Generalization of ML-Based Surrogate Models to Dynamic Injection Schedules using Transfer Learning
For presentation at the FECM/ NETL Carbon Management Project Review Meeting, Pittsburgh, PA, August 28-September 1, 2023.
Riedl, Michael↗
SEARCH · Search NASA
Results for “Dynamic scheduling”
Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.
Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.
Enabling Autonomous Rover Science through Dynamic Planning and Scheduling
This paper describes how dynamic planning and scheduling techniques can be used onboard a rover to autonomously adjust rover activities in support of science goals. These goals could be identified by scientists on the ground or could be identified by onboard data-analysis software. Several different types of dynamic decisions are described, including the handling of opportunistic science goals identified during rover traverses, preserving high priority science targets when resources, such as power, are unexpectedly over-subscribed, and dynamically adding additional, ground-specified science targets when rover actions are executed more quickly than expected. After describing our specific system approach, we discuss some of the particular challenges we have examined to support autonomous rover decision-making. These include interaction with rover navigation and path-planning software and handling large amounts of uncertainty in state and resource estimations.
planning↗
Towards Evolving Electronic Circuits for Autonomous Space Applications
The relatively new field of Evolvable Hardware studies how simulated evolution can reconfigure, adapt, and design hardware structures in an automated manner. Space applications, especially those requiring autonomy, are potential beneficiaries of evolvable hardware. For example, robotic drilling from a mobile platform requires high-bandwidth controller circuits that are difficult to design. In this paper, we present automated design techniques based on evolutionary search that could potentially be used in such applications. First, we present a method of automatically generating analog circuit designs using evolutionary search and a circuit construction language. Our system allows circuit size (number of devices), circuit topology, and device values to be evolved. Using a parallel genetic algorithm, we present experimental results for five design tasks. Second, we investigate the use of coevolution in automated circuit design. We examine fitness evaluation by comparing the effectiveness of four fitness schedules. The results indicate that solution quality is highest with static and co-evolving fitness schedules as compared to the other two dynamic schedules. We discuss these results and offer two possible explanations for the observed behavior: retention of useful information, and alignment of problem difficulty with circuit proficiency.
Lohn, Jason D.↗
Addressing Load Imbalance in Bioinformatics and Biomedical Applications: Efficient Scheduling across Multiple GPUs
Computational bioinformatics and biomedical applications frequently contain heterogeneously sized units of work or tasks, for instance due to variability in the sizes of biological sequences and molecules. Variable-sized workloads lead to load imbalances in parallel implementations which detract from efficiency and performance. Many modern computing resources now have multiple graphics processing units(GPUs) per computer for acceleration. These multiple GPU resources need to be used efficiently through balancing of workloads across the GPUs. OpenMP is a portable directive-based parallel programming API used ubiquitously in bioscience applications to program CPUs; recently, the use of OpenMP directives for GPU acceleration has become possible. Here, motivated by experiences with imbalanced loads in GPU-accelerated bioinformatics applications, we address the load balancing problem using OpenMP task-to-GPU scheduling combined with OpenMP GPU offloading for multiply heterogeneous workloads – loads with both variable input sizes, and simultaneously, variable convergence rates for algorithms with a stochastic component – scheduled across multiple GPUs. We aim to develop strategies which are both easy to use and have lower overheads, and may be incorporated incrementally in existing programs which already make use of OpenMP for CPU-based threading in order to make use of multi-GPU computers. We test different combinations of input size variability and convergence rate variability, and characterize the effects of these different scenarios on the performance of scheduling strategies across multiple GPUs with OpenMP. We present several dynamic scheduling solutions for different parallel patterns, explore optimizations, and provide publicly available example computational kernels to make these strategies easy to use in programs. This work will enable application developers to efficiently and easily use multiple GPUs for imbalanced workloads found in bioinformatics and biomedical applications.
Thavappiragasam, Mathialakan↗
Channel access schemes and fiber optic configurations for integrated-services local area networks
Local Area Networks are in common use for data communications and have enjoyed great success. Recently, there is a growing interest in using a single network to support many applications in addition to traditional data traffic. These additional applications introduce new requirements in terms of volume of traffic and real-time delivery of data which are not met by existing networks. To satisfy these requirements, a high-bandwidth tranmission medium, such as fiber optics, and a distributed channel access scheme for the efficient sharing of the bandwidth among the various applications are needed. As far as the throughput-delay requirements of the various application are concerned, a network structure along with a distributed channel access are proposed which incorporate appropriate scheduling policies for the transmission of outstanding messages on the network. A dynamic scheduling policy was devised which outperforms all existing policies in terms of minimizing the expected cost per message. A broadcast mechanism was devised for the efficient dissemination of all relevant information. Fiber optic technology is considered for the high-bandwidth transmisison medium.
