Search

In 2020 preschool participation rates substantially declined. A recent NIEER survey found that the Fall 2022 preschool participation rates had nearly returned to pre-pandemic 2019 levels. The NIEER survey asked parents of young children who were still not in preschool in the Fall of 2022 to report why they did not enroll their children. The top responses included having one parent home (82%), cost and lack of affordable options (41%), and continuing concerns about the pandemic (35%). In NIEER's Spring 2021 PLA survey, 81 percent of parents of age-eligible children said they would very likely or somewhat likely send their child to a free, universal pre-K program if it became available. In addition, parents with annual incomes below $25,000 more often reported being very likely to participate than those with higher incomes. [For "Impacts of the Pandemic on Young Children and Their Parents: Initial Findings from NIEER's May-June 2021 Preschool Learning Activities Survey. Research Report," see ED616245.]

Factor analysis (FA) and principal component analysis (PCA) are popular statistical methods for summarizing and explaining the variability in multivariate datasets. By default, FA and PCA assume the number of components or factors to be known \emph{a priori}. However, in practice the users first estimate the number of factors or components and then perform FA and PCA analyses using the point estimate. Therefore, in practice the users ignore any uncertainty in the point estimate of the number of factors or components. For datasets where the uncertainty in the point estimate is not ignorable, it is prudent to perform FA and PCA analyses for the range of positive integer values in the confidence intervals for the number of factors or components. We address this problem by proposing a subsampling-based data-intensive approach for estimating confidence intervals for the number of components in FA and PCA. We study the coverage probability of the proposed confidence intervals and provide non-asymptotic theoretical guarantees concerning the accuracy of the confidence intervals. As a byproduct, we derive the first-order \emph{Edgeworth expansion} for spiked eigenvalues of the sample covariance matrix when the data matrix is generated under a factor model. We also demonstrate the usefulness of our approach through numerical simulations and by applying our approach for estimating confidence intervals for the number of factors of the genotyping dataset of the Human Genome Diversity Project.

Meeting the rise of industry demand to incorporate machine learning (ML) components into software systems requires interdisciplinary teams contributing to a shared code base. To maintain consistency, reduce defects and ensure maintainability, developers use code analysis tools to aid them in identifying defects and maintaining standards. With the inclusion of machine learning, tools must account for the cultural differences within the teams which manifests as multiple programming languages, and conflicting definitions and objectives. Existing tools fail to identify these cultural differences and are geared towards software engineering which reduces their adoption in ML projects. In our approach we attempt to resolve this problem by exploring the use of context which includes i) purpose of the source code, ii) technical domain, iii) problem domain, iv) team norms, v) operational environment, and vi) development lifecycle stage to provide contextualised error reporting for code analysis. To demonstrate our approach, we adapt Pylint as an example and apply a set of contextual transformations to the linting results based on the domain of individual project files under analysis. This allows for contextualised and meaningful error reporting for the end-user., Comment: ICSE 2022 NIER track

Planetary obliquity is a first order control on planetary climate and seasonal contrast, which has a number of cascading consequences for life. How moderately high obliquity (obliquities greater than Earth's current obliquity up to 45$^{\circ}$) affects a planet's surface physically has been studied previously, but we lack an understanding of how marine life will respond to these conditions. We couple the ROCKE-3D general circulation model to the cGENIE 3D biogeochemical model to simulate the ocean biosphere's response to various planetary obliquities, bioessential nutrient inventories, and biospheric structure. We find that the net rate of photosynthesis increased by 35$\%$ and sea-to-air flux of biogenic oxygen doubled between the 0$^{\circ}$ and 45$^{\circ}$ obliquity scenarios, which is an equivalent response to doubling bioessential nutrients. Our results suggest that moderately high-obliquity planets have higher potential for biospheric oxygenation than their low-obliquity counterparts and that life on moderately high-obliquity habitable planets may be easier to detect with next generation telescopes. These moderately high-obliquity habitable planets may also be more conducive to the evolution of complex life.

Federated learning (FL) has been widely employed for medical image analysis to facilitate multi-client collaborative learning without sharing raw data. Despite great success, FL's performance is limited for multiple sclerosis (MS) lesion segmentation tasks, due to variance in lesion characteristics imparted by different scanners and acquisition parameters. In this work, we propose the first FL MS lesion segmentation framework via two effective re-weighting mechanisms. Specifically, a learnable weight is assigned to each local node during the aggregation process, based on its segmentation performance. In addition, the segmentation loss function in each client is also re-weighted according to the lesion volume for the data during training. Comparison experiments on two FL MS segmentation scenarios using public and clinical datasets have demonstrated the effectiveness of the proposed method by outperforming other FL methods significantly. Furthermore, the segmentation performance of FL incorporating our proposed aggregation mechanism can exceed centralised training with all the raw data. The extensive evaluation also indicated the superiority of our method when estimating brain volume differences estimation after lesion inpainting., Comment: 10 pages, 3 figures, and 7 tables

The use of `backfilling', assigning additional patients to doses deemed safe, in phase I dose-escalation studies has been used in practice to collect additional information on the safety profile, pharmacokinetics and activity of a drug. These additional patients help ensure that the Maximum Tolerated Dose (MTD) is reliably estimated and give additional information in order to determine the recommended phase II dose (RP2D). In this paper, we study the effect of employing backfilling in a phase I trial on the estimation of the MTD and the duration of the study. We consider the situation where only one cycle of follow-up is used for escalation as well as the case where there may be delayed onset toxicities. We find that, over a range of scenarios, there is an increase in the proportion of correct selections and a notable reduction in the trial duration at the cost of more patients required in the study.

