Computation on dynamic n-tuples of particles is ubiquitous in scientific computing, with an archetypal application in many-body molecular dynamics (MD) simulations. We propose a tuple-decomposition (TD) approach that reorders computations according to dynamically created lists of n-tuples. We analyze the performance characteristics of the TD approach on general purpose graphics processing units for MD simulations involving pair (n = 2) and triplet (n = 3) interactions. The results show superior performance of the TD approach over the conventional particle-decomposition (PD) approach. Detailed analyses reveal the register footprint as the key factor that dictates the performance. Furthermore, the TD approach is found to outperform PD for more intensive computations of quadruplet (n = 4) interactions in first principles-informed reactive MD simulations based on the reactive force-field (ReaxFF) method. This work thus demonstrates the viable performance portability of the TD approach across a wide range of applications.

This paper is focusing on the Facial ExpressionRecognition (FER) problem from a single face image. Inspired by the advances Convolutional Neural Networks (CNNs) haveachieved in image recognition and classification, we propose a CNN-based approach to address this problem. Our model consists of several different structured subnets. Each subnet is a compactCNN model trained separately. The whole network is structured by assembling these subnets together. We trained and evaluated our model on the FER2013 dataset[7]. The best single subnet achieved 62:44% accuracy and the whole model scored 65:03% accuracy, which is ranked 9th and 5th respectively among all other participants.

The development of sophisticated adaptive source coding algorithms together with inherent error sensitivity problems fostered the need for efficient space communication at very low bit error probilbilities(\leq 10^{-6}). This led to the specification and implementation of a concatenated coding system using an interleaved Reed-Solomon code as the outer code and a Viterbi-decoded convolutionai code as the inner code. This paper presents the experimental results of this channel coding system under an emulatedS-band uplink andX-band downlink two-way space communication channel, where both uplink and downlink have strong carrier power. Test results verify that at a bit error probability of 10-6or less, this concatenated coding system does provide a coding gain of 2.5 dB or more over the viterbi-decoded convolutional-only coding system. These tests also show that a desirable interleaving depth for the Reed-Solomon outer code is 8 or more. The imptict of this "virtually" error-free space communication link on the transmission of images is discussed and examples of Simulation results are given.