Quantum computers must operate in concert with classical computers to deliver on the promise of quantum advantage for practical problems. To achieve that, it is important to understand how quantum and classical computing can interact together, and how one can characterize the scalability and efficiency of hybrid quantum–classical workflows. So far, early experiments with quantum-centric supercomputing workflows have been limited in scale and complexity. Here, we use a Heron quantum processor deployed on premises with the entire supercomputer Fugaku to perform the largest computation of electronic structure involving quantum and classical high-performance computing. We design a closed-loop workflow between the quantum processors and 152,064 classical nodes of Fugaku, to approximate the electronic structure of chemistry models beyond the reach of exact diagonalization, with accuracy comparable to some all-classical approximation methods. Our work pushes the limits of the integration of quantum and classical high-performance computing, showcasing computational resource orchestration at the largest scale possible for current classical supercomputers.

Virtual reality (VR) has become an increasingly popular way for learning and training. The assessment of the amount of mental effort, or cognitive load required to perform a task, is essential to create adaptive VR experiences. In this work, we conducted a large-scale study (N=738) to collect behavioral and physiological measures under different cognitive load conditions in a VR environment, and developed a novel machine learning solution to predict cognitive load in real time. Our model predicts cognitive load as a continuous value in the range from 0 to 1, where 0 and 1 correspond to the lowest and highest reported cognitive loads across all participants. On top of the point estimation of cognitive load, our model quantifies prediction uncertainty using a prediction interval. We propose a novel dual-branch attention model to accurately predict the cognitive load. We achieve a MAE (Mean Absolute Error) of 0.11. The result indicates that, with a combination of behavioral and physiological indicators, we can reliably predict cognitive load in real-time, without calibration. To support further research, we are releasing a test dataset comprising data from 100 participants for use by researchers and developers interested in machine learning, virtual reality, learning & memory, cognition, or psychophysiology. This dataset includes recordings from multiple sensors (including pupillometry, eye-tracking, and pulse plethysmography), self-reported cognitive effort, behavioral task performance, and demographic information on the sample.

Quantum computing simulation on a classical computer is difficult due to the exponential runtime and memory overhead. Previous work addresses the difficulty by utilizing multiple Graphical Processing Units (GPUs) and multi-node computers. GPUs are efficient for handling runtime issues but have limited total accessible memory space. Meanwhile, the memory of a multi-node computer can be scaled to the petabytes order, but its bandwidth for access from host computers (CPUs) is narrow. To simultaneously accelerate simulation and enlarge the total memory space, we propose a heterogeneous parallelization approach by combining GPUs and CPUs. Our simulator allocates memory to the GPUs first, and then to the CPUs. It thus accelerates simulation by using the full capabilities of the GPUs if memory for the simulation fits in the GPUs on a cluster. Allocating memory to the CPUs reduces benefits of the GPUs but enlarges the capacity of qubits in the simulation. In such case, it can exploit the memory of the GPUs to add one more qubit in the simulation if the size of memory in a node is the power of two (such as 512GB). We show empirical performance evaluations of our simulator in a distributed environment of POWER9.

Configuration tuning of software is often a good option to improve application performance without any application code modifications. Although we can casually change configurations, it is not easy to apply optimal configurations, as optimal configurations require deep knowledge of the underlying system. This is problematic because applications with suboptimal configuration result in poor performance. As container and container management systems have emerged as an application platform on the cloud, configuration tuning becomes even more challenging because containers add more complexity to the application performance. We need to consider not only fundamental misconfiguration but also container image verification, deployment configuration, application characteristics awareness based on metrics and logs. Although previous knowledge regarding how we should tune configurations for a system software is sometimes available, knowledge about performance tuning practices is neither normalized nor reusable to expand on any advice for misconfiguration to the containers. Even in the cloud-native environment, there is no centralized service to deliver knowledge continuously to application containers nor a framework to develop a misconfiguration fix rule for a container throughout its lifetime. In this paper, we propose a performance-centric configuration tuning framework for containers on Kubernetes, named ConfAdvisor, that enables containers to achieve a higher performance by validating various misconfigurations adaptively. ConfAdivsor gives config tuning advice to application containers, images, and Kubernetes specs and also provides a development framework to build configuration validation rules. We present the design of ConfAdvisor and provide several case studies to tune application containers in the real world.