On 25th February 2022, increased gamma radiation dose rates were reported within the Chornobyl Exclusion Zone (CEZ). This coincided with Russian military vehicles entering the Ukrainian part of the CEZ from neighbouring Belarus. It was speculated that contaminated soil resuspension by vehicle movements or a leak from the Chornobyl Nuclear Power Plant (ChNPP) complex may explain these spikes in radiation dose rates. The gamma dose rate monitoring network in the CEZ provides a crucial early warning system for releases of radioactivity to the environment and is part of the international safeguards for nuclear facilities. With the potential for further military action in the CEZ and concerns over nuclear safety, it is essential that such anomalous readings are investigated. We evaluate the hypotheses suggested to explain the apparent gamma dose rate increases, demonstrating that neither military vehicle-induced soil resuspension nor a leak from the ChNPP are plausible. However, disruption of the Chornobyl base-station's reception of wireless signals from the CRMS network may potentially explain the dose rate increases recorded., Comment: 14 pages including 5 figures. M.D.W. conceived the study; M.D.W. and N.A.B. wrote the paper; All authors discussed results and commented on the manuscript; C.L.B compiled the CRMS gamma dose rate data, which M.D.W. and N.A.B QCd; P.H.B. developed the modelling approach, with input from N.A.B., M.D.W. and S.M.; M.D.W. developed the figures; C.L.B compiled the reference list

A quadratic approximation manifold is presented for performing nonlinear, projection-based, model order reduction (PMOR). It constitutes a departure from the traditional affine subspace approximation that is aimed at mitigating the Kolmogorov barrier for nonlinear PMOR, particularly for convection-dominated transport problems. It builds on the data-driven approach underlying the traditional construction of projection-based reduced-order models (PROMs); is application-independent; is linearization-free; and therefore is robust for highly nonlinear problems. Most importantly, this approximation leads to quadratic PROMs that deliver the same accuracy as their traditional counterparts using however a much smaller dimension -- typically, $n_2 \sim \sqrt n_1$, where $n_2$ and $n_1$ denote the dimensions of the quadratic and traditional PROMs, respectively. The computational advantages of the proposed high-order approach to nonlinear PMOR over the traditional approach are highlighted for the detached-eddy simulation-based prediction of the Ahmed body turbulent wake flow, which is a popular CFD benchmark problem in the automotive industry. For a fixed accuracy level, these advantages include: a reduction of the total offline computational cost by a factor greater than five; a reduction of its online wall clock time by a factor greater than 32; and a reduction of the wall clock time of the underlying high-dimensional model by a factor greater than two orders of magnitude.

In Phase I/II dose-finding trials, the objective is to find the Optimal Biological Dose (OBD), a dose that is both safe and efficacious that maximises some optimality criterion based on safety and efficacy. This is further complicated when the investigated treatment consists of multiple cycles of therapy, and both toxicity and efficacy outcomes may occur at any point throughout the follow up of multiple cycles. In this work we present the Joint TITE-CRM, a model-based design for late onset toxicities and efficacies based on the well-known TITE-CRM. It is found to be superior to both currently available alternative designs that account for late onset bivariate outcomes; a model-assisted method and a bivariate survival design, as well as being both intuitive and computationally feasible.

Stated simply the paradox is as follows: it is clear that the orbital angular momentum of a light beam in its direction of propagation is an intrinsic quantity, and therefore has the same value everywhere in the beam. How then can a Gaussian beam, with precisely zero orbital angular momentum, drive a (single-photon) quadrupole transition which requires the transfer of angular momentum 2$\hbar$ to an absorbing atom?, Comment: Submitted to J.Phys.A. Special Issue "Claritons and the Asymptotics of Ideas: the Physics of Michael Berry"

This brief draws on a study of teacher working conditions and their relationship to teacher retention and school performance in North Carolina. It is part of a series of studies conducted by the Learning Policy Institute (LPI)--in collaboration with WestEd and the Friday Institute for Educational Innovation at North Carolina State University--as part of an action plan developed to inform ongoing efforts to ensure compliance with the North Carolina Supreme Court's decision in "Leandro v. the State of North Carolina." That case affirmed the state's constitutional responsibility to provide every student an equal opportunity for a sound basic education, including access to qualified teachers and administrators. Requested by the court in conjunction with both plaintiffs and defendants, the action plan aims to identify root causes of current inequalities and evidence-based solutions to meet the constitutional standard. This brief is based on an LPI report published in 2019: "How Teaching and Learning Conditions Affect Teacher Retention and School Performance in North Carolina."

