Showing 1–50 of 210 results for author: Kuo, T

A Survey of Generative Information Retrieval

Authors: Tzu-Lin Kuo, Tzu-Wei Chiu, Tzung-Sheng Lin, Sheng-Yang Wu, Chao-Wei Huang, Yun-Nung Chen

Abstract: Generative Retrieval (GR) is an emerging paradigm in information retrieval that leverages generative models to directly map queries to relevant document identifiers (DocIDs) without the need for traditional query processing or document reranking. This survey provides a comprehensive overview of GR, highlighting key developments, indexing and retrieval strategies, and challenges. We discuss various… ▽ More Generative Retrieval (GR) is an emerging paradigm in information retrieval that leverages generative models to directly map queries to relevant document identifiers (DocIDs) without the need for traditional query processing or document reranking. This survey provides a comprehensive overview of GR, highlighting key developments, indexing and retrieval strategies, and challenges. We discuss various document identifier strategies, including numerical and string-based identifiers, and explore different document representation methods. Our primary contribution lies in outlining future research directions that could profoundly impact the field: improving the quality of query generation, exploring learnable document identifiers, enhancing scalability, and integrating GR with multi-task learning frameworks. By examining state-of-the-art GR techniques and their applications, this survey aims to provide a foundational understanding of GR and inspire further innovations in this transformative approach to information retrieval. We also make the complementary materials such as paper collection publicly available at https://github.com/MiuLab/GenIR-Survey/ △ Less

Submitted 4 June, 2024; v1 submitted 3 June, 2024; originally announced June 2024.

Charge transport through the multiple end zigzag edge states of armchair graphene nanoribbons and heterojunctions

Authors: David M T Kuo

Abstract: This comprehensive study investigates charge transport through the multiple end zigzag edge states of finite-size armchair graphene nanoribbons/boron nitride nanoribbons (n-AGNR/w-BNNR) junctions under a longitudinal electric field, where n and w denote the widths of the AGNRs and the BNNRs, respectively. In 13-atom wide AGNR segments, the edge states exhibit a blue Stark shift in response to the… ▽ More This comprehensive study investigates charge transport through the multiple end zigzag edge states of finite-size armchair graphene nanoribbons/boron nitride nanoribbons (n-AGNR/w-BNNR) junctions under a longitudinal electric field, where n and w denote the widths of the AGNRs and the BNNRs, respectively. In 13-atom wide AGNR segments, the edge states exhibit a blue Stark shift in response to the electric field, with only the long decay length zigzag edge states showing significant interaction with the red Stark shift subband states. Charge tunneling through such edge states assisted by the subband states is elucidated in the spectra of the transmission coefficient. In the 13-AGNR/6-BNNR heterojunction, notable influences on the energy levels of the end zigzag edge states of 13-AGNRs induced by BNNR segments are observed. We demonstrate the modulation of these energy levels in resonant tunneling situations, as depicted by bias-dependent transmission coefficient spectra. Intriguing nonthermal broadening of tunneling current shows a significant peak-to-valley ratio. Our findings highlight the promising potential of n-AGNR/w-BNNR heterojunctions with long decay length edge states in the realm of GNR-based single electron transistors at room temperature. △ Less

Submitted 14 June, 2024; v1 submitted 27 May, 2024; originally announced May 2024.

Comments: In the revised manuscript, we have added new references

RAEE: A Training-Free Retrieval-Augmented Early Exiting Framework for Efficient Inference

Authors: Lianming Huang, Shangyu Wu, Yufei Cui, Ying Xiong, Xue Liu, Tei-Wei Kuo, Nan Guan, Chun Jason Xue

Abstract: Deploying large language model inference remains challenging due to their high computational overhead. Early exiting accelerates model inference by adaptively reducing the number of inference layers. Existing methods require training internal classifiers to determine whether to exit at each intermediate layer. However, such classifier-based early exiting frameworks require significant effort to de… ▽ More Deploying large language model inference remains challenging due to their high computational overhead. Early exiting accelerates model inference by adaptively reducing the number of inference layers. Existing methods require training internal classifiers to determine whether to exit at each intermediate layer. However, such classifier-based early exiting frameworks require significant effort to design and train the classifiers. To address these limitations, this paper proposes RAEE, a training-free Retrieval-Augmented Early Exiting framework for efficient inference. First, this paper demonstrates that the early exiting problem can be modeled as a distribution prediction problem, where the distribution is approximated using similar data's existing information. Next, the paper details the process of collecting existing information to build the retrieval database. Finally, based on the pre-built retrieval database, RAEE leverages the retrieved similar data's exiting information to guide the backbone model to exit at the layer, which is predicted by the approximated distribution. Experimental results demonstrate that the proposed RAEE can significantly accelerate inference. RAEE also achieves state-of-the-art zero-shot performance on 8 classification tasks. △ Less

Submitted 24 May, 2024; originally announced May 2024.

