Elastic-net prefiltering for two-class classification

A two-stage linear-in-the-parameter model construction algorithm is proposed aimed at noisy two-class classification problems. The purpose of the first stage is to produce a prefiltered signal that is used as the desired output for the second stage which constructs a sparse linear-in-the-parameter classifier. The prefiltering stage is a two-level process aimed at maximizing a model's generalization capability, in which a new elastic-net model identification algorithm using singular value decomposition is employed at the lower level, and then, two regularization parameters are optimized using a particle-swarm-optimization algorithm at the upper level by minimizing the leave-one-out (LOO) misclassification rate. It is shown that the LOO misclassification rate based on the resultant prefiltered signal can be analytically computed without splitting the data set, and the associated computational cost is minimal due to orthogonality. The second stage of sparse classifier construction is based on orthogonal forward regression with the D-optimality algorithm. Extensive simulations of this approach for noisy data sets illustrate the competitiveness of this approach to classification of noisy data problems.

Runge-Kutta IMEX schemes for the Horizontally Explicit/Vertically Implicit (HEVI) solution of wave equations

Many operational weather forecasting centres use semi-implicit time-stepping schemes because of their good efficiency. However, as computers become ever more parallel, horizontally explicit solutions of the equations of atmospheric motion might become an attractive alternative due to the additional inter-processor communication of implicit methods. Implicit and explicit (IMEX) time-stepping schemes have long been combined in models of the atmosphere using semi-implicit, split-explicit or HEVI splitting. However, most studies of the accuracy and stability of IMEX schemes have been limited to the parabolic case of advection–diffusion equations. We demonstrate how a number of Runge–Kutta IMEX schemes can be used to solve hyperbolic wave equations either semi-implicitly or HEVI. A new form of HEVI splitting is proposed, UfPreb, which dramatically improves accuracy and stability of simulations of gravity waves in stratified flow. As a consequence it is found that there are HEVI schemes that do not lose accuracy in comparison to semi-implicit ones. The stability limits of a number of variations of trapezoidal implicit and some Runge–Kutta IMEX schemes are found and the schemes are tested on two vertical slice cases using the compressible Boussinesq equations split into various combinations of implicit and explicit terms. Some of the Runge–Kutta schemes are found to be beneficial over trapezoidal, especially since they damp high frequencies without dropping to first-order accuracy. We test schemes that are not formally accurate for stiff systems but in stiff limits (nearly incompressible) and find that they can perform well. The scheme ARK2(2,3,2) performs the best in the tests.

Sparse model construction using coordinate descent optimization

We propose a new sparse model construction method aimed at maximizing a model’s generalisation capability for a large class of linear-in-the-parameters models. The coordinate descent optimization algorithm is employed with a modified l1- penalized least squares cost function in order to estimate a single parameter and its regularization parameter simultaneously based on the leave one out mean square error (LOOMSE). Our original contribution is to derive a closed form of optimal LOOMSE regularization parameter for a single term model, for which we show that the LOOMSE can be analytically computed without actually splitting the data set leading to a very simple parameter estimation method. We then integrate the new results within the coordinate descent optimization algorithm to update model parameters one at the time for linear-in-the-parameters models. Consequently a fully automated procedure is achieved without resort to any other validation data set for iterative model evaluation. Illustrative examples are included to demonstrate the effectiveness of the new approaches.

The influence of stratospheric vortex displacements and splits on surface climate

A strong link exists between stratospheric variability and anomalous weather patterns at the earth’s surface. Specifically, during extreme variability of the Arctic polar vortex termed a “weak vortex event,” anomalies can descend from the upper stratosphere to the surface on time scales of weeks. Subsequently the outbreak of cold-air events have been noted in high northern latitudes, as well as a quadrupole pattern in surface temperature over the Atlantic and western European sectors, but it is currently not understood why certain events descend to the surface while others do not. This study compares a new classification technique of weak vortex events, based on the distribution of potential vorticity, with that of an existing technique and demonstrates that the subdivision of such events into vortex displacements and vortex splits has important implications for tropospheric weather patterns on weekly to monthly time scales. Using reanalysis data it is found that vortex splitting events are correlated with surface weather and lead to positive temperature anomalies over eastern North America of more than 1.5 K, and negative anomalies over Eurasia of up to −3 K. Associated with this is an increase in high-latitude blocking in both the Atlantic and Pacific sectors and a decrease in European blocking. The corresponding signals are weaker during displacement events, although ultimately they are shown to be related to cold-air outbreaks over North America. Because of the importance of stratosphere–troposphere coupling for seasonal climate predictability, identifying the type of stratospheric variability in order to capture the correct surface response will be necessary.

The residual circulation of the Southern Ocean: which spatio-temporal scales are needed?