Nassehi, M. Mehdi↗
Integrating production scheduling and process control using latent variable dynamic models
Not provided.
Automation & Control Systems↗
Expert system for on-board satellite scheduling and control
An Expert System is described which Rockwell Satellite and Space Electronics Division (S&SED) is developing to dynamically schedule the allocation of on-board satellite resources and activities. This expert system is the Satellite Controller. The resources to be scheduled include power, propellant and recording tape. The activities controlled include scheduling satellite functions such as sensor checkout and operation. The scheduling of these resources and activities is presently a labor intensive and time consuming ground operations task. Developing a schedule requires extensive knowledge of the system and subsystems operations, operational constraints, and satellite design and configuration. This scheduling process requires highly trained experts anywhere from several hours to several weeks to accomplish. The process is done through brute force, that is examining cryptic mnemonic data off line to interpret the health and status of the satellite. Then schedules are formulated either as the result of practical operator experience or heuristics - that is rules of thumb. Orbital operations must become more productive in the future to reduce life cycle costs and decrease dependence on ground control. This reduction is required to increase autonomy and survivability of future systems. The design of future satellites require that the scheduling function be transferred from ground to on board systems.
Barry, John M.↗
IRIS-GNN: Leveraging Graph Neural Networks for Scheduling on Truly Heterogeneous Runtime Systems
The diversity of accelerators in computer systems poses significant challenges for software developers, such as managing vendor-specific compiler toolchains, code fragmentation requiring different kernel implementations, and performance portability issues. To address these, the Intelligent Runtime System (IRIS) was developed. IRIS works across various systems, from smartphones to supercomputers, enabling automatic performance scaling based on available accelerators. It introduces abstract tasks for seamless execution transitions between accelerators while ensuring memory consistency and task dependencies. Although IRIS simplifies system details, optimal dynamic scheduling still requires user input to understand workload structures. To address this, we introduce a new scheduling policy for IRIS, termed IRIS-GNN, which is the first IRIS hybrid policy that operates in conjunction with the dynamic policies. This policy employs a Graph-Neural Network (GNN) to conduct Graph Classification of any task graphs submitted to IRIS. This GNN analyzes the structure and attributes of the task graph, categorizing it as either locality, concurrency, or mixed. This classification subsequently guides the selection of the dynamic policy used by IRIS. We provide a comparison of the performance of IRIS-GNN against the complete spectrum of IRIS’s dynamic policies, assess the overhead introduced by the GNN within this scheduling framework, and ultimately explore its practical application in real-world scenarios.
Johnston, Beau↗
Co-optimization of repairs and dynamic network reconfiguration for improved distribution system resilience
In this work, a post-disaster distribution system repair and restoration (DSRR) strategy is proposed to improve distribution system resilience. The DSRR strategy is formulated as a two-stage optimization. The first stage is a comprehensive co-optimization of repair crew scheduling, dynamic network reconfiguration, and distributed energy resource (DER) dispatch based on the forecast load profile. The goal is to minimize the accumulative operating cost caused by the load reduction payment as well as DER operating cost. In particular, since the number of available repair crews is usually smaller than the number of faulted lines after a disaster event, the DSRR strategy determines the optimal scheduling for repairing faulted lines. The second stage is a re-dispatch of the DER power output and load shedding based on the real-time load demand of each bus. The proposed algorithm is validated by case studies of the IEEE 33-bus and 123-bus test systems. We consider those scenarios in which faults occur in multiple heavy-loaded feeders. The simulation results demonstrate that the DSRR strategy effectively coordinate the repair scheduling, network reconfiguration and load shedding to minimize the operating cost.
24 POWER TRANSMISSION AND DISTRIBUTION↗
Method for resource control in parallel environments using program organization and run-time support
A system and method for dynamic scheduling and allocation of resources to parallel applications during the course of their execution. By establishing well-defined interactions between an executing job and the parallel system, the system and method support dynamic reconfiguration of processor partitions, dynamic distribution and redistribution of data, communication among cooperating applications, and various other monitoring actions. The interactions occur only at specific points in the execution of the program where the aforementioned operations can be performed efficiently.