A Gauss-Bonnet formula for the renormalized area of minimal submanifolds of Poincaré-Einstein manifolds

Authors: Jeffrey S. Case, C Robin Graham, Tzu-Mo Kuo, Aaron J. Tyrrell, Andrew Waldron

Abstract: Assuming the extrinsic $Q$-curvature admits a decomposition into the Pfaffian, a scalar conformal submanifold invariant, and a tangential divergence, we prove that the renormalized area of an even-dimensional minimal submanifold of a Poincaré-Einstein manifold can be expressed as a linear combination of its Euler characteristic and the integral of a scalar conformal submanifold invariant. We deriv… ▽ More Assuming the extrinsic $Q$-curvature admits a decomposition into the Pfaffian, a scalar conformal submanifold invariant, and a tangential divergence, we prove that the renormalized area of an even-dimensional minimal submanifold of a Poincaré-Einstein manifold can be expressed as a linear combination of its Euler characteristic and the integral of a scalar conformal submanifold invariant. We derive such a decomposition of the extrinsic $Q$-curvature in dimensions two and four, thereby recovering and generalizing results of Alexakis-Mazzeo and Tyrrell, respectively. We also conjecture such a decomposition for general natural submanifold scalars whose integral over compact submanifolds is conformally invariant, and verify our conjecture in dimensions two and four. Our results also apply to the area of a compact even-dimensional minimal submanifold of an Einstein manifold. △ Less

Submitted 25 March, 2024; originally announced March 2024.

Comments: 47 pages

Wikibench: Community-Driven Data Curation for AI Evaluation on Wikipedia

Authors: Tzu-Sheng Kuo, Aaron Halfaker, Zirui Cheng, Jiwoo Kim, Meng-Hsin Wu, Tongshuang Wu, Kenneth Holstein, Haiyi Zhu

Abstract: AI tools are increasingly deployed in community contexts. However, datasets used to evaluate AI are typically created by developers and annotators outside a given community, which can yield misleading conclusions about AI performance. How might we empower communities to drive the intentional design and curation of evaluation datasets for AI that impacts them? We investigate this question on Wikipe… ▽ More AI tools are increasingly deployed in community contexts. However, datasets used to evaluate AI are typically created by developers and annotators outside a given community, which can yield misleading conclusions about AI performance. How might we empower communities to drive the intentional design and curation of evaluation datasets for AI that impacts them? We investigate this question on Wikipedia, an online community with multiple AI-based content moderation tools deployed. We introduce Wikibench, a system that enables communities to collaboratively curate AI evaluation datasets, while navigating ambiguities and differences in perspective through discussion. A field study on Wikipedia shows that datasets curated using Wikibench can effectively capture community consensus, disagreement, and uncertainty. Furthermore, study participants used Wikibench to shape the overall data curation process, including refining label definitions, determining data inclusion criteria, and authoring data statements. Based on our findings, we propose future directions for systems that support community-driven data curation. △ Less

Submitted 21 February, 2024; originally announced February 2024.

Journal ref: Proceedings of the 2024 CHI Conference on Human Factors in Computing Systems (CHI '24)

ReFusion: Improving Natural Language Understanding with Computation-Efficient Retrieval Representation Fusion

Authors: Shangyu Wu, Ying Xiong, Yufei Cui, Xue Liu, Buzhou Tang, Tei-Wei Kuo, Chun Jason Xue

Abstract: Retrieval-based augmentations (RA) incorporating knowledge from an external database into language models have greatly succeeded in various knowledge-intensive (KI) tasks. However, integrating retrievals in non-knowledge-intensive (NKI) tasks is still challenging. Existing works focus on concatenating retrievals with inputs to improve model performance. Unfortunately, the use of retrieval concaten… ▽ More Retrieval-based augmentations (RA) incorporating knowledge from an external database into language models have greatly succeeded in various knowledge-intensive (KI) tasks. However, integrating retrievals in non-knowledge-intensive (NKI) tasks is still challenging. Existing works focus on concatenating retrievals with inputs to improve model performance. Unfortunately, the use of retrieval concatenation-based augmentations causes an increase in the input length, substantially raising the computational demands of attention mechanisms. This paper proposes a new paradigm of RA named \textbf{ReFusion}, a computation-efficient Retrieval representation Fusion with bi-level optimization. Unlike previous works, ReFusion directly fuses the retrieval representations into the hidden states of models. Specifically, ReFusion leverages an adaptive retrieval integrator to seek the optimal combination of the proposed ranking schemes across different model layers. Experimental results demonstrate that the proposed ReFusion can achieve superior and robust performance in various NKI tasks. △ Less

Submitted 27 May, 2024; v1 submitted 4 January, 2024; originally announced January 2024.

Thermoelectric Properties of Armchair Graphene Nanoribbons with Array Characteristics

Authors: David M T Kuo

Abstract: The thermoelectric properties of armchair graphene nanoribbons (AGNRs) with array characteristics are investigated theoretically using the tight-binding model and Green's function technique. The AGNR structures with array characteristics are created by embedding a narrow boron nitride nanoribbon (BNNR) into a wider AGNR, resulting in two narrow AGNRs. This system is denoted as w-AGNR/n-BNNR, where… ▽ More The thermoelectric properties of armchair graphene nanoribbons (AGNRs) with array characteristics are investigated theoretically using the tight-binding model and Green's function technique. The AGNR structures with array characteristics are created by embedding a narrow boron nitride nanoribbon (BNNR) into a wider AGNR, resulting in two narrow AGNRs. This system is denoted as w-AGNR/n-BNNR, where 'w' and 'n' represent the widths of the wider AGNR and narrow BNNR, respectively. We elucidate the coupling effect between two narrow symmetrical AGNRs on the electronic structure of w-AGNR/n-BNNR. A notable discovery is that the power factor of the 15-AGNR/5-BNNR with the minimum width surpasses the quantum limitation of power factor for 1D ideal systems. The energy level degeneracy observed in the first subbands of w-AGNR/n-BNNR structures proves to be highly advantageous in enhancing the electrical power outputs of graphene nanoribbon devices. △ Less

Submitted 16 January, 2024; v1 submitted 10 December, 2023; originally announced December 2023.