Interpretable machine learning has demonstrated impressive performance while preserving explainability. In particular, neural additive models (NAM) offer the interpretability to the black-box deep learning and achieve state-of-the-art accuracy among the large family of generalized additive models. In order to empower NAM with feature selection and improve the generalization, we propose the sparse neural additive models (SNAM) that employ the group sparsity regularization (e.g. Group LASSO), where each feature is learned by a sub-network whose trainable parameters are clustered as a group. We study the theoretical properties for SNAM with novel techniques to tackle the non-parametric truth, thus extending from classical sparse linear models such as the LASSO, which only works on the parametric truth. Specifically, we show that SNAM with subgradient and proximal gradient descents provably converges to zero training loss as $t\to\infty$, and that the estimation error of SNAM vanishes asymptotically as $n\to\infty$. We also prove that SNAM, similar to LASSO, can have exact support recovery, i.e. perfect feature selection, with appropriate regularization. Moreover, we show that the SNAM can generalize well and preserve the `identifiability', recovering each feature's effect. We validate our theories via extensive experiments and further testify to the good accuracy and efficiency of SNAM.

We present a method to measure the optical torque applied to particles of arbitrary shape held in an optical trap, inferred from the change of angular momentum of light induced by the particle. All torque components can be determined from a single interference pattern recorded by a camera in the back focal plane of a high-NA condenser collecting all forward scattered light. We derive explicit expressions mapping the measured complex field in this plane to the torque components. The required phase is retrieved by an iterative algorithm, using the known position of the optical traps as constraints. The torque pertaining to individual particles is accessible, as well as separate spin or orbital parts of the total torque., Comment: V2, with updated figures

Local topological markers are effective tools for determining the topological properties of both homogeneous and inhomogeneous systems. The Chern marker is an established topological marker that has previously been shown to effectively reveal the topological properties of 2D systems. In an earlier work, the present authors have developed a marker that can be applied to 1D time-dependent systems which can be used to explore their topological properties, like charge pumping under the presence of disorder. In this paper, we show how to alter the 1D marker so that it can be applied to quasiperiodic and aperiodic systems. We then verify its effectiveness against different quasicrystal Hamiltonians, some which have been addressed in previous studies using existing methods, and others which possess topological structures that have been largely unexplored. We also demonstrate that the altered 1D marker can be productively applied to systems that are fully aperiodic.

Data collected from wearable devices and smartphones can shed light on an individual's pattern of behavioral and circadian routine. Phone use can be modeled as alternating event process, between the state of active use and the state of being idle. Markov chains and alternating recurrent event models are commonly used to model state transitions in cases such as these, and the incorporation of random effects can be used to introduce diurnal effects. While state labels can be derived prior to modeling dynamics, this approach omits informative regression covariates that can influence state memberships. We instead propose an alternating recurrent event proportional hazards (PH) regression to model the transitions between latent states. We propose an Expectation-Maximization (EM) algorithm for imputing latent state labels and estimating regression parameters. We show that our E-step simplifies to the hidden Markov model (HMM) forward-backward algorithm, allowing us to recover a HMM with logistic regression transition probabilities. In addition, we show that PH modeling of discrete-time transitions implicitly penalizes the logistic regression likelihood and results in shrinkage estimators for the relative risk. We derive asymptotic distributions for our model parameter estimates and compare our approach against competing methods through simulation as well as in a digital phenotyping study that followed smartphone use in a cohort of adolescents with mood disorders.

We introduce a new periodicity detection algorithm for binary time series of event onsets, the Gaussian Mixture Periodicity Detection Algorithm (GMPDA). The algorithm approaches the periodicity detection problem to infer the parameters of a generative model. We specified two models - the Clock and Random Walk - which describe two different periodic phenomena and provide a generative framework. The algorithm achieved strong results on test cases for single and multiple periodicity detection and varying noise levels. The performance of GMPDA was also evaluated on real data, recorded leg movements during sleep, where GMPDA was able to identify the expected periodicities despite high noise levels. The paper's key contributions are two new models for generating periodic event behavior and the GMPDA algorithm for multiple periodicity detection, which is highly accurate under noise.

