A hallmark of human intelligence, but challenging for reinforcement learning (RL) agents, is the ability to compositionally generalise, that is, to recompose familiar knowledge components in novel ...ways to solve new problems. For instance, when navigating in a city, one needs to know the location of the destination and how to operate a vehicle to get there, whether it be pedalling a bike or operating a car. In RL, these correspond to the reward function and transition function, respectively. To compositionally generalize, these two components need to be transferable independently of each other: multiple modes of transport can reach the same goal, and any given mode can be used to reach multiple destinations. Yet there are also instances where it can be helpful to learn and transfer entire structures, jointly representing goals and transitions, particularly whenever these recur in natural tasks (e.g., given a suggestion to get ice cream, one might prefer to bike, even in new towns). Prior theoretical work has explored how, in model-based RL, agents can learn and generalize task components (transition and reward functions). But a satisfactory account for how a single agent can simultaneously satisfy the two competing demands is still lacking. Here, we propose a hierarchical RL agent that learns and transfers individual task components as well as entire structures (particular compositions of components) by inferring both through a non-parametric Bayesian model of the task. It maintains a factorised representation of task components through a hierarchical Dirichlet process, but it also represents different possible covariances between these components through a standard Dirichlet process. We validate our approach on a variety of navigation tasks covering a wide range of statistical correlations between task components and show that it can also improve generalisation and transfer in more complex, hierarchical tasks with goal/subgoal structures. Finally, we end with a discussion of our work including how this clustering algorithm could conceivably be implemented by cortico-striatal gating circuits in the brain.
The clinical syndrome of heart failure (HF) can be described as the reduced capacity of the heart to deliver blood throughout the body. To compensate for inadequate tissue perfusion, the ...renin–angiotensin aldosterone system (RAAS) and the sympathetic nervous system (SNS) become activated, resulting in increased blood pressure, heart rate, and blood volume. Consequent activation of the natriuretic peptide system (NPS) typically balances these effects; however, the NPS is unable to sustain compensation for excessive neurohormonal activation over time. Until recently, mortality benefits have been provided to patients with HF only by therapies that target the RAAS and SNS, including angiotensin-converting enzyme inhibitors (ACEIs), angiotensin receptor blockers (ARBs), mineralocorticoid receptor antagonists, and beta-blockers. Sacubitril/valsartan, the first-in-class angiotensin receptor/neprilysin inhibitor (ARNI), targets both the NPS and RAAS to further improve clinical outcomes. This review discusses the focused management of patients with HF with reduced ejection fraction (HFrEF) and suggests changes to current management paradigms. From this assessment, the evidence supports favoring sacubitril/valsartan over ACEIs or ARBs, and this therapy should be used in conjunction with beta-blockers to further decrease morbidity and mortality in patients with HFrEF.
This paper examines the behavior of closed "lattice universes" wherein masses are distributed in a regular lattice on the Cauchy surfaces of closed vacuum universes. Such universes are approximated ...using a form of Regge calculus originally developed by Collins and Williams to model closed Friedmann-Lemaitre-Robertson-Walker universes. We consider two types of lattice universes, one where all masses are identical to each other and another where one mass gets perturbed in magnitude. In the unperturbed universe, we consider the possible arrangements of the masses in the Regge Cauchy surfaces and demonstrate that the model will only be stable if each mass lies within some spherical region of convergence. We also briefly discuss the existence of Regge models that are dual to the ones we have considered. We then model a perturbed lattice universe and demonstrate that the model's evolution is well behaved, with the expansion increasing in magnitude as the perturbation is increased.
Western diet (WD) intake, aging, and inactivation of farnesoid X receptor (FXR) are risk factors for metabolic and chronic inflammation-related health issues ranging from metabolic ...dysfunction-associated steatotic liver disease (MASLD) to dementia. The progression of MASLD can be escalated when those risks are combined. Inactivation of FXR, the receptor for bile acid (BA), is cancer prone in both humans and mice. The current study used multi-omics including hepatic transcripts, liver, serum, and urine metabolites, hepatic BAs, as well as gut microbiota from mouse models to classify those risks using machine learning. A linear support vector machine with
-fold cross-validation was used for classification and feature selection. We have identified that increased urine sucrose alone achieved 91% accuracy in predicting WD intake. Hepatic lithocholic acid and serum pyruvate had 100% and 95% accuracy, respectively, to classify age. Urine metabolites (decreased creatinine and taurine as well as increased succinate) or increased gut bacteria (
,
, and
) could predict FXR deactivation with greater than 90% accuracy. Human disease relevance is partly revealed using the metabolite-disease interaction network. Transcriptomics data were also compared with the human liver disease datasets. WD-reduced hepatic
(cytochrome P450 family 39 subfamily a member 1) and increased
(GRAM domain containing 1B) were also changed in human liver cancer and metabolic liver disease, respectively. Together, our data contribute to the identification of noninvasive biomarkers within the gut-liver axis to predict metabolic status.