Comments: 6 pages and 5 figures, revised manuscript

DMLR: Data-centric Machine Learning Research -- Past, Present and Future

Authors: Luis Oala, Manil Maskey, Lilith Bat-Leah, Alicia Parrish, Nezihe Merve Gürel, Tzu-Sheng Kuo, Yang Liu, Rotem Dror, Danilo Brajovic, Xiaozhe Yao, Max Bartolo, William A Gaviria Rojas, Ryan Hileman, Rainier Aliment, Michael W. Mahoney, Meg Risdal, Matthew Lease, Wojciech Samek, Debojyoti Dutta, Curtis G Northcutt, Cody Coleman, Braden Hancock, Bernard Koch, Girmaw Abebe Tadesse, Bojan Karlaš , et al. (13 additional authors not shown)

Abstract: Drawing from discussions at the inaugural DMLR workshop at ICML 2023 and meetings prior, in this report we outline the relevance of community engagement and infrastructure development for the creation of next-generation public datasets that will advance machine learning science. We chart a path forward as a collective effort to sustain the creation and maintenance of these datasets and methods tow… ▽ More Drawing from discussions at the inaugural DMLR workshop at ICML 2023 and meetings prior, in this report we outline the relevance of community engagement and infrastructure development for the creation of next-generation public datasets that will advance machine learning science. We chart a path forward as a collective effort to sustain the creation and maintenance of these datasets and methods towards positive scientific, societal and business impact. △ Less

Submitted 1 June, 2024; v1 submitted 21 November, 2023; originally announced November 2023.

Comments: Published in the Journal of Data-centric Machine Learning Research (DMLR) at https://data.mlr.press/assets/pdf/v01-5.pdf

Measuring Adversarial Datasets

Authors: Yuanchen Bai, Raoyi Huang, Vijay Viswanathan, Tzu-Sheng Kuo, Tongshuang Wu

Abstract: In the era of widespread public use of AI systems across various domains, ensuring adversarial robustness has become increasingly vital to maintain safety and prevent undesirable errors. Researchers have curated various adversarial datasets (through perturbations) for capturing model deficiencies that cannot be revealed in standard benchmark datasets. However, little is known about how these adver… ▽ More In the era of widespread public use of AI systems across various domains, ensuring adversarial robustness has become increasingly vital to maintain safety and prevent undesirable errors. Researchers have curated various adversarial datasets (through perturbations) for capturing model deficiencies that cannot be revealed in standard benchmark datasets. However, little is known about how these adversarial examples differ from the original data points, and there is still no methodology to measure the intended and unintended consequences of those adversarial transformations. In this research, we conducted a systematic survey of existing quantifiable metrics that describe text instances in NLP tasks, among dimensions of difficulty, diversity, and disagreement. We selected several current adversarial effect datasets and compared the distributions between the original and their adversarial counterparts. The results provide valuable insights into what makes these datasets more challenging from a metrics perspective and whether they align with underlying assumptions. △ Less

Submitted 6 November, 2023; originally announced November 2023.

Comments: ART of Safety workshop (AACL 2023)

A Collaborative Filtering-Based Two Stage Model with Item Dependency for Course Recommendation

Authors: Eric L. Lee, Tsung-Ting Kuo, Shou-De Lin

Abstract: Recommender systems have been studied for decades with numerous promising models been proposed. Among them, Collaborative Filtering (CF) models are arguably the most successful one due to its high accuracy in recommendation and elimination of privacy-concerned personal meta-data from training. This paper extends the usage of CF-based model to the task of course recommendation. We point out several… ▽ More Recommender systems have been studied for decades with numerous promising models been proposed. Among them, Collaborative Filtering (CF) models are arguably the most successful one due to its high accuracy in recommendation and elimination of privacy-concerned personal meta-data from training. This paper extends the usage of CF-based model to the task of course recommendation. We point out several challenges in applying the existing CF-models to build a course recommendation engine, including the lack of rating and meta-data, the imbalance of course registration distribution, and the demand of course dependency modeling. We then propose several ideas to address these challenges. Eventually, we combine a two-stage CF model regularized by course dependency with a graph-based recommender based on course-transition network, to achieve AUC as high as 0.97 with a real-world dataset. △ Less

Submitted 1 November, 2023; originally announced November 2023.

Comments: 8 pages, 2 figures, 2017 IEEE International Conference on Data Science and Advanced Analytics (DSAA)

Journal ref: In 2017 IEEE DSAA, pp. 496-503. IEEE, 2017

Electronic structures and transport properties of cove-edged graphene nanoribbons

