigations into the possibility that the dark matter particles themselves self interact and scatter. Such self-interacting dark matter (SIDM) particles can smooth out the dark-matter profile in high-density regions. We also review the theoretical proposal that self-interacting dark matter may arise as an additional Higgs scalar in the 3-3-1 extension of the standard model. We present new simulations of galaxy formation and evolution for this formulation of self-interacting dark matter. Current constraints on this self-interacting dark matter are then summarized. △ Less

Submitted 3 December, 2020; originally announced December 2020.

Comments: 11 pages, 5 figures, accepted in MPLA

Fuzzy Simplicial Networks: A Topology-Inspired Model to Improve Task Generalization in Few-shot Learning

Authors: Henry Kvinge, Zachary New, Nico Courts, Jung H. Lee, Lauren A. Phillips, Courtney D. Corley, Aaron Tuor, Andrew Avila, Nathan O. Hodas

Abstract: Deep learning has shown great success in settings with massive amounts of data but has struggled when data is limited. Few-shot learning algorithms, which seek to address this limitation, are designed to generalize well to new tasks with limited data. Typically, models are evaluated on unseen classes and datasets that are defined by the same fundamental task as they are trained for (e.g. category… ▽ More Deep learning has shown great success in settings with massive amounts of data but has struggled when data is limited. Few-shot learning algorithms, which seek to address this limitation, are designed to generalize well to new tasks with limited data. Typically, models are evaluated on unseen classes and datasets that are defined by the same fundamental task as they are trained for (e.g. category membership). One can also ask how well a model can generalize to fundamentally different tasks within a fixed dataset (for example: moving from category membership to tasks that involve detecting object orientation or quantity). To formalize this kind of shift we define a notion of "independence of tasks" and identify three new sets of labels for established computer vision datasets that test a model's ability to generalize to tasks which draw on orthogonal attributes in the data. We use these datasets to investigate the failure modes of metric-based few-shot models. Based on our findings, we introduce a new few-shot model called Fuzzy Simplicial Networks (FSN) which leverages a construction from topology to more flexibly represent each class from limited data. In particular, FSN models can not only form multiple representations for a given class but can also begin to capture the low-dimensional structure which characterizes class manifolds in the encoded space of deep networks. We show that FSN outperforms state-of-the-art models on the challenging tasks we introduce in this paper while remaining competitive on standard few-shot benchmarks. △ Less

Submitted 23 September, 2020; originally announced September 2020.

Comments: 17 pages

A framework to evaluate whether to pool or separate behaviors in a multilayer network

Authors: Annemarie van der Marel, Sanjay Prasher, Chelsea Carminito, Claire O'Connell, Alexa Phillips, Bryan M. Kluever, Elizabeth A. Hobson

Abstract: A multilayer network approach combines different network layers, which are connected by interlayer edges, to create a single mathematical object. These networks can contain a variety of information types and represent different aspects of a system. However, the process for selecting which information to include is not always straightforward. Using data on two agonistic behaviors in a captive popul… ▽ More A multilayer network approach combines different network layers, which are connected by interlayer edges, to create a single mathematical object. These networks can contain a variety of information types and represent different aspects of a system. However, the process for selecting which information to include is not always straightforward. Using data on two agonistic behaviors in a captive population of monk parakeets (Myiopsitta monachus), we developed a framework for investigating how pooling or splitting behaviors at the scale of dyadic relationships (between two individuals) affects individual- and group-level social properties. We designed two reference models to test whether randomizing the number of interactions across behavior types results in similar structural patterns as the observed data. Although the behaviors were correlated, the first reference model suggests that the two behaviors convey different information about some social properties and should therefore not be pooled. However, once we controlled for data sparsity, we found that the observed measures corresponded with those from the second reference model. Hence, our initial result may have been due to the unequal frequencies of each behavior. Overall, our findings support pooling the two behaviors. Awareness of how selected measurements can be affected by data properties is warranted, but nonetheless our framework disentangles these efforts and as a result can be used for myriad types of behaviors and questions. This framework will help researchers make informed and data-driven decisions about which behaviors to pool or separate, prior to using the data in subsequent multilayer network analyses. △ Less

Submitted 20 December, 2020; v1 submitted 19 July, 2020; originally announced July 2020.