Parallel copying garbage collection (GC) is widely used in the de facto Java virtual machines such as OpenJDK and OpenJ9. OpenJDK uses work-stealing for copying objects in the Parallel GC and Garbage-First (G1) GC policies to balance the copying task among GC threads. When a thread has no task in its own queue, it tries to steal a task from another thread's queue as a thief. When a thief succeeds in stealing a task, it processes the task and enqueues the children of the task into its queue, which is accessible from other thieves.Unfortunately, the overhead of the work-stealing framework becomes non-negligible when we aim to achieve a minimum GC pause time by increasing the number of GC threads. Since the number of tasks processed per thread decreases, thieves frequently try to steal tasks from others at a low success rate. When a thief fails in steals continuously, it needs to wait in a spin loop on the termination protocol of the work-stealing framework. Spinning in a loop frequently results in high CPU utilization, which is not acceptable in a large-scale data center where severe power management is required. This paper proposes two approaches named steal-best-of-many selection and spin-less termination to reduce the overhead in the work-stealing framework. Steal-best-of-many selection reduces steal failures by changing the number of queue selections to steal in accordance with the number of GC threads. Spin-less termination moves a part of the object copies into a spin loop by changing the procedure of copying GC. It reduces part of the GC pause time for the object copy as well as the CPU utilization for the spin loop. We developed a prototype on OpenJDK8 and evaluated it using SPECjbb2015 and SPECjvm2008 benchmarks. Critical-jOPS performance of SPECjbb2015 improved by 18% at maximum and scores of the SPECjvm2008 benchmarks improved by 1-5%.

While Hadoop is the de facto standard big-data middleware, many frameworks have been developed on top of it. Since many SQL-on-Hadoop systems are available, we often consider which engine is best for our queries. We can choose not only query engines but also Java virtual machines (JVMs) as well. As their systems become more complex, however, it is not always true that a single system performs best at any time. Moreover, the performance of a mismatched system may degrade greatly. To exploit the best performance, it is important to know what type of queries are suitable for a system and then to schedule queries for the appropriate system. In this paper, we evaluated the TPC-DS benchmark on a combination of query engines (Spark and Tez) and JVMs (J9 and OpenJDK). We found that using different engines lead to a drawback of over 10 times and that using different JVMs leads to a drawback of 3 times. We also analyzed the characteristics of each combination and then proposed classification models for selecting the best combination of systems with a generated query plan. As a result, we achieved a performance improvement of up to two times in total with the classifier.

Columnar storage is a data source for data analytics in distributed computing frameworks. For portability and scalability, columnar storage is built on top of existing distributed file systems with columnar data representations such as Parquet, RCFile, and ORC. However, these representations fail to utilize high-level information (e.g., columnar formats) for low-level disk buffer management in operating systems. As a result, data analytics workloads suffer from redundant memory buffers with expensive garbage collections, unnecessary disk readahead, and cache pollution in the operating system buffer cache.We propose column cache, which unifies and re-structures the buffers and caches of multiple software layers from columnar storage to operating systems. Column cache leverages high-level information such as file formats and query plans for enabling adaptive disk reads and cache eviction policies. We have developed a column cache prototype for Apache Parquet and observed that our prototype reduced redundant resource utilization in Apache Spark. Specifically, with our prototype, Spark showed a maximum speedup of 1.28x in TPC-DS workloads while increasing Linux page cache size by 18%, reducing total disk reads by 43%, and reducing garbage collection time in a Java virtual machine by 76%.