Network-based clustering methods frequently require the number of communities to be specified \emph{a priori}. Moreover, most of the existing methods for estimating the number of communities assume the number of communities to be fixed and not scale with the network size $n$. The few methods that assume the number of communities to increase with the network size $n$ are only valid when the average degree $d$ of a network grows at least as fast as $O(n)$ (i.e., the dense case) or lies within a narrow range. This presents a challenge in clustering large-scale network data, particularly when the average degree $d$ of a network grows slower than the rate of $O(n)$ (i.e., the sparse case). To address this problem, we proposed a new sequential procedure utilizing multiple hypothesis tests and the spectral properties of Erd\"{o}s R\'{e}nyi graphs for estimating the number of communities in sparse stochastic block models (SBMs). We prove the consistency of our method for sparse SBMs for a broad range of the sparsity parameter. As a consequence, we discover that our method can estimate the number of communities $K^{(n)}_{\star}$ with $K^{(n)}_{\star}$ increasing at the rate as high as $O(n^{(1 - 3\gamma)/(4 - 3\gamma)})$, where $d = O(n^{1 - \gamma})$. Moreover, we show that our method can be adapted as a stopping rule in estimating the number of communities in binary tree stochastic block models. We benchmark the performance of our method against other competing methods on six reference single-cell RNA sequencing datasets. Finally, we demonstrate the usefulness of our method through numerical simulations and by using it for clustering real single-cell RNA-sequencing datasets.

Recent advances in unsupervised domain adaptation (UDA) techniques have witnessed great success in cross-domain computer vision tasks, enhancing the generalization ability of data-driven deep learning architectures by bridging the domain distribution gaps. For the UDA-based cross-domain object detection methods, the majority of them alleviate the domain bias by inducing the domain-invariant feature generation via adversarial learning strategy. However, their domain discriminators have limited classification ability due to the unstable adversarial training process. Therefore, the extracted features induced by them cannot be perfectly domain-invariant and still contain domain-private factors, bringing obstacles to further alleviate the cross-domain discrepancy. To tackle this issue, we design a Domain Disentanglement Faster-RCNN (DDF) to eliminate the source-specific information in the features for detection task learning. Our DDF method facilitates the feature disentanglement at the global and local stages, with a Global Triplet Disentanglement (GTD) module and an Instance Similarity Disentanglement (ISD) module, respectively. By outperforming state-of-the-art methods on four benchmark UDA object detection tasks, our DDF method is demonstrated to be effective with wide applicability., Comment: Accepted to appear in IEEE Transactions on Multimedia; source code: https://github.com/dliu5812/DDF

Jupiter's enhancement in nitrogen relative to hydrogen when compared to the Sun has been interpreted as evidence that its early formation occurred beyond the N$_{2}$ snowline ($\sim$ 20-40 AU). However, the rapid growth necessary to form Jupiter before the dissipation of the solar nebula would lead to the forming planet's core reaching very high temperatures ($>$1000 K), which would lead to it warming its surroundings. Here, we explore the effects of a luminous planetary core on the solids that it ultimately accretes. We find that a critical transition occurs where very hot (rapidly accreting) cores drive off volatiles prior to accretion, while cool cores (slowly accreting) are able to inherit volatile rich solids. Given Jupiter's nitrogen enrichment, if it formed beyond the N$_{2}$ snowline, its core could not have accreted solids at a rate above 10$^{-10}$ M$_{\odot}$ yr$^{-1}$. Our results suggest that either Jupiter formed in more distal regions of the solar nebula, or nitrogen loss was suppressed, either by its incorporation in more refractory carriers or because it was trapped within ices which devolatilized at higher temperatures., Comment: Accepted to the Astrophysical Journal, 11 pages, 5 figures

Any treatment of magnetic interactions between atoms, molecules and optical media must start at the form of the interaction energy. This forms the base on which predictions about any number of magnetic atom-light properties stands -- spontaneous decay rates and forces included. As is well-known, the Heaviside-Larmor duality symmetry of Maxwell's equations, where electric and magnetic quantities are exchanges, is broken by the usual form of the magnetic interaction energy. We argue that this symmetry can be restored by including general local-field effects, and that local fields should be treated as a necessity for correctly translating between the microscopic world of the dipole and the macroscopic world of the measured fields. This may additionally aid in resolving a long standing debate over the form of the force on a dipole in a medium. Finally, we compute the magnetic dipole decay rate in a magneto-dielectric with local-field effects taken into account, and show that macroscopic quantum electrodynamics can be made to be dual symmetric at an operator level, instead of only for expectation values., Comment: 9 pages, 4 figures, plus appendices. Comments welcome

Language is not only used to transmit neutral information; we often seek to persuade by arguing in favor of a particular view. Persuasion raises a number of challenges for classical accounts of belief updating, as information cannot be taken at face value. How should listeners account for a speaker's "hidden agenda" when incorporating new information? Here, we extend recent probabilistic models of recursive social reasoning to allow for persuasive goals and show that our model provides a pragmatic account for why weakly favorable arguments may backfire, a phenomenon known as the weak evidence effect. Critically, this model predicts a systematic relationship between belief updates and expectations about the information source: weak evidence should only backfire when speakers are expected to act under persuasive goals and prefer the strongest evidence. We introduce a simple experimental paradigm called the Stick Contest to measure the extent to which the weak evidence effect depends on speaker expectations, and show that a pragmatic listener model accounts for the empirical data better than alternative models. Our findings suggest further avenues for rational models of social reasoning to illuminate classical decision-making phenomena., Comment: in press at Open Mind

