Original Iteration Research Articles

Upgrading <em>The L Word: Generation Q</em>

Upgrading <em>The L Word: Generation Q</em>

A sparse iteration space transformation framework for sparse tensor algebra

We address the problem of optimizing sparse tensor algebra in a compiler and show how to define standard loop transformations---split, collapse, and reorder---on sparse iteration spaces. The key idea is to track the transformation functions that map the original iteration space to derived iteration spaces. These functions are needed by the code generator to emit code that maps coordinates between iteration spaces at runtime, since the coordinates in the sparse data structures remain in the original iteration space. We further demonstrate that derived iteration spaces can tile both the universe of coordinates and the subset of nonzero coordinates: the former is analogous to tiling dense iteration spaces, while the latter tiles sparse iteration spaces into statically load-balanced blocks of nonzeros. Tiling the space of nonzeros lets the generated code efficiently exploit heterogeneous compute resources such as threads, vector units, and GPUs. We implement these concepts by extending the sparse iteration theory implementation in the TACO system. The associated scheduling API can be used by performance engineers or it can be the target of an automatic scheduling system. We outline one heuristic autoscheduling system, but other systems are possible. Using the scheduling API, we show how to optimize mixed sparse-dense tensor algebra expressions on CPUs and GPUs. Our results show that the sparse transformations are sufficient to generate code with competitive performance to hand-optimized implementations from the literature, while generalizing to all of the tensor algebra.

Observations on the Correlation of Basic Science Content in Year 1 With the Clinical Science Content in Year 2 Within a Modified System Based Integrated Spiraled Medical Curriculum

In August 2016, Instruction began at the Burrell College of Osteopathic Medicine. Curriculum planning commenced in 2015 and has been constant in preparation of teach subsequent year. The class that began as the inaugural class (class of 2020) will graduate in May of 2020. Now, the first year curriculum is being presented in its fourth iteration and the second year curriculum is presented in its third iteration. Input from students, faculty and administration has been incorporated as the curriculum has evolved into its present form. This presentation will highlight the observations and inputs that have contributed to the growth of the curriculum from planning to present form. The original iteration of the curriculum was modeled after the two pass integrated curriculum that was developed at Texas College of Osteopathic Medicine and employed at Rocky Vista College of Osteopathic Medicine. Faculty designed the basic order of systems for presentation and the weekly design of the sessions in each system over the fall and spring of 2015–2016. The number of faculty involved continually grew as the faculty were hired during this period. The basic science faculty designed the systems courses in the Year 1 curriculum. Clinical faculty designed the Clinical Skills and Osteopathic Manipulative Medicine components of both years 1 and 2. The clinical faculty also designed the systems courses of Year 2 of the curriculum. As a result, the first year of the curriculum is based in the general science that serves as a basis for understanding medicine while the second year of the curriculum has a distinctly clinical focus related to diagnosis and basic treatment of diseases. Connection of the content between year 1 and year 2 is the area of biggest concern related to the curriculum. Planning and execution do not always align and leads to many gaps that are seen in the knowledge base of the students. Communication between the faculties teaching each year is vital but is often less than optimal. Bloating in each year of the curriculum is the result, but often exacerbates problems as opposed to resolving issues. Administration and faculty opinions often differ as to the cause of the problem, as well as, how to correct the situation. Administrative organization, curriculum order, length of semester, and faculty turnover have all be looked at to address various issues. We will address what has been done to facilitate improvements in the program as the first four years have progressed.

Existence and Uniqueness of Positive Solutions to Nonlinear Systems of Equations With M-Type Functions

This paper considers the existence and uniqueness of positive solutions to a class of nonlinear systems of equations, which has wide applications in fluid mechanics and electrical circuit theory. In this paper, we define a new class of functions, called M-type function, and then, we show some good properties of M-type function as well as the relationship between this kind of functions and other kinds of functions. In particular, by fully exploiting the properties of M-type function, we present that under a reasonable condition, the nonlinear system of equations with an M-type function, which can be seen as a generalization of the multilinear system with a nonsingular M-tensor, and a positive right-hand side vector has a unique positive solution. Meanwhile, for this class of problems, we give a preliminary and simple algorithm framework. What is more, under an additional assumption, we illustrate the feasibility of using Newton step when improving the original iteration algorithm to obtain better convergence. Numerical experiments show the stability and efficiency of the proposed algorithms.