Work-stealing is promising for scheduling and balancing parallel workloads. It has a wide range of applicability on middleware, libraries, and runtime systems of programming languages. OpenJDK uses work-stealing for copying garbage collection (GC) to balance copying tasks among GC threads. Each thread has its own queue to store tasks. When a thread has no task in its queue, it acts as a thief and attempts to steal a task from another thread's queue. However, this work-stealing algorithm requires expensive memory fences for pushing, popping, and stealing tasks, especially on weak memory models such as POWER and ARM. To address this problem, we propose a work-stealing algorithm that uses double queues. Each GC thread has a public queue that is accessible from other GC threads and a private queue that is only accessible by itself. Pushing and popping tasks in the private queue are free from expensive memory fences. The most significant point in our algorithm is providing a mechanism to maintain the load balance on the basis of the use of double queues. We developed a prototype implementation for parallel GC in OpenJDK8 for ppc64le. We evaluated our algorithm by using SPECjbb2015, SPECjvm2008, TPC-DS, and Apache DayTrader.

Haptic feedback in VR is important for realistic simulation in virtual reality. However, recreating the haptic experience for hand tools in VR traditionally requires hardware with precise actuators, adding complexity to the system. We propose Ungrounded Haptic Retargeting, an interaction technique that provides a realistic haptic experience for grabbing tools using only passive mechanisms. This technique leverages the ungrounded feedback inherent in grabbing tools combined with dynamic visual adjustments of their position in virtual reality to create an illusion of physical presence for virtual objects. To demonstrate the capabilities of this technique, we created VR Grabbers, an exemplary passive VR controller, similar to training chopsticks, with haptic feedback for precise object selection and manipulation. We conducted two user studies based on VR Grabbers. The first study probed the perceptual limits of the illusion; we found that the maximum position difference between the virtual and physical world acceptable to the user is (-1.48, 1.95) cm. The second study showed that task performance of the VR Grabbers controller with Ungrounded Haptic Retargeting enabled outperforms the same controller with Ungrounded Haptic Retargeting disabled.

The microservice architecture is essential for agile development and deployment of the application components; however, designing microservices for a web application is not a straight-forward task. One of the best ways to design microservices is to decompose a monolithic prototype of an application into microservices on the basis of both the complexity in engineering and the component boundaries of the application in the early phase of development. We propose a visualization tool allowing developers to interactively design microservice applications on the basis of the characteristics of source codes and the behaviors of a monolithic prototype. This visualization tool first constructs a calling-context tree from profile data taken in a dry-run of the application. Next, it generates an initial microservice design while considering keyword features in the source codes or amount of function calls between components. Developers can interactively refine this design via this visual interface by taking four-choice actions to revise boundaries of microservices while considering expected communications between them. This interaction will have a significant impact on runtime performance. Case studies of two open-source benchmark applications demonstrate the proposed tool enables interactive design of microservices. The results of the demonstration show that compared to the official microservice designs of the applications, the proposed tool can effectively design microservice applications.

Electric Agents is a cross-media game that presents new ways for children to actively engage with television content. In the game Manny, a member of the Pranksters, steals words out of the mouth of Hector, a member of The Electric Company team, and hides them to prevent the story from progressing. Children collaborate through a mobile augmented reality experience to find and collect the missing vocabulary words relevant to the show narrative. The players return the stolen words back to the show by "throwing" the words towards the television using their mobile devices. This blend of a narrative and a game strives to make educational television content more engaging and participatory while fostering collaborative play and use of specific vocabulary words.

This paper outlines a series of experiments to develop asynchronous messaging systems for preschool aged children. Three unique systems build on a foundational design called Toaster, a jack-in-the box toy with embedded mobile phone that allows children to playfully take and share electronic media. Orange Toaster allows children to create and share self-portraits; Family Toast allows children to browse family photos with physical tokens, and shares their self-portrait reactions with remote family members; Play with Elmo allows children and distant adults to asynchronously share playful video messages. Observations with over 30 children suggest that asynchronous photographic or video messaging with very young children is possible. The results of these studies indicate specific guidelines including (1) children's UI's need to be playful and immediate (2) UI designs for children should create the "here and now" feel of real-time interaction, and (3) adults' UI's must provide emotionally meaningful feedback from children to engage adult users.