The "State of Preschool 2022" annual report covers the 2021-2022 school year and is NIEER's 20th report tracking preschool enrollment, funding, and policies state-by-state. This report focuses on the recovery since the 2020-2021 school year, where we are compared to before the pandemic, where we've come over the last two decades, and what needs to come next. Across the 44 states, DC, and Guam that funded a preschool program in 2021-2022, 32% of 4-year-olds and 6% of 3-year-olds were enrolled, an increase if 13% from the prior year, but still 8% below the pre-pandemic high. This pattern was seen in nearly every state. States spent $9.51 billion on preschool and also used $393 million in COVID-19 relief funding, without which inflation-adjusted spending would have declined. A big concern is the preschool workforce where we found unprecedented teacher shortages as well as waivers to education and specialized training requirements resulting in fewer qualified teachers in preschool classrooms. Yet few states provided incentives for teacher retention or recruitment. Over the last 20 years, preschool enrollment has more than doubled and so has inflation-adjusted spending. Spending per child in 2021-2022 was $6,571, essentially the same as it was 20 years ago after adjusting for inflation. Most children continue to lack access to high quality preschool. Though seven states currently working towards universal preschool offer a glimmer of hope - California, Colorado, Hawaii, Illinois, Michigan, New Jersey, and New Mexico. Success in these states could dramatically alter the early childhood education (ECE) landscape nationally. In the report, NIEER calls on states to take stock of their state's ECE system and calls on the federal government to provide more support to states. [For "The State of Preschool 2021: State Preschool Yearbook," see ED620548.]

In the United States, the prevailing high school mathematics course sequence begins with a year of Algebra I, followed by a year of geometry and a year of Algebra II. Educators and others have raised concerns about the extent to which this sequence, which prioritizes the mastery of algebra, is appropriate for the longer-term education and career goals of students who do not intend to pursue STEM degrees in college. These concerns have impelled educators and policymakers to reexamine the prominence of algebra in high school mathematics curricula and to consider new approaches that provide students with more mathematics course options better aligned with their academic and career goals. In this paper, we explore existing approaches to high school mathematics curricula as well as new developments in the field. To start, we discuss a range of high school mathematics course sequences that are currently offered across the country and look at some of the systemic challenges embedded within the traditional paradigm. Then we explore federal and state changes to the provision of high school mathematics in the early 21st century, which we follow with a look at the influence of postsecondary institutions on high school math curricula. We then introduce short case studies of innovative high school math reforms that are occurring in five states. We conclude the paper by considering the Charles A. Dana Center's new initiative, Launch Years, and how this project works to reimagine high school mathematics and its relationship to postsecondary education and careers.

This report presents findings on the implementation of a technology-mediated advising redesign within the context of a randomized controlled trial (RCT) at three institutions--the University of North Carolina at Charlotte, California State University at Fresno, and Montgomery County Community College. These colleges aimed to improve their advising systems using a three-pronged approach: expanding informational communications to students, identifying and supporting students who are struggling, and improving advising sessions. Though the RCT did not find significant changes in quantitative student outcomes such as grade point average, qualitative fieldwork suggests that stakeholders across the three institutions gained valuable experiences that may provide lessons for other colleges considering advising reforms. In this report, the authors explore how advisors adjusted their approach to advising work, how they used relevant tools and resources, and some of the challenges that arose during the redesign's implementation. Based on these implementation findings, the authors offer several suggestions for other colleges entering this work: (1) Acknowledge up front that technology alone cannot resolve advising and student support issues, nor can it replace face-to-face conversations about education and career planning; (2) Consider ways to encourage or make it easier for students to self-report concerns; (3) Call for input from advisors, faculty, students, and communications or marketing teams on the content of risk-related messages and alerts; (4) Consider alternatives to registration holds to incentivize students to meet with an advisor, as registration holds may have unintended consequences; (5) Give advisors the time, resources, and professional development to proactively engage with students; and (6) Provide ongoing support for staff in implementing reformed advising practices.

Virtually all community colleges and more than 90 percent of public four-year colleges use the results of placement tests--either alone or in concert with other information--to determine whether students are ready for college-level coursework or need remedial help in math or English. Evidence suggests that placement tests do a poor job of indicating which students need remediation. The Center for the Analysis of Postsecondary Readiness (CAPR) is studying an alternative placement system that uses multiple measures--including both placement test scores and high school GPAs--in predictive algorithms to place incoming students into remedial or college-level courses. Seven community colleges in the State University of New York system participated in the random assignment study to determine whether multiple measures placement leads to better student outcomes than a system based on test scores alone. Using multiple measures placement, many more students were assigned to college-level courses. In math, gains in college-level enrollment and completion were small and short-lived. But in English, the effects were much larger and lasted through at least three semesters. Regardless of whether they were predicted to succeed, students did better when they were allowed to start in college-level courses. A report on longer-term outcomes from the study will be released in summer 2022.