Improvements in DDA program for rockslides with local in-circle contact method and modified open-close iteration

Discontinuous deformation analysis (DDA) is capable of modeling rockslides with substantial debris flow. This study improves DDA modeling accuracy by introducing two important enhancements. Firstly, a local in-circle method is proposed to obtain round corners, which eliminates contact indeterminacy and the associated unreasonable sliding behaviors. The radii of the round corners prove having influence on motion modes and sliding distances, offering another perspective for back analysis. Secondly, the original open-close iteration (OCI) is modified to enforce stricter convergence criteria of contact analysis. Accurate contact states and friction forces are obtained using the modified OCI, which is conducive to more precise analysis of rockslides. Several numerical tests were conducted in validating the necessity and accuracy of these two improvements. The results also imply that rock blocks with rounder corners are more likely to cross bench slopes and exhibit longer sliding distances. By overcoming two critical issues in DDA, this work could contribute to rockslide analysis in rock engineering.

Efficient large-scale geometric verification for structure from motion

Geometric verification is a fundamental problem in epipolar geometry, which estimates fundamental matrix and homography matrix to confirm that image pairs share a common 3D structure. However, it still suffers from low efficiency when it encounters large-scale Structure from Motion (SfM). In this paper, we adopt the linear congruence algorithm to sample point-pairs in parallel. Then, we propose to simultaneously estimate a certain number of candidate fundamental/homography matrices in GPU to avoid the iterative random point sets sampling and perform matrix estimation based on these point sets, followed by best matrix selection and further parallel refinement. Experiments on extensive datasets show that our GPU-based geometric verification is up to seventy times faster than original iteration method while maintaining comparable satisfactory 3D reconstruction results.

<em>The Simpsons</em> Do the Nineties

<em>The Simpsons</em> Do the Nineties

Iterative phase retrieval by combining modulus constraints and angle relationships

In coherent diffraction imaging, modulus is recorded and phase is lost. Phase retrieval is to recover the lost phase from the recorded modulus. Iterative Fourier algorithm (IFA) solves this problem by imposing modulus constraints to make the phase converge. Here we show a modulus angle joint phase retrieval method. When both amplitude and phase modulation are applied, the angle relationship, which denotes the Fourier phase difference, can be exactly derived from the measured intensities. Based on this angle relationship, a new form of IFA called angle iteration is developed, which exhibits a faster but unstable convergence compared with tradition IFA (modulus iteration). By modulus-angle combined iteration, the proposed method reaches a quick and stable convergence. With the same intensity measurements, it drops nearly 80% the phase retrieval errors by traditional IFA but costs the same CPU time. This work reveals the fact that changing the constraint format would improve the original iteration process.

Automatic voltage regulator design using a modified adaptive optimal approach

In this paper, an online adaptive optimal controller is firstly designed to optimize the performance of an automatic voltage regulator (AVR). Towards this end, an optimal quadratic tracking problem is defined based on the error between the synchronous generator’s terminal voltage and its desired value. Then, this optimal control problem is solved using an adaptive dynamic programming (ADP) method, called the policy iteration technique. Using this technique, the optimal performance is achieved without knowledge of the AVR parameters. To eliminate the steady state error between the AVR output and its desired value, a modification is made in the original policy iteration technique base on the integral action control and another adaptive optimal controller is designed. In addition, an approach is proposed to represent a large-scale power system through several Single-Machine Infinite-Bus (SMIB) systems. The equivalent SMIB system then may be linearized through modal analysis to be employed by the proposed control scheme. Simulation results show that the designed adaptive optimal controllers are so effective and can be used in practical power systems.

Tradeoffs Between Convergence Speed and Reconstruction Accuracy in Inverse Problems

Solving inverse problems with iterative algorithms is popular, especially for large data. Due to time constraints, the number of possible iterations is usually limited, potentially affecting the achievable accuracy. Given an error one is willing to tolerate, an important question is whether it is possible to modify the original iterations to obtain faster convergence to a minimizer achieving the allowed error without increasing the computational cost of each iteration considerably. Relying on recent recovery techniques developed for settings in which the desired signal belongs to some low-dimensional set, we show that using a coarse estimate of this set may lead to faster convergence at the cost of an additional reconstruction error related to the accuracy of the set approximation. Our theory ties to recent advances in sparse recovery, compressed sensing, and deep learning. Particularly, it may provide a possible explanation to the successful approximation of the $\ell _1$ -minimization solution by neural networks with layers representing iterations, as practiced in the learned iterative shrinkage-thresholding algorithm.