The Southern Ocean circulation consists of a complicated mixture of processes and phenomena that arise at different time and spatial scales which need to be parametrized in the state-of-the-art climate models. The temporal and spatial scales that give rise to the present-day residual mean circulation are here investigated by calculating the Meridional Overturning Circulation (MOC) in density coordinates from an eddy-permitting global model. The region sensitive to the temporal decomposition is located between 38°S and 63°S, associated with the eddy-induced transport. The ‘‘Bolus’’ component of the residual circulation corresponds to the eddy-induced transport. It is dominated by timescales between 1 month and 1 year. The temporal behavior of the transient eddies is examined in splitting the ‘‘Bolus’’ component into a ‘‘Seasonal’’, an ‘‘Eddy’’ and an ‘‘Inter-monthly’’ component, respectively representing the correlation between density and velocity fluctuations due to the average seasonal cycle, due to mesoscale eddies and due to large-scale motion on timescales longer than one month that is not due to the seasonal cycle. The ‘‘Seasonal’’ bolus cell is important at all latitudes near the surface. The ‘‘Eddy’’ bolus cell is dominant in the thermocline between 50°S and 35°S and over the whole ocean depth at the latitude of the Drake Passage. The ‘‘Inter-monthly’’ bolus cell is important in all density classes and is maximal in the Brazil–Malvinas Confluence and the Agulhas Return Current. The spatial decomposition indicates that a large part of the Eulerian mean circulation is recovered for spatial scales larger than 11.25°, implying that small-scale meanders in the Antarctic Circumpolar Current (ACC), near the Subantarctic and Polar Fronts, and near the Subtropical Front are important in the compensation of the Eulerian mean flow.

Galactic cosmic ray modulation near the heliospheric current sheet

Galactic cosmic rays (GCRs) are modulated by the heliospheric magnetic field (HMF) both over decadal time scales (due to long-term, global HMF variations), and over time scales of a few hours (associated with solar wind structures such as coronal mass ejections or the heliospheric current sheet, HCS). Due to the close association between the HCS, the streamer belt, and the band of slow solar wind, HCS crossings are often associated with corotating interaction regions where fast solar wind catches up and compresses slow solar wind ahead of it. However, not all HCS crossings are associated with strong compressions. In this study we categorize HCS crossings in two ways: Firstly, using the change in magnetic polarity, as either away-to-toward (AT) or toward-to-away (TA) magnetic field directions relative to the Sun and, secondly, using the strength of the associated solar wind compression, determined from the observed plasma density enhancement. For each category, we use superposed epoch analyses to show differences in both solar wind parameters and GCR flux inferred from neutron monitors. For strong-compression HCS crossings, we observe a peak in neutron counts preceding the HCS crossing, followed by a large drop after the crossing, attributable to the so-called ‘snow-plough’ effect. For weak-compression HCS crossings, where magnetic field polarity effects are more readily observable, we instead observe that the neutron counts have a tendency to peak in the away magnetic field sector. By splitting the data by the dominant polarity at each solar polar region, we find that the increase in GCR flux prior to the HCS crossing is primarily from strong compressions in cycles with negative north polar fields due to GCR drift effects. Finally, we report on unexpected differences in GCR behavior between TA weak compressions during opposing polarity cycles.

Elastic net orthogonal forward regression

An efficient two-level model identification method aiming at maximising a model׳s generalisation capability is proposed for a large class of linear-in-the-parameters models from the observational data. A new elastic net orthogonal forward regression (ENOFR) algorithm is employed at the lower level to carry out simultaneous model selection and elastic net parameter estimation. The two regularisation parameters in the elastic net are optimised using a particle swarm optimisation (PSO) algorithm at the upper level by minimising the leave one out (LOO) mean square error (LOOMSE). There are two elements of original contributions. Firstly an elastic net cost function is defined and applied based on orthogonal decomposition, which facilitates the automatic model structure selection process with no need of using a predetermined error tolerance to terminate the forward selection process. Secondly it is shown that the LOOMSE based on the resultant ENOFR models can be analytically computed without actually splitting the data set, and the associate computation cost is small due to the ENOFR procedure. Consequently a fully automated procedure is achieved without resort to any other validation data set for iterative model evaluation. Illustrative examples are included to demonstrate the effectiveness of the new approaches.

The effect of the tensor force on the predicted stability of superheavy nuclei

The effect of the tensor component of the Skyrme effective nucleon-nucleon interaction on the single-particle structure in superheavy elements is studied. A selection of the available Skyrme forces has been chosen and their predictions for the proton and neutron shell closures investigated. The inclusion of the tensor term with realistic coupling strength parameters leads to a small increase in the spin-orbit splitting between the proton 2f7/2 and 2f5/2 partners, opening the Z=114 shell gap over a wide range of nuclei. The Z=126 shell gap, predicted by these models in the absence of the tensor term, is found to be stongly dependent on neutron number with a Z=138 gap opening for large neutron numbers, having a consequent implication for the synthesis of neutron-rich superheavy elements. The predicted neutron shell structures remain largely unchanged by inclusion of the tensor component.