Community college students invest considerable time and money into attending college, but too few of them complete their programs of study. This paper discusses how the development of an overall framing vision for student success, the implementation of evidence-based practices, and the establishment of a culture that is both committed to student success and conducive to innovation are useful, complementary approaches that colleges can pursue to improve student outcomes.

Aspen and the Community College Research Center studied nine dual enrollment programs across three states--Florida, Ohio, and Washington--that have narrowed or closed equity gaps in dual enrollment for Black, Latinx, Indigenous, and Pacific Islander students. Based on this research, "The Dual Enrollment Playbook" describes the five principles that college and K-12 leaders can follow to design equitable dual enrollment programs. These five principles reflect the lessons that will make the most difference for historically underrepresented students of color, but these strategies may benefit students from low-income and other backgrounds as well. For each principle, the authors name several strategies, with examples, that the colleges and high schools they studied use to advance equitable access to and success in high-quality dual enrollment courses. When leaders apply these principles and champion practices and policies that prioritize equity as the ultimate goal, dual enrollment can truly accelerate the path to college and meaningful careers for all students. [For "Highlights from the Dual Enrollment Playbook: A Guide to Equitable Acceleration for Students," see ED608547.]

In 2015, 26 two- and four-year institutions received three-year grants to implement reforms consistent with the Integrated Planning and Advising for Student Success (iPASS) initiative. Grantees committed to launch or enhance existing technologies and undertake related organizational changes that would enable them to provide more effective advising and support to all students. CCRC was engaged as a research partner to document the iPASS colleges' implementation efforts and outcomes. CCRC analyzed key performance indicators (KPIs)--including student retention from first to second semester and first-term grade point average--using administrative data collected from all 26 participating institutions. This report describes trends in KPIs, aggregated by college sector and cohort, in the participating colleges from 2011 to 2017--from the period before the initiative started until about two years after the grants were awarded. Yet because iPASS reforms cannot easily be isolated from other reforms the colleges were carrying out, it is difficult to directly associate the modest changes over time in KPIs that we observe with iPASS efforts. This report also includes college scores on an iPASS development index, a weighted measure we created to gain insights into the level of adoption of iPASS technologies and practices across the 26 grantee institutions as of fall 2017, when colleges provided answers to an institutional survey about their efforts. We find that while no institution had fully implemented iPASS, at most institutions, substantial progress was being made. [For an earlier report, "Baseline Trends in Key Performance Indicators among Colleges Participating in a Technology-Mediated Advising Reform Initiative," see ED586484.]

As NIET enters its 20th year, research continues to point to the effectiveness of NIET's support and the positive impact of investing in human capital. The Winter 2020 Research Summary captures both NIET's key long-term impact, shares promising results highlighting that the TAP System [The System for Teacher and Student Advancement] sustains effectiveness, and spotlights specific districts' successes. Among the highlights is DeSoto Parish Schools in Louisiana. Since partnering with NIET to implement the TAP System in 2008, DeSoto's decade-long commitment to excellent teaching has improved student performance, and the work has deepened even across leadership transitions.

Increasing numbers of high schools are offering senior-year transition curricula in math and English to better prepare graduating students for college. These transition curricula are typically full-year, high school credit-bearing courses taken by students at risk of being placed into developmental (also known as remedial) courses upon enrollment in college. In many cases, students who successfully complete a transition course are officially designated as "college ready" and therefore bypass developmental education altogether. Educators who are interested in developing transition courses currently have limited information about how others have done so. Based on the experiences of those who have developed or overseen transition courses, this brief provides a typology of transition course design options for those thinking about how to best plan and implement these offerings. Among other topics discussed, the brief considers alternative course goals, curricular and instructional approaches, which kinds of students are targeted for course enrollment, mechanisms for determining college readiness, and teacher supports. The information shared here is based on research carried out in the 2017-18 academic year by the Community College Research Center. CCRC researchers conducted interviews with key stakeholders--generally high school and college personnel who have helped to develop transition courses or others who are knowledgeable about them--from 12 randomly selected states, chosen from among states previously identified as having implemented transition curricula statewide or in local settings (see Fay, Barnett, & Chavarín, 2017). The authors also sought out resources available online and used relevant literature to create this typology.

Colleges throughout the United States are evaluating the effectiveness of the strategies used to decide whether to place students into college-level or developmental education courses. Developmental, or remedial, courses are designed to develop the reading, writing, or math skills of students deemed underprepared for college-level courses, a determination usually made through standardized placement tests. However, increasing numbers of colleges are using multiple measures to place students, including additional types of placement tests, high school transcripts, and evaluations of student motivation. The current study was developed to add to the understanding about the implementation, cost, and efficacy of an multiple measures assessment (MMA) system using locally determined rules. As part of a randomized controlled trial, the study team evaluated MMA programs and interviewed and observed staff at five colleges in Minnesota and Wisconsin; it also wrote a short case study about one Wisconsin college. The five colleges in the random assignment study targeted all students taking placement tests in the months before the fall 2018 semester. In the four colleges included in the current analysis, 5,282 students participated in the study; of these, 3,677 were tested for English, and 4,487 were tested for math. The findings suggest that while implementation (especially automation) was not easy, it was possible; and using the new MMA systems became much easier once they were established.