Authors: David M T Kuo

Abstract: In this comprehensive study, we undertake a thorough theoretical examination of the electronic subband structures within cove-edged zigzag graphene nanoribbons (CZGNRs) using the tight-binding model. These unique nanostructures arise from the systematic removal of carbon atoms along the zigzag edges of conventional zigzag graphene nanoribbons (ZGNRs). Notably, CZGNRs that exhibit intriguing band g… ▽ More In this comprehensive study, we undertake a thorough theoretical examination of the electronic subband structures within cove-edged zigzag graphene nanoribbons (CZGNRs) using the tight-binding model. These unique nanostructures arise from the systematic removal of carbon atoms along the zigzag edges of conventional zigzag graphene nanoribbons (ZGNRs). Notably, CZGNRs that exhibit intriguing band gaps can be conceptualized as interconnected graphene quantum dots (GQDs). An essential finding of our investigation is the inverse relationship between the size of GQDs and the band gaps of CZGNRs, a relationship that remains consistent regardless of the number of GQDs present. Additionally, we delve into the examination of electron effective masses in proximity to the edges of the first conduction subband of CZGNRs as GQD sizes expand. We observe a significant increase in electron effective masses as GQDs become larger, which is attributed to the increasing similarity between larger GQDs and ZGNRs. To further understand the practical implications, we explore the transport properties of finite CZGNRs when connected to electrodes through line contacts. The presence of edge defects introduces intriguing asymmetries in the tunneling current, leading to a significant reduction in its magnitude. Notably, we observe that the saturation current magnitude is less influenced by the length of CZGNRs and is instead more sensitive to the choice of materials used for the contacted electrodes. Lastly, we investigate the tunneling currents through GQDs featuring boron nitride textures within the Coulomb blockade region, unveiling an irregular staircase-like pattern in the tunneling current behavior. △ Less

Submitted 24 October, 2023; originally announced October 2023.

Comments: 10 pages and 11 figures

Tunable non-Lifshitz-Kosevich temperature dependence of Shubnikov-de Haas oscillation amplitudes in SmSb

Authors: Wei Zhang, C. N. Kuo, S. T. Kuo, Chun Wa So, Jianyu Xie, Kwing To Lai, Wing Chi Yu, C. S. Lue, Hoi Chun Po, Swee K. Goh

Abstract: The Lifshitz-Kosevich (LK) theory is the pillar of magnetic quantum oscillations, which have been extensively applied to characterize a wide range of metallic states. In this study, we focus on the Shubnikov-de Haas (SdH) effect observed in SmSb, a rare-earth monopnictide. We observed a significant departure from the expected LK theory near $T_N=2.4$~K: both a peak-like anomaly and an enhancement… ▽ More The Lifshitz-Kosevich (LK) theory is the pillar of magnetic quantum oscillations, which have been extensively applied to characterize a wide range of metallic states. In this study, we focus on the Shubnikov-de Haas (SdH) effect observed in SmSb, a rare-earth monopnictide. We observed a significant departure from the expected LK theory near $T_N=2.4$~K: both a peak-like anomaly and an enhancement in the temperature dependence of quantum oscillation amplitude are seen in SmSb. Moreover, we discovered a remarkable sensitivity of the SdH amplitudes to sample purity. By adjusting the sample purity, we were able to tune the temperature dependence of the $α$ band's SdH amplitudes from a peak-like anomalous behavior to an enhancement. Therefore, SdH oscillations from the $α$ band connect the two well-known non-LK behaviours, controllable through varying the sample purity, paving the way for developing further understanding of the mechanism leading to the anomalous quantum oscillations. △ Less

Submitted 10 October, 2023; originally announced October 2023.

Comments: 4 figures

Journal ref: npj Quantum Materials 8, 55 (2023)

Thermal rectification through the topological states of asymmetrical length armchair graphene nanoribbons heterostructures with vacancies

Authors: David M T Kuo

Abstract: We present a theoretical investigation of electron heat current in asymmetrical length armchair graphene nanoribbon (AGNR) heterostructures with vacancies, focusing on the topological states (TSs). In particular, we examine the 9-7-9 AGNR heterostructures where the TSs are well-isolated from the conduction and valence subbands. This isolation effectively mitigates thermal noise of subbands arising… ▽ More We present a theoretical investigation of electron heat current in asymmetrical length armchair graphene nanoribbon (AGNR) heterostructures with vacancies, focusing on the topological states (TSs). In particular, we examine the 9-7-9 AGNR heterostructures where the TSs are well-isolated from the conduction and valence subbands. This isolation effectively mitigates thermal noise of subbands arising from temperature fluctuations during charge transport. Moreover, when the TSs exhibit an orbital off-set, intriguing electron heat rectification phenomena are observed, primarily attributed to inter-TS electron Coulomb interactions. To enhance the heat rectification ratio ($η_Q$), we manipulate the coupling strengths between the heat sources and the TSs by introducing asymmetrical lengths in the 9-AGNRs. This approach offers control over the rectification properties, enabling significant enhancements. Additionally, we introduce vacancies strategically positioned between the heat sources and the TSs to suppress phonon heat current. This arrangement effectively reduces the overall phonon heat current, while leaving the TSs unaffected. Our findings provide valuable insights into the behavior of electron heat current in AGNR heterostructures, highlighting the role of topological states, inter-TS electron Coulomb interactions, and the impact of structural modifications such as asymmetrical lengths and vacancy positioning. These results pave the way for the design and optimization of graphene-based devices with improved thermal management and efficient control of electron heat transport. △ Less

Submitted 2 October, 2023; v1 submitted 12 September, 2023; originally announced September 2023.