Intolerance of uncertainty predicts fear extinction in amygdala-ventromedial prefrontal cortical circuitry

Background: Coordination of activity between the amygdala and ventromedial prefrontal cortex (vmPFC) is important for fear-extinction learning. Aberrant recruitment of this circuitry is associated with anxiety disorders. Here, we sought to determine if individual differences in future threat uncertainty sensitivity, a potential risk factor for anxiety disorders, underly compromised recruitment of fear extinction circuitry. Twenty-two healthy subjects completed a cued fear conditioning task with acquisition and extinction phases. During the task, pupil dilation, skin conductance response, and functional magnetic resonance imaging were acquired. We assessed the temporality of fear extinction learning by splitting the extinction phase into early and late extinction. Threat uncertainty sensitivity was measured using self-reported intolerance of uncertainty (IU). Results: During early extinction learning, we found low IU scores to be associated with larger skin conductance responses and right amygdala activity to learned threat vs. safety cues, whereas high IU scores were associated with no skin conductance discrimination and greater activity within the right amygdala to previously learned safety cues. In late extinction learning, low IU scores were associated with successful inhibition of previously learned threat, reflected in comparable skin conductance response and right amgydala activity to learned threat vs. safety cues, whilst high IU scores were associated with continued fear expression to learned threat, indexed by larger skin conductance and amygdala activity to threat vs. safety cues. In addition, high IU scores were associated with greater vmPFC activity to threat vs. safety cues in late extinction. Similar patterns of IU and extinction learning were found for pupil dilation. The results were specific for IU and did not generalize to self-reported trait anxiety. Conclusions: Overall, the neural and psychophysiological patterns observed here suggest high IU individuals to disproportionately generalize threat during times of uncertainty, which subsequently compromises fear extinction learning. More broadly, these findings highlight the potential of intolerance of uncertainty-based mechanisms to help understand pathological fear in anxiety disorders and inform potential treatment targets.

Electronic structure of porphyrin-based metal-organic frameworks and their suitability for solar fuel production photocatalysis

Metal-organic frameworks (MOFs) can be exceptionally good catalytic materials thanks to the presence of active metal centres and a porous structure that is advantageous for molecular adsorption and confinement. We present here a first-principles investigation of the electronic structure of a family of MOFs based on porphyrins connected through phenyl-carboxyl ligands and AlOH species, in order to assess their suitability for the photocatalysis of fuel production reactions using sunlight. We consider structures with protonated porphyrins and those with the protons exchanged with late 3d metal cations (Fe2+, Co2+, Ni2+, Cu2+, Zn2+), a process that we find to be thermodynamically favorable from aqueous solution for all these metals. Our band structure calculations, based on an accurate screened hybrid functional, reveal that the bandgaps are in a favorable range (2.0 to 2.6 eV) for efficient adsorption of solar light. Furthermore, by approximating the vacuum level to the pore center potential, we provide the alignment of the MOFs’ band edges with the redox potentials for water splitting and carbon dioxide reduction, and show that the structures studied here have band edges positions suitable for these reactions at neutral pH.

Temporally stable feature clusters for maritime object tracking in visible and thermal imagery

This paper describes a new approach to detect and track maritime objects in real time. The approach particularly addresses the highly dynamic maritime environment, panning cameras, target scale changes, and operates on both visible and thermal imagery. Object detection is based on agglomerative clustering of temporally stable features. Object extents are first determined based on persistence of detected features and their relative separation and motion attributes. An explicit cluster merging and splitting process handles object creation and separation. Stable object clus- ters are tracked frame-to-frame. The effectiveness of the approach is demonstrated on four challenging real-world public datasets.

Tracer Leakage from Modeled Agulhas Rings

In a numerical, isopycnal, ocean model the mixing is investigated with the environment of two idealized Agulhas rings, one that splits, and one that remains coherent. The evolution of a passive tracer , initially contained within the rings, shows that tracer leakage is associated with the for mation of filaments in the early stage of ring evolution. These filaments reach down to the ther mocline. In the deepest layers leakage occurs on a larger scale. Self-advection of the rings is ver y irregular , and it is not possible to compute a Lagrangian boundar y i n order to estimate the transport of leakage from the rings. T o describe the processes that gover n tracer leakage, in a coordinate frame moving with the ring a kinematic separatrix is defined in the streamfunction field for the nondivergent flow . Initially , filaments arise because of the elongation of the ring, which is mainly gover ned by an m 5 2 instability that is collaborating with differential rotation. Because of beta, the symmetr y i s destroyed related to the separatrix associated with a stagnation point in the flow . The filament upstream of the stagnation point grows much faster and is associated with the bulk of tracer leakage. Mixing is enhanced by time dependence of the separatrix. As a result, there are no large differences between the leakage from a coherent ring, where the m 5 2 instability equilibrates, and from a splitting ring, where the m 5 2 instability keeps growing, which confir ms that the amount of leakage is mainly gover ned by the ring’ s initial defor mation combined with unsteady self-advection of the ring and not by the splitting of the ring. The decay of tracer content in the ther mocline shows that in the first months up to 40% of the ring water can be mixed with the environment. In deeper layers the decay of tracer content may reach up to 90%.