Seismograms are convolution results between seismic sources and the media that seismic waves propagate through, and, therefore, the primary observations for studying seismic source parameters and the Earth interior. The routine earthquake location and travel‐time tomography rely on accurate seismic phase picks (e.g., P and S arrivals). As data increase, reliable automated seismic phase‐picking methods are needed to analyze data and provide timely earthquake information. However, most traditional autopickers suffer from low signal‐to‐noise ratio and usually require additional efforts to tune hyperparameters for each case. In this study, we proposed a deep‐learning approach that adapted soft attention gates (AGs) and recurrent‐residual convolution units (RRCUs) into the backbone U‐Net for seismic phase picking. The attention mechanism was implemented to suppress responses from waveforms irrelevant to seismic phases, and the cooperating RRCUs further enhanced temporal connections of seismograms at multiple scales. We used numerous earthquake recordings in Taiwan with diverse focal mechanisms, wide depth, and magnitude distributions, to train and test our model. Setting the picking errors within 0.1 s and predicted probability over 0.5, the AG with recurrent‐residual convolution unit (ARRU) phase picker achieved the F1 score of 98.62% for P arrivals and 95.16% for S arrivals, and picking rates were 96.72% for P waves and 90.07% for S waves. The ARRU phase picker also shown a great generalization capability, when handling unseen data. When applied the model trained with Taiwan data to the southern California data, the ARRU phase picker shown no cognitive downgrade. Comparing with manual picks, the arrival times determined by the ARRU phase picker shown a higher consistency, which had been evaluated by a set of repeating earthquakes. The arrival picks with less human error could benefit studies, such as earthquake location and seismic tomography.

We examine the plausibility of using an Artificial Neural Network (ANN) and an Importance-Aided Neural Network (IANN) for the refinement of the structural model used to create full-wave tomography images. Specifically, we apply the machine learning techniques to classifying segments of observed data wave seismograms and synthetic data wave seismograms as either usable for iteratively refining the structural model or not usable for refinement. Segments of observed and synthetic seismograms are considered usable if they are not too di_erent, a heuristic observation made by a human expert, which is considered a match. The use of the ANN and the IANN for classification of the data wave segments removes the human computational cost of the classification process and removes the need for an expert to oversee all such classifications. Our experiments on the seismic data for Southern California have shown this technique to be promising for both classification accuracy and the reduction of the time required to compute the classification of observed data wave segment and synthetic data wave segment matches

Summary. A central problem of seismology is the inversion of regional waveform data for models of earthquake sources. In regions such as Southern California, p

LSQR (Sparse Equations and Least Squares) is a widely used Krylov subspace method to solve large-scale linear systems in seismic tomography. This paper presents a parallel MPI-CUDA implementation for LSQR solver. On CUDA level, our contributions include: (1) utilize CUBLAS and CUSPARSE to compute major steps in LSQR; (2) optimize memory copy between host memory and device memory; (3) develop a CUDA kernel to perform transpose SpMV without transposing the matrix in memory or preserving additional copy. On MPI level, our contributions include: (1) decompose both matrix and vector to increase parallelism; (2) design a static load balancing strategy. In our experiment, the single GPU code achieves up to 17.6x speedup with 15.7 GFlops in single precision and 15.2x speedup with 12.0 GFlops in double precision compared with the original serial CPU code. The MPI-GPU code achieves up to 3.7x speedup with 268 GFlops in single precision and 3.8x speedup with 223 GFlops in double precision on 135 MPI tasks compared with the corresponding MPI-CPU code. The MPI-GPU code scales on both strong and weak scaling tests. In addition, our parallel implementations have better performance than the LSQR subroutine in PETSc library.

Sparse Matrix-Vector Multiplication (SpMV) is very common to scientific computing. The Graphics Processing Unit (GPU) has recently emerged as a high-performance computing platform due to its massive processing capability. This paper presents an innovative performance-model driven approach for partitioning sparse matrix into appropriate formats, and auto-tuning configurations of CUDA kernels to improve the performance of SpMV on GPUs. This paper makes the following contributions: (1) Propose an empirical CUDA performance model to predict the execution time of SpMV CUDA kernels. (2) Design and implement a model-driven partitioning framework to predict how to partition the target sparse matrix into one or more partitions and transform each partition into appropriate storage format, which is based on the fact that the different storage formats of sparse matrix can significantly affect the performance of SpMV. (3) Integrate the model-driven partitioning with our previous auto-tuning framework to automatically adjust CUDA-specific parameters to optimize performance on specific GPUs. Compared to the NVIDIA's existing implementations, our approach shows a substantial performance improvement. It has 222%, 197%, and 33% performance improvement on the average for CSR vector kernel, ELL kernel and HYB kernel, respectively.