Comments: 13 pages and 13 figures

Sharp results for spherical metric on flat tori with conical angle 6$π$ at two symmetric points

Authors: Ting-Jung Kuo

Abstract: In this paper, we investigate the following curvature equation: \begin{equation} Δu+e^{u}=8π(δ_{0}+δ_{\frac{ω_{k}}{2}})\text{ in } E_{τ}\text{, }τ\in \mathbb{H} (0.1) \label{a} \end{equation} Here $E_{τ}$ represents a flat torus and $\frac{ω_{k}}{2}$ is one of the half periods of $E_{τ}$. Our primary objective is to establish a necessary and sufficient criterion for the existence of a non-even fam… ▽ More In this paper, we investigate the following curvature equation: \begin{equation} Δu+e^{u}=8π(δ_{0}+δ_{\frac{ω_{k}}{2}})\text{ in } E_{τ}\text{, }τ\in \mathbb{H} (0.1) \label{a} \end{equation} Here $E_{τ}$ represents a flat torus and $\frac{ω_{k}}{2}$ is one of the half periods of $E_{τ}$. Our primary objective is to establish a necessary and sufficient criterion for the existence of a non-even family of solutions (see the definition in Section 1). Remarkably, this is equivalent to determining the presence of solutions for the equation with a single conical singularity: \begin{equation*} Δu+e^{u}=8πδ_{0}\text{ in }E_{τ}\text{, }τ\in \mathbb{ H}\text{.} \end{equation*} This study marks the first exploration of the structure of non-even families of solutions to the curvature equation with multiple singular sources in the literature. Building on our findings, we provide a comprehensive analysis of the solution structure for equation (0.1) for all $τ$. This analysis is facilitated by Theorem 1.3, which will play a central role in our exploration of cases involving general parameters in the future, such as: \begin{equation*} Δu+e^{u}=8πn(δ_{0}+δ_{\frac{ω_{k}}{2}})\text{ in } E_{τ},\text{ }n\in \mathbb{N}\text{.} \end{equation*} As an application, we offer explicit descriptions for solutions to equation (0.1) in the context of both rectangle tori and rhombus tori. See Corollary 1.4 as well as Corollary 1.5. △ Less

Submitted 24 January, 2024; v1 submitted 15 August, 2023; originally announced August 2023.

MSC Class: 58J10; 53A10

Effects of Coulomb blockade on the charge transport through the topological states of finite armchair graphene nanoribbons and heterostructures

Authors: David M T Kuo

Abstract: In this study, we investigate the charge transport properties of semiconducting armchair graphene nanoribbons (AGNRs) and heterostructures through their topological states (TSs), with a specific focus on the Coulomb blockade region. Our approach employs a two-site Hubbard model that takes into account both intra- and inter-site Coulomb interactions. Using this model, we calculate the electron ther… ▽ More In this study, we investigate the charge transport properties of semiconducting armchair graphene nanoribbons (AGNRs) and heterostructures through their topological states (TSs), with a specific focus on the Coulomb blockade region. Our approach employs a two-site Hubbard model that takes into account both intra- and inter-site Coulomb interactions. Using this model, we calculate the electron thermoelectric coefficients and tunneling currents of serially coupled TSs (SCTSs). In the linear response regime, we analyze the electrical conductance ($G_e$), Seebeck coefficient ($S$), and electron thermal conductance ($κ_e$) of finite AGNRs. Our results reveal that at low temperatures, the Seebeck coefficient is more sensitive to many-body spectra than the electrical conductance. Furthermore, we observe that the optimized $S$ at high temperature is less sensitive to electron Coulomb interactions than $G_e$ and $κ_e$. In the nonlinear response regime, we observe a tunneling current with negative differential conductance through the SCTSs of finite AGNRs. This current is generated by electron inter-site Coulomb interactions rather than intra-site Coulomb interactions. Additionally, we observe current rectification behavior in asymmetrical junction systems of SCTSs of AGNRs. Notably, we also uncover the remarkable current rectification behavior of SCTSs of 9-7-9 AGNR heterostructure in the Pauli spin blockade configuration. Overall, our study provides valuable insights into the charge transport properties of TSs in finite AGNRs and heterostructures. We emphasize the importance of considering electron-electron interactions in understanding the behavior of these materials. △ Less

Submitted 22 July, 2023; originally announced July 2023.

Comments: corrected some typos

Journal ref: Nanomaterials 2023, 13(11), 1757

LLMs as Workers in Human-Computational Algorithms? Replicating Crowdsourcing Pipelines with LLMs

Authors: Tongshuang Wu, Haiyi Zhu, Maya Albayrak, Alexis Axon, Amanda Bertsch, Wenxing Deng, Ziqi Ding, Bill Guo, Sireesh Gururaja, Tzu-Sheng Kuo, Jenny T. Liang, Ryan Liu, Ihita Mandal, Jeremiah Milbauer, Xiaolin Ni, Namrata Padmanabhan, Subhashini Ramkumar, Alexis Sudjianto, Jordan Taylor, Ying-Jui Tseng, Patricia Vaidos, Zhijin Wu, Wei Wu, Chenyang Yang

Abstract: LLMs have shown promise in replicating human-like behavior in crowdsourcing tasks that were previously thought to be exclusive to human abilities. However, current efforts focus mainly on simple atomic tasks. We explore whether LLMs can replicate more complex crowdsourcing pipelines. We find that modern LLMs can simulate some of crowdworkers' abilities in these "human computation algorithms," but… ▽ More LLMs have shown promise in replicating human-like behavior in crowdsourcing tasks that were previously thought to be exclusive to human abilities. However, current efforts focus mainly on simple atomic tasks. We explore whether LLMs can replicate more complex crowdsourcing pipelines. We find that modern LLMs can simulate some of crowdworkers' abilities in these "human computation algorithms," but the level of success is variable and influenced by requesters' understanding of LLM capabilities, the specific skills required for sub-tasks, and the optimal interaction modality for performing these sub-tasks. We reflect on human and LLMs' different sensitivities to instructions, stress the importance of enabling human-facing safeguards for LLMs, and discuss the potential of training humans and LLMs with complementary skill sets. Crucially, we show that replicating crowdsourcing pipelines offers a valuable platform to investigate (1) the relative strengths of LLMs on different tasks (by cross-comparing their performances on sub-tasks) and (2) LLMs' potential in complex tasks, where they can complete part of the tasks while leaving others to humans. △ Less

Submitted 19 July, 2023; v1 submitted 19 July, 2023; originally announced July 2023.