Comments: accepted for Current Zoology

Journal ref: Current Zoology, 2021

CheXphoto: 10,000+ Photos and Transformations of Chest X-rays for Benchmarking Deep Learning Robustness

Authors: Nick A. Phillips, Pranav Rajpurkar, Mark Sabini, Rayan Krishnan, Sharon Zhou, Anuj Pareek, Nguyet Minh Phu, Chris Wang, Mudit Jain, Nguyen Duong Du, Steven QH Truong, Andrew Y. Ng, Matthew P. Lungren

Abstract: Clinical deployment of deep learning algorithms for chest x-ray interpretation requires a solution that can integrate into the vast spectrum of clinical workflows across the world. An appealing approach to scaled deployment is to leverage the ubiquity of smartphones by capturing photos of x-rays to share with clinicians using messaging services like WhatsApp. However, the application of chest x-ra… ▽ More Clinical deployment of deep learning algorithms for chest x-ray interpretation requires a solution that can integrate into the vast spectrum of clinical workflows across the world. An appealing approach to scaled deployment is to leverage the ubiquity of smartphones by capturing photos of x-rays to share with clinicians using messaging services like WhatsApp. However, the application of chest x-ray algorithms to photos of chest x-rays requires reliable classification in the presence of artifacts not typically encountered in digital x-rays used to train machine learning models. We introduce CheXphoto, a dataset of smartphone photos and synthetic photographic transformations of chest x-rays sampled from the CheXpert dataset. To generate CheXphoto we (1) automatically and manually captured photos of digital x-rays under different settings, and (2) generated synthetic transformations of digital x-rays targeted to make them look like photos of digital x-rays and x-ray films. We release this dataset as a resource for testing and improving the robustness of deep learning algorithms for automated chest x-ray interpretation on smartphone photos of chest x-rays. △ Less

Submitted 11 December, 2020; v1 submitted 13 July, 2020; originally announced July 2020.

Constant-Depth and Subcubic-Size Threshold Circuits for Matrix Multiplication

Authors: Ojas Parekh, Cynthia A. Phillips, Conrad D. James, James B. Aimone

Abstract: Boolean circuits of McCulloch-Pitts threshold gates are a classic model of neural computation studied heavily in the late 20th century as a model of general computation. Recent advances in large-scale neural computing hardware has made their practical implementation a near-term possibility. We describe a theoretical approach for multiplying two $N$ by $N$ matrices that integrates threshold gate lo… ▽ More Boolean circuits of McCulloch-Pitts threshold gates are a classic model of neural computation studied heavily in the late 20th century as a model of general computation. Recent advances in large-scale neural computing hardware has made their practical implementation a near-term possibility. We describe a theoretical approach for multiplying two $N$ by $N$ matrices that integrates threshold gate logic with conventional fast matrix multiplication algorithms, that perform $O(N^ω)$ arithmetic operations for a positive constant $ω< 3$. Our approach converts such a fast matrix multiplication algorithm into a constant-depth threshold circuit with approximately $O(N^ω)$ gates. Prior to our work, it was not known whether the $Θ(N^3)$-gate barrier for matrix multiplication was surmountable by constant-depth threshold circuits. Dense matrix multiplication is a core operation in convolutional neural network training. Performing this work on a neural architecture instead of off-loading it to a GPU may be an appealing option. △ Less

Submitted 25 June, 2020; originally announced June 2020.