Differences in gait patterns of children with Duchenne muscular dystrophy (DMD) and typically developing (TD) peers are visible to the eye, but quantifications of those differences outside of the ...gait laboratory have been elusive. In this work, we measured vertical, mediolateral, and anteroposterior acceleration using a waist-worn iPhone accelerometer during ambulation across a typical range of velocities. Fifteen TD and fifteen DMD children from 3 to 16 years of age underwent eight walking/running activities, including five 25 m walk/run speed-calibration tests at a slow walk to running speeds (SC-L1 to SC-L5), a 6-min walk test (6MWT), a 100 m fast walk/jog/run (100MRW), and a free walk (FW). For clinical anchoring purposes, participants completed a Northstar Ambulatory Assessment (NSAA). We extracted temporospatial gait clinical features (CFs) and applied multiple machine learning (ML) approaches to differentiate between DMD and TD children using extracted temporospatial gait CFs and raw data. Extracted temporospatial gait CFs showed reduced step length and a greater mediolateral component of total power (TP) consistent with shorter strides and Trendelenberg-like gait commonly observed in DMD. ML approaches using temporospatial gait CFs and raw data varied in effectiveness at differentiating between DMD and TD controls at different speeds, with an accuracy of up to 100%. We demonstrate that by using ML with accelerometer data from a consumer-grade smartphone, we can capture DMD-associated gait characteristics in toddlers to teens.
•A new volume of solid based implicit forcing immersed boundary (VOS-IFIB) method is proposed.•The modified pressure Poisson equation (MPPE) is derived to couple the forcing and pressure terms.•A ...prediction-correction procedure is adopted to solve the MPPE.•The predicted velocity field within the solid region is almost identical to the body velocity and satisfies the divergence-free condition simultaneously.•The proposed VOS-IFIB method can simulate problem with sharp angle domain accurately.
In this study, a new implicit forcing immersed boundary (IFIB) method is proposed to solve incompressible viscous fluid flow problems involving complex domains. In the conventional immersed boundary (IB) method, a forcing term computed from the volume of solid (VOS) is added to the incompressible Navier-Stokes equations in order to satisfy the velocity condition within the embedded solid body. However, the velocity boundary condition and the divergence-free condition are enforced at different time levels, i.e. intermediate and new time levels. Penetration of streamlines into the stationary solid body is visible as the velocity boundary condition inside the solid body is not strictly enforced at the new time level. In the current work, the proposed IFIB method can ensure velocity field which satisfies both the velocity boundary condition and the divergence-free condition at the same (new) time level. This is accomplished by solving the pressure equation and calculating the forcing term simultaneously (implicitly). A modified pressure Poisson equation (MPPE) is derived in order to couple the pressure and the forcing terms by treating the forcing term as part of the source term of MPPE. Also, a new cell-based method is proposed to compute the VOS for getting a better parallel efficiency. The accuracy of the present IFIB method is then demonstrated by solving several benchmark problems. No penetration of streamlines has been found in the solid body.
In this article, the high-order upwinding combined compact difference scheme developed in a three-point grid stencil is applied to solve the incompressible Navier-Stokes (NS) and energy equations in ...three dimensions. The time integrator with symplectic property is employed to approximate the temporal derivative term in inviscid Euler equation so as to numerically retain the embedded Hamiltions and Casimir to get long-time accurate solutions. For the sake of computational efficiency in solving the three-dimensional NS equations, all the calculations will be accelerated using the hybrid CUDA and OpenAcc GPU programing models. The parallel speedup performance compared to the multicore of an Intel Xeon E5-2690V5 CPU is reported.
In the community of computational fluid dynamics, pressure Poisson equation with Neumann boundary condition is usually encountered when solving the incompressible Navier-Stokes equations in a ...segregated approach such as SIMPLE, PISO, and projection methods. To deal with Neumann boundary conditions more naturally and to retain high order spatial accuracy as well, a sixth-order accurate combined compact difference scheme developed on staggered grids (NSCCD6) is adopted to solve the parabolic and elliptic equations subject to Neumann boundary conditions. The staggered grid system is usually used when solving the incompressible Navier-Stokes equations. By adopting the combined compact difference concept, there is no need to discretize Neumann boundary conditions with one-sided discretization scheme which is of lower accuracy order. The conventional Crank-Nicolson scheme is applied in this study for temporal discretization. For two-dimensional cases, D'yakonov alternating direction implicit scheme is adopted. A newly proposed time step changing strategy is adopted to improve convergence rate when solving the steady state solutions of the parabolic equation. High accuracy order of the currently proposed NSCCD6 scheme for one- and two-dimensional cases are shown in this article.
•The IMLE method is proposed to solve the incompressible Navier–Stokes equations.•Multiple GPUs are adopted to accelerate the computation.•A data decomposition strategy is proposed to achieve higher ...speedup ratio.•The speedup ratio can up to 70x for adopting four GPU cards.
In this study, a GPU-accelerated improved mixed Lagrangian–Eulerian (IMLE) method is proposed to solve the three-dimensional incompressible Navier–Stokes equations. To improve the prediction accuracy, the proposed IMLE method approximates the total derivative term in Lagragian sense, and the spatial derivative terms are approximated on Eulerian coordinates. Transfer of data from Lagrangian particles to data on Eulerian grids is accurately carried out by adopting moving least squares (MLS) interpolation method. The velocity-pressure decoupling issue is overcome by adopting pressure-free projection method in which the pressure field is calculated by solving a pressure Poisson equation (PPE). It is noted that the MLS interpolation is time consuming since this procedure belongs to a pointwise scheme in which a local matrix equation shall be solved on each grid point. In addition, the discretized PPE forms a large sparse matrix and it is computationally intensive to solve by using the conjugate gradient (CG) method. Therefore, we are aimed to resort to CUDA- and OpenMP-programming means to accelerate the computation. In this study, the performance of the multiple GPUs code can reach up to 27 times faster with respect to multi-threads CPU performance.
PDF