Extrinsic GJMS operators for submanifolds

Authors: Jeffrey S. Case, C Robin Graham, Tzu-Mo Kuo

Abstract: We derive extrinsic GJMS operators and $Q$-curvatures associated to a submanifold of a conformal manifold. The operators are conformally covariant scalar differential operators on the submanifold with leading part a power of the Laplacian in the induced metric. Upon realizing the conformal manifold as the conformal infinity of an asymptotically Poincaré--Einstein space and the submanifold as the b… ▽ More We derive extrinsic GJMS operators and $Q$-curvatures associated to a submanifold of a conformal manifold. The operators are conformally covariant scalar differential operators on the submanifold with leading part a power of the Laplacian in the induced metric. Upon realizing the conformal manifold as the conformal infinity of an asymptotically Poincaré--Einstein space and the submanifold as the boundary of an asymptotically minimal submanifold thereof, these operators arise as obstructions to smooth extension as eigenfunctions of the Laplacian of the induced metric on the minimal submanifold. We derive explicit formulas for the operators of orders 2 and 4. We prove factorization formulas when the original submanifold is a minimal submanifold of an Einstein manifold. We also show how to reformulate the construction in terms of the ambient metric for the conformal manifold, and use this to prove that the operators defined by the factorization formulas are conformally invariant for all orders in all dimensions. △ Less

Submitted 19 March, 2024; v1 submitted 20 June, 2023; originally announced June 2023.

Comments: 33 pages, v2 has minor editing, Remark 4.3 added, references updated

MSC Class: 53A31 (Primary) 53A55; 53A10; 58J70; 53C18 (Secondary)

BiTrackGAN: Cascaded CycleGANs to Constraint Face Aging

Authors: Tsung-Han Kuo, Zhenge Jia, Tei-Wei Kuo, Jingtong Hu

Abstract: With the increased accuracy of modern computer vision technology, many access control systems are equipped with face recognition functions for faster identification. In order to maintain high recognition accuracy, it is necessary to keep the face database up-to-date. However, it is impractical to collect the latest facial picture of the system's user through human effort. Thus, we propose a bottom… ▽ More With the increased accuracy of modern computer vision technology, many access control systems are equipped with face recognition functions for faster identification. In order to maintain high recognition accuracy, it is necessary to keep the face database up-to-date. However, it is impractical to collect the latest facial picture of the system's user through human effort. Thus, we propose a bottom-up training method for our proposed network to address this challenge. Essentially, our proposed network is a translation pipeline that cascades two CycleGAN blocks (a widely used unpaired image-to-image translation generative adversarial network) called BiTrackGAN. By bottom-up training, it induces an ideal intermediate state between these two CycleGAN blocks, namely the constraint mechanism. Experimental results show that BiTrackGAN achieves more reasonable and diverse cross-age facial synthesis than other CycleGAN-related methods. As far as we know, it is a novel and effective constraint mechanism for more reason and accurate aging synthesis through the CycleGAN approach. △ Less

Submitted 22 April, 2023; originally announced April 2023.

Comments: V1.0

A Reconfigurable Linear RF Analog Processor for Realizing Microwave Artificial Neural Network

Authors: Minning Zhu, Tzu-Wei Kuo, Chung-Tse Michael Wu

Abstract: Owing to the data explosion and rapid development of artificial intelligence (AI), particularly deep neural networks (DNNs), the ever-increasing demand for large-scale matrix-vector multiplication has become one of the major issues in machine learning (ML). Training and evaluating such neural networks rely on heavy computational resources, resulting in significant system latency and power consumpt… ▽ More Owing to the data explosion and rapid development of artificial intelligence (AI), particularly deep neural networks (DNNs), the ever-increasing demand for large-scale matrix-vector multiplication has become one of the major issues in machine learning (ML). Training and evaluating such neural networks rely on heavy computational resources, resulting in significant system latency and power consumption. To overcome these issues, analog computing using optical interferometric-based linear processors have recently appeared as promising candidates in accelerating matrix-vector multiplication and lowering power consumption. On the other hand, radio frequency (RF) electromagnetic waves can also exhibit similar advantages as the optical counterpart by performing analog computation at light speed with lower power. Furthermore, RF devices have extra benefits such as lower cost, mature fabrication, and analog-digital mixed design simplicity, which has great potential in realizing affordable, scalable, low latency, low power, near-sensor radio frequency neural network (RFNN) that may greatly enrich RF signal processing capability. In this work, we propose a 2X2 reconfigurable linear RF analog processor in theory and experiment, which can be applied as a matrix multiplier in an artificial neural network (ANN). The proposed device can be utilized to realize a 2X2 simple RFNN for data classification. An 8X8 linear analog processor formed by 28 RFNN devices are also applied in a 4-layer ANN for Modified National Institute of Standards and Technology (MNIST) dataset classification. △ Less

Submitted 24 July, 2023; v1 submitted 14 April, 2023; originally announced April 2023.