Comments: Appears in the proceedings of the ACM Symposium on Parallelism in Algorithms and Architectures (SPAA), 2018

Fractional Decomposition Tree Algorithm: A tool for studying the integrality gap of Integer Programs

Authors: Robert D. Carr, Arash Haddadan, Cynthia A. Phillips

Abstract: We present a new algorithm, Fractional Decomposition Tree (FDT) for finding a feasible solution for an integer program (IP) where all variables are binary. FDT runs in polynomial time and is guaranteed to find a feasible integer solution provided the integrality gap is bounded. The algorithm gives a construction for Carr and Vempala's theorem that any feasible solution to the IP's linear-programmi… ▽ More We present a new algorithm, Fractional Decomposition Tree (FDT) for finding a feasible solution for an integer program (IP) where all variables are binary. FDT runs in polynomial time and is guaranteed to find a feasible integer solution provided the integrality gap is bounded. The algorithm gives a construction for Carr and Vempala's theorem that any feasible solution to the IP's linear-programming relaxation, when scaled by the instance integrality gap, dominates a convex combination of feasible solutions. FDT is also a tool for studying the integrality gap of IP formulations. We demonstrate that with experiments studying the integrality gap of two problems: optimally augmenting a tree to a 2-edge-connected graph and finding a minimum-cost 2-edge-connected multi-subgraph (2EC). We also give a simplified algorithm, Dom2IP, that more quickly determines if an instance has an unbounded integrality gap. We show that FDT's speed and approximation quality compare well to that of feasibility pump on moderate-sized instances of the vertex cover problem. For a particular set of hard-to-decompose fractional 2EC solutions, FDT always gave a better integer solution than the best previous approximation algorithm (Christofides). △ Less

Submitted 11 August, 2020; v1 submitted 12 June, 2020; originally announced June 2020.

LAReQA: Language-agnostic answer retrieval from a multilingual pool

Authors: Uma Roy, Noah Constant, Rami Al-Rfou, Aditya Barua, Aaron Phillips, Yinfei Yang

Abstract: We present LAReQA, a challenging new benchmark for language-agnostic answer retrieval from a multilingual candidate pool. Unlike previous cross-lingual tasks, LAReQA tests for "strong" cross-lingual alignment, requiring semantically related cross-language pairs to be closer in representation space than unrelated same-language pairs. Building on multilingual BERT (mBERT), we study different strateg… ▽ More We present LAReQA, a challenging new benchmark for language-agnostic answer retrieval from a multilingual candidate pool. Unlike previous cross-lingual tasks, LAReQA tests for "strong" cross-lingual alignment, requiring semantically related cross-language pairs to be closer in representation space than unrelated same-language pairs. Building on multilingual BERT (mBERT), we study different strategies for achieving strong alignment. We find that augmenting training data via machine translation is effective, and improves significantly over using mBERT out-of-the-box. Interestingly, the embedding baseline that performs the best on LAReQA falls short of competing baselines on zero-shot variants of our task that only target "weak" alignment. This finding underscores our claim that languageagnostic retrieval is a substantively new kind of cross-lingual evaluation. △ Less

Submitted 11 April, 2020; originally announced April 2020.

Probing a Set of Trajectories to Maximize Captured Information

Authors: Sándor P. Fekete, Alexander Hill, Dominik Krupke, Tyler Mayer, Joseph S. B. Mitchell, Ojas Parekh, Cynthia A. Phillips

Abstract: We study a trajectory analysis problem we call the Trajectory Capture Problem (TCP), in which, for a given input set ${\cal T}$ of trajectories in the plane, and an integer $k\geq 2$, we seek to compute a set of $k$ points (``portals'') to maximize the total weight of all subtrajectories of ${\cal T}$ between pairs of portals. This problem naturally arises in trajectory analysis and summarization.… ▽ More We study a trajectory analysis problem we call the Trajectory Capture Problem (TCP), in which, for a given input set ${\cal T}$ of trajectories in the plane, and an integer $k\geq 2$, we seek to compute a set of $k$ points (``portals'') to maximize the total weight of all subtrajectories of ${\cal T}$ between pairs of portals. This problem naturally arises in trajectory analysis and summarization. We show that the TCP is NP-hard (even in very special cases) and give some first approximation results. Our main focus is on attacking the TCP with practical algorithm-engineering approaches, including integer linear programming (to solve instances to provable optimality) and local search methods. We study the integrality gap arising from such approaches. We analyze our methods on different classes of data, including benchmark instances that we generate. Our goal is to understand the best performing heuristics, based on both solution time and solution quality. We demonstrate that we are able to compute provably optimal solutions for real-world instances. △ Less