StoryVisit allows children and long-distance adults to experience a sense of togetherness by reading children's story books together over a distance. StoryVisit combines video conferencing and connected books: remote grown-up and child readers can see and hear each other, and can also see and control the same e-book. We report on research with 61 families - over 200 users including parents, children and long-distance readers - who used StoryVisit in their homes with a long-distance reader for at least one reading session. In addition, we report qualitative findings regarding nineteen of the families who participated in telephone interviews and four families who were monitored and interviewed by researchers at home. Results show that connected e-book video chat sessions last about five times as long as the typical video chats reported in previous research on families with young children. Moreover, the addition of an animated character increased session lengths by another 50%. StoryVisit usage peaked for families with three year olds, showing that sustained distance interactions with very young children are possible if communication technologies incorporate joint activities that engage children and adults.

Users often engage in tasks that span multiple personal devices. Although many current solutions exist to provide ubiquitous access to one's data, users continue to struggle with cross-device tasks. These solutions often require them to plan ahead for their information needs. In this paper, we present Myngle, a device-agnostic system that lets users quickly find the information they are looking for from previously visited web pages without having to plan ahead. Myngle provides a unified web history from multiple personal devices, and allows users to filter their history based on high-level categories influenced by common mobile information need categories (e.g., address, phone number). We evaluated Myngle with 32 users and found that our category-based method of filtering eases the burden of continuing cross-device tasks.

In this work, we introduce Interactive Rich Reading, a new enhanced book experience designed to run on smartphones and a tablet device. Interactive Rich Reading is characterized by a video-based conversational agent that asks questions or makes comments about the current page and is specifically designed to promote engagement with the contents of children's books. We use video compositing techniques to overlay the conversational agent directly over the book contents, creating a magical experience for children by bringing the book to life. We describe technical issues related to enabling this experience on mobile platforms for easier adoption of this technique by other researchers and practitioners.

People interact with numerous personal devices on a daily basis. Sharing content among these devices is often done depending on the device capabilities and context of use; following turn-by-turn directions is more appropriate when mobile. Although several solutions exist to share content among one's devices, these solutions rely on the user planning ahead for the data he may need on another device. In this paper, we describe a system that addresses the unplanned activities, by automatically extracting addresses and points of interest that users view in their web browser and making those readily available through an in-car interface.

We introduce Family Story Play, a system that supports grandparents to read books together with their grandchildren over the Internet. Family Story Play is designed to improve communication across generations and over a distance, and to support parents and grandparents in fostering the literacy development of young children. The interface encourages active child participation in the book reading experience by combining a paper book, a sensor-enhanced frame, video conferencing technology, and video content of a Sesame Street Muppet (Elmo). Results with users indicate that Family Story Play improves child engagement in long-distance communication and increases the quality of interaction between young children and distant grandparents. Additionally, Family Story Play encourages dialogic reading styles that are linked with literacy development. Ultimately, reading with Family Story Play becomes a creative shared activity that suggests a new kind of collaborative story telling.

Increasingly, smartphones are being used to access all manner of information: email messages, Facebook status updates, tweets, RSS feeds, photographs and more. Approaches to dealing with this multi-faceted information stream developed on the desktop, such as switching between multiple applications or multiple browser windows, are unwieldy and scale poorly for mobile devices. In this paper, we propose the combination of the universal inbox and a system called 'Lenses' for extracting information of interest as part of a solution to this problem. These mechanisms allow the user to easily specify ways to sort, filter and manage their universal inbox in an intuitive way. We culminate with a discussion of implications for mobile phone interface design.

SMS messages provide an easy and simple method to com-municate with others. These short messages are useful, but can sometimes feel restricted due to the limitations of textual communication. The ability to express subtle nuances and contexts around the message could help add enjoyment and amplify the emotions being expressed by the mobile user. We present SensorComix, a new way of creating comics us-ing SMS messages combined with gestures on mobile phones. Comics are automatically generated from users ’ SMS mes-sages, and augmented with visual icons based on the per-formed gestures. We demonstrate that gestures mapped to comics can help influence the expressiveness of messages sent by mobile users.