Comments: 11 pages, 16 figures

Hermitian Matrix Diagonalization and its Symmetry Properties

Authors: S. H. Chiu, T. K. Kuo

Abstract: A hermitian matrix can be parametrized by a set consisting of its determinant and the eigenvalues of its submatrices. We established a group of equations which connect these variables with the mixing parameters of diagonalization. These equations are simple in structure and manifestly invariant in form under the symmetry operations of dilatation, translation, rephasing and permutation. When applie… ▽ More A hermitian matrix can be parametrized by a set consisting of its determinant and the eigenvalues of its submatrices. We established a group of equations which connect these variables with the mixing parameters of diagonalization. These equations are simple in structure and manifestly invariant in form under the symmetry operations of dilatation, translation, rephasing and permutation. When applied to the problem of neutrino oscillation in matter they produced two new ``matter invariants" which are confirmed by available data. △ Less

Submitted 14 August, 2023; v1 submitted 29 March, 2023; originally announced March 2023.

Comments: 9 pages, 1 figure. Appendix added

Understanding Frontline Workers' and Unhoused Individuals' Perspectives on AI Used in Homeless Services

Authors: Tzu-Sheng Kuo, Hong Shen, Jisoo Geum, Nev Jones, Jason I. Hong, Haiyi Zhu, Kenneth Holstein

Abstract: Recent years have seen growing adoption of AI-based decision-support systems (ADS) in homeless services, yet we know little about stakeholder desires and concerns surrounding their use. In this work, we aim to understand impacted stakeholders' perspectives on a deployed ADS that prioritizes scarce housing resources. We employed AI lifecycle comicboarding, an adapted version of the comicboarding me… ▽ More Recent years have seen growing adoption of AI-based decision-support systems (ADS) in homeless services, yet we know little about stakeholder desires and concerns surrounding their use. In this work, we aim to understand impacted stakeholders' perspectives on a deployed ADS that prioritizes scarce housing resources. We employed AI lifecycle comicboarding, an adapted version of the comicboarding method, to elicit stakeholder feedback and design ideas across various components of an AI system's design. We elicited feedback from county workers who operate the ADS daily, service providers whose work is directly impacted by the ADS, and unhoused individuals in the region. Our participants shared concerns and design suggestions around the AI system's overall objective, specific model design choices, dataset selection, and use in deployment. Our findings demonstrate that stakeholders, even without AI knowledge, can provide specific and critical feedback on an AI system's design and deployment, if empowered to do so. △ Less

Submitted 16 March, 2023; originally announced March 2023.

Journal ref: Proceedings of the 2023 CHI Conference on Human Factors in Computing Systems (CHI '23)

Effects of metallic electrodes on the thermoelectric properties of zigzag graphene nanoribbons with periodic vacancies

Authors: David M T Kuo

Abstract: We theoretically analyze the thermoelectric properties of graphene quantum dot arrays (GQDAs) with line- or surface-contacted metal electrodes. Such GQDAs are realized as zigzag graphene nanoribbons (ZGNRs) with periodic vacancies. Gaps and minibands are formed in these GQDAs, which can have metallic and semiconducting phases. The electronic states of the first conduction (valence) miniband with n… ▽ More We theoretically analyze the thermoelectric properties of graphene quantum dot arrays (GQDAs) with line- or surface-contacted metal electrodes. Such GQDAs are realized as zigzag graphene nanoribbons (ZGNRs) with periodic vacancies. Gaps and minibands are formed in these GQDAs, which can have metallic and semiconducting phases. The electronic states of the first conduction (valence) miniband with nonlinear dispersion may have long coherent lengths along the zigzag edge direction. With line-contacted metal electrodes, the GQDAs have the characteristics of serially coupled quantum dots (SCQDs) if the armchair edge atoms of the ZGNRs are coupled to the electrodes. By contrast, the GQDAs have the characteristics of parallel QDs if the zigzag edge atoms are coupled to the electrodes. The maximum thermoelectric power factors of SCQDs with line-contacted electrodes of Cu, Au, Pt, Pd, or Ti at room temperature were similar or greater than $0.186~nW/K$; their figures of merit were greater than three. GQDAs with line-contacted metal electrodes have much better thermoelectric performance than surface contacted metal electrodes. △ Less

Submitted 5 May, 2023; v1 submitted 4 December, 2022; originally announced December 2022.

Comments: 9 pages, 12 figures and improve Fig. 2

Improving Federated Learning Communication Efficiency with Global Momentum Fusion for Gradient Compression Schemes

Authors: Chun-Chih Kuo, Ted Tsei Kuo, Chia-Yu Lin

Abstract: Communication costs within Federated learning hinder the system scalability for reaching more data from more clients. The proposed FL adopts a hub-and-spoke network topology. All clients communicate through the central server. Hence, reducing communication overheads via techniques such as data compression has been proposed to mitigate this issue. Another challenge of federated learning is unbalanc… ▽ More Communication costs within Federated learning hinder the system scalability for reaching more data from more clients. The proposed FL adopts a hub-and-spoke network topology. All clients communicate through the central server. Hence, reducing communication overheads via techniques such as data compression has been proposed to mitigate this issue. Another challenge of federated learning is unbalanced data distribution, data on each client are not independent and identically distributed (non-IID) in a typical federated learning setting. In this paper, we proposed a new compression compensation scheme called Global Momentum Fusion (GMF) which reduces communication overheads between FL clients and the server and maintains comparable model accuracy in the presence of non-IID data. GitHub repository: https://github.com/tony92151/global-momentum-fusion-fl △ Less

Submitted 16 November, 2022; originally announced November 2022.