Submitted 7 April, 2020; originally announced April 2020.

Colossal pressure-induced softening in scandium fluoride

Authors: Zhongsheng Wei, Lei Tan, Guanqun Cai, Anthony E Phillips, Ivan da Silva, Mark G Kibble, Martin T Dove

Abstract: The counter-intuitive phenomenon of pressure-induced softening in materials is likely to be caused by the same dynamical behaviour that produces negative thermal expansion. Through a combination of molecular dynamics simulation on an idealised model and neutron diffraction at variable temperature and pressure, we show the existence of extraordinary and unprecedented pressure-induced softening in t… ▽ More The counter-intuitive phenomenon of pressure-induced softening in materials is likely to be caused by the same dynamical behaviour that produces negative thermal expansion. Through a combination of molecular dynamics simulation on an idealised model and neutron diffraction at variable temperature and pressure, we show the existence of extraordinary and unprecedented pressure-induced softening in the negative thermal expansion material scandium fluoride, ScF$_3$, with values of the pressure-derivative of the bulk modulus $B$, $B^\prime = \partial B / \partial P$, reaching as low as $-40 \pm 1$. △ Less

Submitted 30 March, 2020; originally announced March 2020.

Journal ref: Phys. Rev. Lett. 124, 255502 (2020)

Efficient Amortised Bayesian Inference for Hierarchical and Nonlinear Dynamical Systems

Authors: Geoffrey Roeder, Paul K. Grant, Andrew Phillips, Neil Dalchau, Edward Meeds

Abstract: We introduce a flexible, scalable Bayesian inference framework for nonlinear dynamical systems characterised by distinct and hierarchical variability at the individual, group, and population levels. Our model class is a generalisation of nonlinear mixed-effects (NLME) dynamical systems, the statistical workhorse for many experimental sciences. We cast parameter inference as stochastic optimisation… ▽ More We introduce a flexible, scalable Bayesian inference framework for nonlinear dynamical systems characterised by distinct and hierarchical variability at the individual, group, and population levels. Our model class is a generalisation of nonlinear mixed-effects (NLME) dynamical systems, the statistical workhorse for many experimental sciences. We cast parameter inference as stochastic optimisation of an end-to-end differentiable, block-conditional variational autoencoder. We specify the dynamics of the data-generating process as an ordinary differential equation (ODE) such that both the ODE and its solver are fully differentiable. This model class is highly flexible: the ODE right-hand sides can be a mixture of user-prescribed or "white-box" sub-components and neural network or "black-box" sub-components. Using stochastic optimisation, our amortised inference algorithm could seamlessly scale up to massive data collection pipelines (common in labs with robotic automation). Finally, our framework supports interpretability with respect to the underlying dynamics, as well as predictive generalization to unseen combinations of group components (also called "zero-shot" learning). We empirically validate our method by predicting the dynamic behaviour of bacteria that were genetically engineered to function as biosensors. Our implementation of the framework, the dataset, and all code to reproduce the experimental results is available at https://www.github.com/Microsoft/vi-hds . △ Less

Submitted 1 October, 2019; v1 submitted 28 May, 2019; originally announced May 2019.

Comments: Published in "Proceedings of Machine Learning Research, Volume 97: International Conference on Machine Learning, 9-15 June 2019, Long Beach, California, USA"