To inform the "Leandro v. State of North Carolina" Action Plan, the Learning Policy Institute examined the within-school and out-of-school barriers to obtaining a sound basic education facing students who attend North Carolina's high-poverty schools. High-poverty schools are schools where 75% or more students are at risk because their families' low incomes qualify them for federally subsidized free or reduced-cost meal programs at school. Because of the concentration of at-risk students at high-poverty schools, these schools deserve focused attention as the state seeks to remedy its failure to provide all students with a constitutionally required opportunity for a sound basic education. The primary goal of this study is to understand the experiences of students attending high-poverty schools. This report provides dozens of findings about the challenges to obtaining a sound basic education experienced by students attending high-poverty schools in North Carolina. Several overarching themes emerge from these findings that form a coherent narrative into which the specific findings and recommendations should be placed and interpreted. [For the research brief, "Adequate and Equitable Education in High-Poverty Schools: Barriers and Opportunities in North Carolina," see ED614422.]

In "Leandro v. the State of North Carolina" ("Leandro"), the Supreme Court of North Carolina found that children have a right to qualified teachers and principals who can prepare students for college and careers and meet the needs of those placed at risk. Providing such high-quality educators for each child demands an adequate supply that is equitably distributed, along with supports for ongoing professional learning that enables educators to meet children's needs. This study documents the current status of educator supply, demand, and quality in North Carolina for teachers and school leaders. It also examines current and past policies that influence teacher and leader development and supply in the state and makes recommendations for how to ensure that all children have access to well-prepared educators. In response to the "Leandro" tenets, this study focuses on: (1) well-placed teachers; and (2) well-prepared leaders. The research methods for this study included analyses of existing administrative data sets, collection of perceptual data from teachers and principals through interviews and focus groups, and administration of a survey to a sample of all North Carolina principals. [For the research brief, see ED617603.]

We present a deformable prototypical part network (Deformable ProtoPNet), an interpretable image classifier that integrates the power of deep learning and the interpretability of case-based reasoning. This model classifies input images by comparing them with prototypes learned during training, yielding explanations in the form of "this looks like that." However, while previous methods use spatially rigid prototypes, we address this shortcoming by proposing spatially flexible prototypes. Each prototype is made up of several prototypical parts that adaptively change their relative spatial positions depending on the input image. Consequently, a Deformable ProtoPNet can explicitly capture pose variations and context, improving both model accuracy and the richness of explanations provided. Compared to other case-based interpretable models using prototypes, our approach achieves state-of-the-art accuracy and gives an explanation with greater context. The code is available at https://github.com/jdonnelly36/Deformable-ProtoPNet., Comment: This was published in CVPR 2022

Impurities immersed into a surrounding ultra-cold Bose gas experience interactions mediated by the surrounding many-body environment. If one focuses on two impurities that are sufficiently close to each other, they can form a bipolaron pair. Here, we discuss how the standard methods based on linearizing the condensate field lead to results only valid in the weak coupling regime and for sufficiently large impurity separations. We show how those shortcomings can be remedied within the Born-Oppenheimer approximation by accounting for boson-boson interactions already on the mean-field level.

A study of initial-state gluon radiation (ISR) in hadron collisions is presented using Drell-Yan (DY) events produced in proton-antiproton collisions by the Tevatron collider at a center-of-mass energy of 1.96 TeV. This paper adopts a novel approach which uses the mean value of the Z/$\gamma^*$ transverse momentum $$ in DY events as a powerful observable to characterize the effect of ISR. In a data sample corresponding to an integrated luminosity of 9.4 fb$^{-1}$ collected with the CDF Run II detector, $$ is measured as a function of the Z/$\gamma^*$ invariant mass. It is found that these two observables have a dependence, $ = -8 + 2.2 \ln m_{DY}^2$ [GeV/c], where $m_{DY}$ is the value of the Z/$\gamma^*$ mass measured in units of GeV/$c^2$. This linear dependence is observed for the first time in this analysis. It may be exploited to model the effect of ISR and constrain its impact in other processes., Comment: 14 pages, 14 figures

