Joint radar receive filter and waveform design is non-convex, but is individually convex for a fixed receiver filter while optimizing the waveform, and vice versa. Such classes of problems are fre- quently encountered in optimization, and are referred to biconvex programs. Alternating minimization (AM) is perhaps the most popu- lar, effective, and simplest algorithm that can deal with bi-convexity. In this paper we consider new perspectives on this problem via older, well established problems in the optimization literature. It is shown here specifically that the radar waveform optimization may be cast as constrained least squares, semi-definite programs (SDP), and quadratically constrained quadratic programs (QCQP). The bi-convex constraint introduces sets which vary for each iteration in the alternat- ing minimization. We prove convergence of alternating minimization for biconvex problems with biconvex constraints by showing the equivalence of this to a biconvex problem with constrained Cartesian product convex sets but for convex hulls of small diameter.
We investigate an alternative solution method to the joint signal-beamformer optimization problem considered in a recent publication devoted to this topic. First, we directly demonstrate that the problem, which minimizes the received noise, interference, and clutter power under a minimum variance distortionless response constraint, is generally nonconvex and we provide insight into the nature of the nonconvexity. Second, we employ the theory of biquadratic optimization and semidefinite relaxations to produce a relaxed version of the problem, which we show to be convex . The optimality conditions of this relaxed problem are examined and a variety of potential solutions are found, both analytically and numerically. These solutions are then compared to existing alternating minimization schemes.
Efforts to taxonomically delineate species are often confounded with conflicting information and subjective interpretation. Advances in genomic methods have resulted in a new approach to taxonomic identification that stands to greatly reduce much of this conflict. This approach is ideal for species complexes, where divergence times are recent (evolutionarily) and lineages less well defined. The California Roach/Hitch fish species complex is an excellent example, experiencing a convoluted geologic history, diverse habitats, conflicting species designations and potential admixture between species. Here we use this fish complex to illustrate how genomics can be used to better clarify and assign taxonomic categories. We performed restriction-site associated DNA (RAD) sequencing on 255 Roach and Hitch samples collected throughout California to discover and genotype thousands of single nucleotide polymorphism (SNPs). Data were then used in hierarchical principal component, admixture, and FST analyses to provide results that consistently resolved a number of ambiguities and provided novel insights across a range of taxonomic levels. At the highest level, our results show that the CA Roach/Hitch complex should be considered five species split into two genera (4 + 1) as opposed to two species from distinct genera (1 +1). Subsequent levels revealed multiple subspecies and distinct population segments within identified species. At the lowest level, our results indicate Roach from a large coastal river are not native but instead introduced from a nearby river. Overall, this study provides a clear demonstration of the power of genomic methods for informing taxonomy and serves as a model for future studies wishing to decipher difficult species questions. By allowing for systematic identification across multiple scales, taxonomic structure can then be tied to historical and contemporary ecological, geographic or anthropogenic factors.
Abstract Extended laboratory culture and antimicrobial susceptibility testing timelines hinder rapid species identification and susceptibility profiling of bacterial pathogens associated with bovine respiratory disease, the most prevalent cause of cattle mortality in the United States. Whole-genome sequencing offers a culture-independent alternative to current bacterial identification methods, but requires a library of bacterial reference genomes for comparison. To contribute new bacterial genome assemblies and evaluate genetic diversity and variation in antimicrobial resistance genotypes, whole-genome sequencing was performed on bovine respiratory disease–associated bacterial isolates (Histophilus somni, Mycoplasma bovis, Mannheimia haemolytica, and Pasteurella multocida) from dairy and beef cattle. One hundred genomically distinct assemblies were added to the NCBI database, doubling the available genomic sequences for these four species. Computer-based methods identified 11 predicted antimicrobial resistance genes in three species, with none being detected in M. bovis. While computer-based analysis can identify antibiotic resistance genes within whole-genome sequences (genotype), it may not predict the actual antimicrobial resistance observed in a living organism (phenotype). Antimicrobial susceptibility testing on 64 H. somni, M. haemolytica, and P. multocida isolates had an overall concordance rate between genotype and phenotypic resistance to the associated class of antimicrobials of 72.7% (P < 0.001), showing substantial discordance. Concordance rates varied greatly among different antimicrobial, antibiotic resistance gene, and bacterial species combinations. This suggests that antimicrobial susceptibility phenotypes are needed to complement genomically predicted antibiotic resistance gene genotypes to better understand how the presence of antibiotic resistance genes within a given bacterial species could potentially impact optimal bovine respiratory disease treatment and morbidity/mortality outcomes.