Ekanadham, Kattamuri↗
Method for resource control in parallel environments using program organization and run-time support
A system and method for dynamic scheduling and allocation of resources to parallel applications during the course of their execution. By establishing well-defined interactions between an executing job and the parallel system, the system and method support dynamic reconfiguration of processor partitions, dynamic distribution and redistribution of data, communication among cooperating applications, and various other monitoring actions. The interactions occur only at specific points in the execution of the program where the aforementioned operations can be performed efficiently.
Ekanadham, Kattamuri↗
Efficient mapping algorithms for scheduling robot inverse dynamics computation on a multiprocessor system
Two efficient mapping algorithms for scheduling the robot inverse dynamics computation consisting of m computational modules with precedence relationship to be executed on a multiprocessor system consisting of p identical homogeneous processors with processor and communication costs to achieve minimum computation time are presented. An objective function is defined in terms of the sum of the processor finishing time and the interprocessor communication time. The minimax optimization is performed on the objective function to obtain the best mapping. This mapping problem can be formulated as a combination of the graph partitioning and the scheduling problems; both have been known to be NP-complete. Thus, to speed up the searching for a solution, two heuristic algorithms were proposed to obtain fast but suboptimal mapping solutions. The first algorithm utilizes the level and the communication intensity of the task modules to construct an ordered priority list of ready modules and the module assignment is performed by a weighted bipartite matching algorithm. For a near-optimal mapping solution, the problem can be solved by the heuristic algorithm with simulated annealing. These proposed optimization algorithms can solve various large-scale problems within a reasonable time. Computer simulations were performed to evaluate and verify the performance and the validity of the proposed mapping algorithms. Finally, experiments for computing the inverse dynamics of a six-jointed PUMA-like manipulator based on the Newton-Euler dynamic equations were implemented on an NCUBE/ten hypercube computer to verify the proposed mapping algorithms. Computer simulation and experimental results are compared and discussed.
Lee, C. S. G.↗
Resource constrained design of artificial neural networks using comparator neural network
We present a systematic design method executed under resource constraints for automating the design of artificial neural networks using the back error propagation algorithm. Our system aims at finding the best possible configuration for solving the given application with proper tradeoff between the training time and the network complexity. The design of such a system is hampered by three related problems. First, there are infinitely many possible network configurations, each may take an exceedingly long time to train; hence, it is impossible to enumerate and train all of them to completion within fixed time, space, and resource constraints. Second, expert knowledge on predicting good network configurations is heuristic in nature and is application dependent, rendering it difficult to characterize fully in the design process. A learning procedure that refines this knowledge based on examples on training neural networks for various applications is, therefore, essential. Third, the objective of the network to be designed is ill-defined, as it is based on a subjective tradeoff between the training time and the network cost. A design process that proposes alternate configurations under different cost-performance tradeoff is important. We have developed a Design System which schedules the available time, divided into quanta, for testing alternative network configurations. Its goal is to select/generate and test alternative network configurations in each quantum, and find the best network when time is expended. Since time is limited, a dynamic schedule that determines the network configuration to be tested in each quantum is developed. The schedule is based on relative comparison of predicted training times of alternative network configurations using comparator network paradigm. The comparator network has been trained to compare training times for a large variety of traces of TSSE-versus-time collected during back-propagation learning of various applications.
Wah, Benjamin W.↗
Elastic distributed training with fast convergence and efficient resource utilization
Distributed learning is now routinely conducted on cloud as well as dedicated clusters. Training with elastic resources brings new challenges and design choices. Prior studies focus on runtime performance and assume a static algorithmic behavior. In this work, by analyzing the impact of of resource scaling on convergence, we introduce schedules for synchronous stochastic gradient descent that proactively adapt the number of learners to reduce training time and improve convergence. Our approach no longer assumes a constant number of processors throughout training. In our experiment, distributed stochastic gradient descent with dynamic schedules and reduction momentum achieves better convergence and significant speedups over prior static ones. Numerous distributed training jobs running on cloud may benefit from our approach.