A modified Newton iteration for finding nonnegative Z-eigenpairs of a nonnegative tensor

We propose a modified Newton iteration for finding some nonnegative Z-eigenpairs of a nonnegative tensor. When the tensor is irreducible, all nonnegative eigenpairs are known to be positive. We prove local quadratic convergence of the new iteration to any positive eigenpair of a nonnegative tensor, under the usual assumption guaranteeing the local quadratic convergence of the original Newton iteration. A big advantage of the modified Newton iteration is that it seems capable of finding a nonnegative eigenpair starting with any positive unit vector. Special attention is paid to transition probability tensors.

Analytical Approaches to Improve Accuracy in Solving the Protein Topology Problem.

To take advantage of recent advances in genomics and proteomics it is critical that the three-dimensional physical structure of biological macromolecules be determined. Cryo-Electron Microscopy (cryo-EM) is a promising and improving method for obtaining this data, however resolution is often not sufficient to directly determine the atomic scale structure. Despite this, information for secondary structure locations is detectable. De novo modeling is a computational approach to modeling these macromolecular structures based on cryo-EM derived data. During de novo modeling a mapping between detected secondary structures and the underlying amino acid sequence must be identified. DP-TOSS (Dynamic Programming for determining the Topology Of Secondary Structures) is one tool that attempts to automate the creation of this mapping. By treating the correspondence between the detected structures and the structures predicted from sequence data as a constraint graph problem DP-TOSS achieved good accuracy in its original iteration. In this paper, we propose modifications to the scoring methodology of DP-TOSS to improve its accuracy. Three scoring schemes were applied to DP-TOSS and tested: (i) a skeleton-based scoring function; (ii) a geometry-based analytical function; and (iii) a multi-well potential energy-based function. A test of 25 proteins shows that a combination of these schemes can improve the performance of DP-TOSS to solve the topology determination problem for macromolecule proteins.

An evolutionary scheduling approach for trading-off accuracy vs. verifiable energy in multicore processors

Abstract This work addresses the problem of energy-efficient scheduling and allocation of tasks in multicore environments, where the tasks can allow a certain loss in accuracy in the output, while still providing proper functionality and meeting an energy budget. This margin for accuracy loss is exploited by using computing techniques that reduce the work load, and thus can also result in significant energy savings. To this end, we use the technique of loop perforation, that transforms loops to execute only a subset of their original iterations, and integrate this technique into our existing optimization tool for energy-efficient scheduling. To verify that a schedule meets an energy budget, both safe upper and lower bounds on the energy consumption of the tasks involved are needed. For this reason, we use a parametric approach to estimate safe (and tight) energy bounds that are practical for energy verification (and optimization applications). This approach consists in dividing a program into basic (‘branchless’) blocks, establishing the maximal (resp. minimal) energy consumption for each block using an evolutionary algorithm, and combining the obtained values according to the program control flow, by using static analysis to produce energy bound functions on input data sizes. The scheduling tool uses evolutionary algorithms coupled with the energy bound functions for estimating the energy consumption of different schedules. The experiments with our prototype implementation were performed on multicore XMOS chips, but our approach can be adapted to any multicore environment with minor changes. The experimental results show that our new scheduler enhanced with loop perforation improves on the previous one, achieving significant energy savings (31% on average for the test programs) for acceptable levels of accuracy loss.