Author(s): O'Rourke, Sean | Advisor(s): Swindlehurst, A. Lee | Abstract: This dissertation investigates joint design of transmit and receive resources in radar systems, given prior knowledge of a signal-dependent clutter environment, based on simultaneous convex relaxation. Motivated by waveform-adaptive space-time adaptive processing(WA-STAP), we first analyze traditional approaches to joint trans-receive design, which alternate between optimal signal designs for fixed filters and vice versa to maximize signal-to-interference-plus-noise ratio (SINR). First, we demonstrate that SINR a its variants arenon-convex bi-quadratic programs (BQP). Next, we provide perspectives on current alternating methods used to solve BQPs in terms of linear semidefinite (SDP) a quadratically-constrained quadratic (QCQP) programs. We then develop a novel method of performing thisjoint design based on quadratic semidefinite programming (QSDP). Our proposed relaxation scheme first analyzes the power-constrained problem, and is accompanied by exploration of operator structures for signal-dependent clutter. This technique is then extended to morechallenging waveform constraints -- in particular, constant modulus and similarity constrained designs. A refinement scheme is devised to improve recovery of vector solutions from therelaxed matrix solution for the constant modulus problem. Finally, we apply the techniqueto a system model other than WA-STAP; namely, the reverberant sonar/radar channel. Numerical simulations are provided to show the superior performance of our method for SINR, detection, a resolution across multiple environmental scenarios.
Two Saccharomyces cerevisiae plasma membrane-spanning proteins, Sho1 and Sln1, function during increased osmolarity to activate a mitogen-activated protein (MAP) kinase cascade. One of these proteins, Sho1, utilizes the MAP kinase kinase kinase Ste11 to activate Pbs2. We previously used the FUS1 gene of the pheromone response pathway as a reporter to monitor cross talk in hog1 mutants. Cross talk requires the Sho1-Ste11 branch of the HOG pathway, but some residual signaling, which is STE11 dependent, still occurs in the absence of Sho1. These observations led us to propose the existence of another osmosensor upstream of Ste11. To identify such an osmosensor, we screened for mutants in which the residual signaling in a hog1 sho1 mutant was further reduced. We identified the MSB2 gene, which encodes a protein with a single membrane-spanning domain and a large presumptive extracellular domain. Assay of the FUS1-lacZ reporter (in a hog1 mutant background) showed that sho1 and msb2 mutations both reduced the expression of the reporter partially and that the hog1 sho1 msb2 mutant was severely defective in the expression of the reporter. The use of DNA microarrays to monitor gene expression revealed that Sho1 and Msb2 regulate identical gene sets in hog1 mutants. A role for MSB2 in HOG1 strains was also seen in strains defective in the two known branches that activate Pbs2: an ssk1 sho1 msb2 strain was more osmosensitive than an ssk1 sho1 MSB2 strain. These observations indicate that Msb2 is partially redundant with the Sho1 osmosensing branch for the activation of Ste11.
The Saccharomyces cerevisiae high osmolarity glycerol (HOG) mitogen-activated protein kinase pathway is required for osmoadaptation and contains two branches that activate a mitogen-activated protein kinase (Hog1) via a mitogen-activated protein kinase kinase (Pbs2). We have characterized the roles of common pathway components (Hog1 and Pbs2) and components in the two upstream branches (Ste11, Sho1, and Ssk1) in response to elevated osmolarity by using whole-genome expression profiling. Several new features of the HOG pathway were revealed. First, Hog1 functions during gene induction and repression, cross talk inhibition, and in governing the regulatory period. Second, the phenotypes of pbs2 and hog1 mutants are identical, indicating that the sole role of Pbs2 is to activate Hog1. Third, the existence of genes whose induction is dependent on Hog1 and Pbs2 but not on Ste11 and Ssk1 suggests that there are additional inputs into Pbs2 under our inducing conditions. Fourth, the two upstream pathway branches are not redundant: the Sln1-Ssk1 branch has a much more prominent role than the Sho1-Ste11 branch for activation of Pbs2 by modest osmolarity. Finally, the general stress response pathway and both branches of the HOG pathway all function at high osmolarity. These studies demonstrate that cells respond to increased osmolarity by using different signal transduction machinery under different conditions.
Vital for Space Situational Awareness, Initial Orbit Determination (IOD) may be used to initialize object tracking and associate observations with a tracked satellite. Classical IOD algorithms provide only a point solution and are sensitive to noisy measurements and to certain target-observer geometry. This work examines the ability of a Multivariate GPR (MV-GPR) to accurately perform IOD and quantify the associated uncertainty. Given perfect test inputs, MV-GPR performs comparably to a simpler multitask learning GPR algorithm and the classical Gauss–Gibbs IOD in terms of prediction accuracy. It significantly outperforms the multitask learning GPR algorithm in uncertainty quantification due to the direct handling of output dimension correlations. A moment-matching algorithm provides an analytic solution to the input noise problem under certain assumptions. The algorithm is adapted to the MV-GPR formulation and shown to be an effective tool to accurately quantify the added input uncertainty. This work shows that the MV-GPR can provide a viable solution with quantified uncertainty which is robust to observation noise and traditionally challenging orbit-observer geometries.
This paper introduces the radar text data set (RadarTD) for technical language modeling. This data set is comprised of sentences containing radar parameters, values, and units determined from real-world values. This data set is created based on values determined from published academic research. Additionally, each statement is assigned a sentiment label and goal priority label. Preliminary investigations into the applicability of this data set are explored using the BERT model and several bi-directional LSTM models. These models are evaluated on text classification and named entity recognition tasks. This study evaluates the applicability of technical language modeling using neural networks to analyze input statements for cognitive radar applications. These findings suggest that this data set can be used to achieve reasonable performance for both text classification and named entity recognition for autonomous radar applications.