Cong, Guojing↗
User-Assisted Store Recycling for Dynamic Task Graph Schedulers
The emergence of the multi-core era has led to increased interest in designing effective yet practical parallel programming models. Models based on task graphs that operate on single-assignment data are attractive in several ways: they can support dynamic applications and precisely represent the available concurrency. However, they also require nuanced algorithms for scheduling and memory management for efficient execution. In this paper, we consider memory-efficient dynamic scheduling of task graphs. Specifically, we present a novel approach for dynamically recycling the memory locations assigned to data items as they are produced by tasks. We develop algorithms to identify memory-efficient store recycling functions by systematically evaluating the validity of a set of (user-provided or automatically generated) alternatives. Because recycling function can be input data-dependent, we have also developed support for continued correct execution of a task graph in the presence of a potentially incorrect store recycling function. Experimental evaluation demonstrates that our approach to automatic store recycling incurs little to no overheads, achieves memory usage comparable to the best manually derived solutions, often produces recycling functions valid across problem sizes and input parameters, and efficiently recovers from an incorrect choice of store recycling functions.
97 MATHEMATICS AND COMPUTING↗
Vehicle to Grid Frequency Regulation Capacity Optimal Scheduling for Battery Swapping Station Using Deep Q-Network
Battery swapping stations (BSSs) are ideal candidates for fast frequency regulation services (FFRS) due to their large battery stock capacity. In addition, BSSs can precharge batteries for customers and the batteries that are not in charging can provide a stable regulation capacity to the market. However, uncertainties, such as ACE signals and the EV per-hour visit counts, introduce stochastic nonlinear dynamics into the operation of a BSS-based FFRS. Currently, there is no quantification method to ensure its optimal economical operation. To close this gap, in this article, we propose a novel deep Q-learning-based FFRS capacity dynamic scheduling strategy. This method can autonomously schedule the hourly regulation capacity in real time to maximize the BSSx0027;s revenue for providing FFRS. Case studies using real-world data verify the efficacy of the proposed work.
25 ENERGY STORAGE↗
On the adequacy of lightweight thread approaches for high-level parallel programming models
High-level parallel programming models (PMs) are becoming crucial in order to extract the computational power of current on-node multi-threaded parallelism. The most popular PMs, such as OpenMP or OmpSs, are directive-based: the complexity of the hardware is hidden by the underlying runtime system, improving coding productivity. The implementations of OpenMP usually rely on POSIX threads (pthreads), offering excellent performance for coarse-grained parallelism and a perfect match with the current hardware. OmpSs is a task oriented PM based on an ad hoc runtime solution called Nanos++; it is the precursor of the tasking parallelism in the OpenMP tasking specification. A recent trend in runtimes and applications points to leveraging massive on-node parallelism in conjunction with fine-grained and dynamic scheduling paradigms. In this paper we analyze the behavior of the OpenMP and OmpSs PMs on top of the recently emerged Generic Lightweight Threads (GLT) API. GLT exposses a common API for lightweight thread (LWT) libraries that offers the possibility of running the same application over different native LWT solutions. We describe the design details of those high-level PMs implemented on top of GLT and analyze different scenarios in order to assess where the use of LWTs may benefit application performance. Our work reveals those scenarios where LWTs overperform pthread-based solutions and compares the performance between an ad hoc solution and a generic implementation.
97 MATHEMATICS AND COMPUTING↗
Cross-scale efficient tensor contractions for coupled cluster computations through multiple programming model backends
Coupled-cluster methods provide highly accurate models of molecular structure through explicit numerical calculation of tensors representing the correlation between electrons. These calculations are dominated by a sequence of tensor contractions, motivating the development of numerical libraries for such operations. While based on matrix–matrix multiplication, these libraries are specialized to exploit symmetries in the molecular structure and in electronic interactions, and thus reduce the size of the tensor representation and the complexity of contractions. The resulting algorithms are irregular and their parallelization has been previously achieved via the use of dynamic scheduling or specialized data decompositions. We introduce our efforts to extend the Libtensor framework to work in the distributed memory environment in a scalable and energy-efficient manner. We achieve up to 240× speedup compared with the optimized shared memory implementation of Libtensor. We attain scalability to hundreds of thousands of compute cores on three distributed-memory architectures (Cray XC30 and XC40, and IBM Blue Gene/Q), and on a heterogeneous GPU-CPU system (Cray XK7). As the bottlenecks shift from being compute-bound DGEMM's to communication-bound collectives as the size of the molecular system scales, we adopt two radically different parallelization approaches for handling load-imbalance, tasking and bulk synchronous models. Nevertheless, we preserve a unified interface to both programming models to maintain the productivity of computational quantum chemists.
97 MATHEMATICS AND COMPUTING↗