The efficiency of Hamiltonian Monte Carlo (HMC) can suffer when sampling a distribution with a wide range of length scales, because the small step sizes needed for stability in high-curvature regions are inefficient elsewhere. To address this we present a delayed rejection variant: if an initial HMC trajectory is rejected, we make one or more subsequent proposals each using a step size geometrically smaller than the last. We extend the standard delayed rejection framework by allowing the probability of a retry to depend on the probability of accepting the previous proposal. We test the scheme in several sampling tasks, including multiscale model distributions such as Neal's funnel, and statistical applications. Delayed rejection enables up to five-fold performance gains over optimally-tuned HMC, as measured by effective sample size per gradient evaluation. Even for simpler distributions, delayed rejection provides increased robustness to step size misspecification. Along the way, we provide an accessible but rigorous review of detailed balance for HMC., Comment: 30 pages, 10 figures

Well-conditioned boundary integral methods for the solution of elliptic boundary value problems (BVPs) are powerful tools for static and dynamic physical simulations. When there are many close-to-touching boundaries (eg, in complex fluids) or when the solution is needed in the bulk, nearly-singular integrals must be evaluated at many targets. We show that precomputing a linear map from surface density to an effective source representation renders this task highly efficient, in the common case where each object is "simple", ie, its smooth boundary needs only moderately many nodes. We present a kernel-independent method needing only an upsampled smooth surface quadrature, and one dense factorization, for each distinct shape. No (near-)singular quadrature rules are needed. The resulting effective sources are drop-in compatible with fast algorithms, with no local corrections nor bookkeeping. Our extensive numerical tests include 2D FMM-based Helmholtz and Stokes BVPs with up to 1000 objects (281000 unknowns), and a 3D Laplace BVP with 10 ellipsoids separated by $1/30$ of a diameter. We include a rigorous analysis for analytic data in 2D and 3D., Comment: 41 pages, 12 figures; submitted to Adv. Comput. Math. (topical collection on integral equations)

The initial-final mass relation (IFMR) represents the total mass lost by a star during the entirety of its evolution from the zero age main sequence to the white dwarf cooling track. The semi-empirical IFMR is largely based on observations of DA white dwarfs, the most common spectral type of white dwarf and the simplest atmosphere to model. We present a first derivation of the semi-empirical IFMR for hydrogen deficient white dwarfs (non-DA) in open star clusters. We identify a possible discrepancy between the DA and non-DA IFMRs, with non-DA white dwarfs $\approx 0.07 M_\odot$ less massive at a given initial mass. Such a discrepancy is unexpected based on theoretical models of non-DA formation and observations of field white dwarf mass distributions. If real, the discrepancy is likely due to enhanced mass loss during the final thermal pulse and renewed post-AGB evolution of the star. However, we are dubious that the mass discrepancy is physical and instead is due to the small sample size, to systematic issues in model atmospheres of non-DAs, and to the uncertain evolutionary history of Procyon B (spectral type DQZ). A significantly larger sample size is needed to test these assertions. In addition, we also present Monte Carlo models of the correlated errors for DA and non-DA white dwarfs in the initial-final mass plane. We find the uncertainties in initial-final mass determinations for individual white dwarfs can be significantly asymmetric, but the recovered functional form of the IFMR is grossly unaffected by the correlated errors., Comment: 26 pages, 13 figures, accepted for publication in the Astronomical Journal

When we deploy machine learning models in high-stakes medical settings, we must ensure these models make accurate predictions that are consistent with known medical science. Inherently interpretable networks address this need by explaining the rationale behind each decision while maintaining equal or higher accuracy compared to black-box models. In this work, we present a novel interpretable neural network algorithm that uses case-based reasoning for mammography. Designed to aid a radiologist in their decisions, our network presents both a prediction of malignancy and an explanation of that prediction using known medical features. In order to yield helpful explanations, the network is designed to mimic the reasoning processes of a radiologist: our network first detects the clinically relevant semantic features of each image by comparing each new image with a learned set of prototypical image parts from the training images, then uses those clinical features to predict malignancy. Compared to other methods, our model detects clinical features (mass margins) with equal or higher accuracy, provides a more detailed explanation of its prediction, and is better able to differentiate the classification-relevant parts of the image., Comment: 10 pages, 6 figures, accepted for oral presentation at the IJCAI-21 Workshop on Deep Learning, Case-Based Reasoning, and AutoML: Present and Future Synergies. arXiv admin note: substantial text overlap with arXiv:2103.12308

At the Fermilab Tevatron proton-antiproton ($p\bar{p}$) collider, high-mass electron-neutrino ($e\nu$) pairs are produced predominantly in the process $p \bar{p} \rightarrow W(\rightarrow e\nu) + X$. The asymmetry of the electron and positron yield as a function of their pseudorapidity constrain the slope of the ratio of the $u$- to $d$-quark parton distributions versus the fraction of the proton momentum carried by the quarks. This paper reports on the measurement of the electron-charge asymmetry using the full data set recorded by the Collider Detector at Fermilab in 2001--2011 and corresponding to 9.1~fb$^{-1}$ of integrated luminosity. The measurement significantly improves the precision of the Tevatron constraints on the parton-distribution functions of the proton. Numerical tables of the measurement are provided., Comment: 27 pages, 25 figures. To be published in PRD