The health information national trends survey (HINTS): A resource for consumer engagement and health communication research

The contemporary healthcare system can help improve health literacy outcomes in two ways: first, by nurturing the skills and motivations needed for patients to be actively engaged in their own health and healthcare decisions; and, second, by creating a prepared and proactive healthcare system that adapts to patients' capacities and needs in efficacious ways. In 2001, the National Cancer Institute launched the Health Information National Trends Survey (HINTS) as a way for researchers and planners to understand how the public is interacting with a rapidly changing health information environment. Original iterations of the HINTS national probability sampling strategies took place on a biennial basis, but in subsequent years the protocol moved to a yearly administration. This yields a rich resource of cross-sectional, national surveillance data to evaluate for trends across and within vulnerable populations. Sixteen studies are presented from the published literature to illustrate how HINTS data were used to explore constructs of direct interest to health literacy researchers. Suggestions are given for how this ongoing public surveillance mechanism can be used: (a) to provide a sentinel view of how the public is interacting with information in the environment to address their health needs; (b) to generate research questions and hypotheses for further exploration using complementary methodologies; and

High-precision analytic solution of the problem on bending vibrations of a clamped square plate

An original iteration algorithm is used to construct new analytic expressions for computing approximate natural frequencies and shape modes of bending vibrations of a square homogeneous plate clamped along its contour. The errors are estimated by comparing with the results of well-known numerical high-precision computations. The results of analytic computations are also compared with experimental data obtained by the author by the resonance method. The proposed research technique and the obtained high-precision expressions for the natural shape modes can be used in the case of rectangular plates and for other types of boundary conditions. A numerical-analytical method is used to show that the small isoperimetric theorem holds.

Breast Cancer in the Age of Personalized Medicine: A CURE Design for Aspiring Physicians

Modern medicine is greatly impacted by the use of emergent technologies to diagnose and treat disease processes at the molecular level ‐from rapid sequencing of human genomes to remarkable progress from genome‐wide association studies. The potential for genetics and genomics to provide personalized approaches for diagnosing and treating human disease in the clinical setting is imminent. How can we prepare undergraduate students to fully engage these issues at the forefront of molecular medicine? We designed a course‐based undergraduate research experience (CURE) for an upper‐division molecular biology course illustrating the use of personalized medicine in the problem of breast cancer. Students experimentally and computationally analyzed the transcriptome of biopsied human breast cancer tissue using DNA microarrays, and determined a molecular diagnosis and prognosis for the patient. Synthesizing clinical pathological data with gene expression data, students reproducibly sub‐typed breast tumors as ER+/− and/or HER2+/− and BRCA1‐2 +/−, and determined a prognostic “fingerprint” for the patient. The experience culminated in a mock molecular tumor board event, wherein students role‐played as physicians to create a recommendation for personalized chemotherapy. In addition to measuring cognitive outcomes, we report the extent to which this CURE impacted non‐cognitive student outcomes.The module's efficacy was evaluated in terms of five design elements ‐ the degree to which students (1) made novel scientific discoveries, (2) iterated and revised work, (3) collaborated with peers, (4) owned the project, and (5) gained confidence as a scientist. The Laboratory Course Assessment Survey (LCAS) and Project Ownership Survey (POS) indicated significant gains in aggregated measures of collaboration (T‐test, p<0.001) and project ownership (T‐test, p=0.004). In addition, pre‐post analysis of measures of science self‐efficacy and attitude indicated a favorable shift as a result of the course (CLASS‐BIO, T‐test, p=0.01), with significant gains in student self‐reported intention to pursue a science‐related research career (T‐test, p=0.01). Interestingly, these affective gains are not associated with the degree to which students make novel scientific contributions (T‐test, p=.28) or iterate the investigative process (T‐test, p=.53 for aggregated measures). Content analysis of open‐ended student comments corroborated survey results, and emphasized appreciation for the applicability of the course to societal issues and decision‐making. Together these data suggest that CURE design elements that include highly relevant proof‐of‐concept inquiry can be as effective as CURES centered on original research and iteration, signifying an alternative pathway toward increased science identity, career clarification, and persistence in science.

Observation of potentially troublesome (2) JCC correlations in 1,1-ADEQUATE spectra.

Despite the tremendous usage of HMBC to establish long-range (1) H-(13) C and (1) H-(15) N heteronuclear correlations, an inherent drawback of the experiment is the indeterminate nature of the (n) JXH correlations afforded by the experiment. A priori there is no reliable way of determining whether a given (n) JCH correlation is, for example, via two-, three-, or sometimes even four-bonds. This limitation of the HMBC experiment spurred the development of the ADEQUATE family of NMR experiments that rely on, in the case of 1,1-ADEQUATE, an out-and-back transfer of magnetization via the (1) JCC homonuclear coupling constant, which is significantly larger than (n) JCC (where n = 2-4) couplings in most cases. Hence, the 1,1-ADEQUATE experiment has generally been assumed to unequivocally provide the equivalent of (2) JCH correlations. The recent development of the 1,1- and 1,n-HD-ADEQUATE experiments that can provide homodecoupling for certain (1) JCC and (n) JCC correlations has increased the sensitivity of the ADEQUATE experiments significantly and can allow acquisition of these data in a fraction of the time required for the original iterations of this pulse sequence. With these gains in sensitivity, however, there occasionally come unanticipated consequences. We have observed that the collapse of proton multiplets, in addition to providing better s/n for the desired (1) JCC correlations can facilitate the observation of typically weaker (2) JCC correlations across intervening carbonyl resonances in 1,1-HD-ADEQUATE spectra. Several examples are shown, with the results supported by the measurement of the (2) JCC coupling constants in question using J-modulated-HD-ADEQUATE and DFT calculations. Copyright © 2016 John Wiley & Sons, Ltd.

Kai hea kai hea te pū o te mate? Reclaiming the power of pūrākau

The pūrākau our ancestors told about the universe and our place within it have been bowdlerised through the process of colonisation. These narratives, as they were transmitted over generations, were transformed by the European settlers, missionaries and educators, from ‘myths’ – oral traditions imbued with the power of the sacred – into ‘fables’ and ‘folktales’. As such, they have largely been neutered of their epistemological power, and their role in sustaining our culture has been substantially diminished. For example, in its original iterations, the ancient story of Māui in which his quest for immortality was foiled by the Tīrairaka contained a fundamental lesson: Māui dies in the act of penetrating Hine-nui-te-pō; from the sex act comes both new life and the sure knowledge of mortality; women are a source of power, life and death. In translation, this story was sanitised; in particular, the description of Māui’s fatal entrance as a lizard into Hine-nui-te-pō’s vagina, were euphemised and displaced, shifted to less controversial body parts. In my paper and presentation, I propose strategies for restoring the power of performance of our pūrākau through the reclamation of the act of storytelling in diverse media. In making new platforms for performing these old stories, we can revitalise the Reo and tikanga, and in so doing reconnect ourselves and our young people to the world that our tīpuna created.

Convergence Proof of Approximate Policy Iteration for Undiscounted Optimal Control of Discrete-Time Systems

Approximate policy iteration (API) is studied to solve undiscounted optimal control problems in this paper. A discrete-time system with the continuous-state space and the finite-action set is considered. As approximation technique is used for the continuous-state space, approximation errors exist in the calculation and disturb the convergence of the original policy iteration. In our research, we analyze and prove the convergence of API for undiscounted optimal control. We use an iterative method to implement approximate policy evaluation and demonstrate that the error between approximate and exact value functions is bounded. Then, with the finite-action set, the greedy policy in policy improvement is generated directly. Our main theorem proves that if a sufficiently accurate approximator is used, API converges to the optimal policy. For implementation, we introduce a fuzzy approximator and verify the performance on the puddle world problem.

A constraint and algorithm for stress-based evolutionary structural optimization of the tie-beam problem

Purpose – The purpose of this paper is to present a constraint and corresponding algorithm enhancing the evolutionary structural optimization (ESO) method, aiming to circumvent its structure break down problem in some special cases, such as the tie-beam problem. Design/methodology/approach – A virtual soft material introduced to an element will change the stiffness of the element and may consequently change the stress distribution of that element and its neighbors. With this property, the virtual stiffness of the selected element is calculated and the threshold of the stress changes is derived. The stress threshold is used to evaluate the role of an element on the load path and therefore decide the contribution of the element to the structure. Adding this checking operation into the original ESO iterations enables validation of element removal. Findings – The reason for structure break down with the ESO method is that the element removal criterion of ESO only works for certain optimal objectives. It cannot guarantee that the structure does not fail. The proposed operation offers a stronger and stricter constraint condition for ESO’s element removal process, preventing the structure from breaking down in some special cases. Originality/value – The tests on several examples reported in the literature show that the proposed method has the same ability to achieve an optimum solution as the original ESO methods do, while avoiding incorrect deletion of structurally important elements. The benchmark tie-beam problem is solved successfully with this algorithm. The method can be used in other situations as well.