Conformal Circles and Local Diffeomorphisms

Authors: Tzu-Mo Kuo

Abstract: We study unparametrized conformal circles, or called conformal geodesics, study diffeomorphisms mapping conformal circles to conformal circles in pseudo-Riemannian conformal manifolds. We show that such local diffeomorphisms are conformal local diffeomorphisms. Our result extends the result of Yano and Tomonaga. We also present a holographic interpretation for our result on Poincaré-Einstein manif… ▽ More We study unparametrized conformal circles, or called conformal geodesics, study diffeomorphisms mapping conformal circles to conformal circles in pseudo-Riemannian conformal manifolds. We show that such local diffeomorphisms are conformal local diffeomorphisms. Our result extends the result of Yano and Tomonaga. We also present a holographic interpretation for our result on Poincaré-Einstein manifolds. The proofs take suitable variations of conformal circles. △ Less

Submitted 12 November, 2023; v1 submitted 11 October, 2022; originally announced October 2022.

Comments: 13 pages

MSC Class: 53

Contact effects on thermoelectric properties of textured graphene nanoribbons

Authors: David M T Kuo, Yia-Chung Chang

Abstract: Transport and thermoelectric properties of finite textured graphene nanoribbons (t-GNRs) connected to electrodes with various coupling strengths are theoretically studied in the framework of the tight-binding model and Green's function approach. Due to quantum constriction induced by the indented edges, such t-GNRs behave like serially-coupled graphene quantum dots (SGQDs). These types of SGQDs ca… ▽ More Transport and thermoelectric properties of finite textured graphene nanoribbons (t-GNRs) connected to electrodes with various coupling strengths are theoretically studied in the framework of the tight-binding model and Green's function approach. Due to quantum constriction induced by the indented edges, such t-GNRs behave like serially-coupled graphene quantum dots (SGQDs). These types of SGQDs can be formed by tailoring zigzag GNRs (ZGNRs) or armchair GNRs (AGNRs). Their bandwidths and gaps can be engineered by varying the size of the quantum dot and the neck width at indented edges. Effects of defects and contact junction on electrical conductance, Seebeck coefficient and electron thermal conductance of t-GNRs are calculated. When a defect occurs in the interior site of textured ZGNRs (t-ZGNRs), the maximum power factor within the central gap or near the band edges is found to be insensitive to the defect scattering. Furthermore, we found that SGQDs formed by t-ZGNRs have significantly better electrical power outputs than those of textured ANGRs due to the improved functional shape of the transmission coefficient in t-ZGNRs. With a proper design of contact the maximum power factor ( figure of merit) of t-ZGNRs could reach $90\%$ ($95\%$) of the theoretical limit. △ Less

Submitted 24 September, 2022; v1 submitted 23 August, 2022; originally announced August 2022.

Comments: 12 pages, 14 figures, typos and new references

DataPerf: Benchmarks for Data-Centric AI Development

Authors: Mark Mazumder, Colby Banbury, Xiaozhe Yao, Bojan Karlaš, William Gaviria Rojas, Sudnya Diamos, Greg Diamos, Lynn He, Alicia Parrish, Hannah Rose Kirk, Jessica Quaye, Charvi Rastogi, Douwe Kiela, David Jurado, David Kanter, Rafael Mosquera, Juan Ciro, Lora Aroyo, Bilge Acun, Lingjiao Chen, Mehul Smriti Raje, Max Bartolo, Sabri Eyuboglu, Amirata Ghorbani, Emmett Goodman , et al. (20 additional authors not shown)

Abstract: Machine learning research has long focused on models rather than datasets, and prominent datasets are used for common ML tasks without regard to the breadth, difficulty, and faithfulness of the underlying problems. Neglecting the fundamental importance of data has given rise to inaccuracy, bias, and fragility in real-world applications, and research is hindered by saturation across existing datase… ▽ More Machine learning research has long focused on models rather than datasets, and prominent datasets are used for common ML tasks without regard to the breadth, difficulty, and faithfulness of the underlying problems. Neglecting the fundamental importance of data has given rise to inaccuracy, bias, and fragility in real-world applications, and research is hindered by saturation across existing dataset benchmarks. In response, we present DataPerf, a community-led benchmark suite for evaluating ML datasets and data-centric algorithms. We aim to foster innovation in data-centric AI through competition, comparability, and reproducibility. We enable the ML community to iterate on datasets, instead of just architectures, and we provide an open, online platform with multiple rounds of challenges to support this iterative development. The first iteration of DataPerf contains five benchmarks covering a wide spectrum of data-centric techniques, tasks, and modalities in vision, speech, acquisition, debugging, and diffusion prompting, and we support hosting new contributed benchmarks from the community. The benchmarks, online evaluation platform, and baseline implementations are open source, and the MLCommons Association will maintain DataPerf to ensure long-term benefits to academia and industry. △ Less

Submitted 13 October, 2023; v1 submitted 20 July, 2022; originally announced July 2022.

Comments: NeurIPS 2023 Datasets and Benchmarks Track

Eigenvector-eigenvalue identities and an application to flavor physics

Authors: S. H. Chiu, T. K. Kuo

Abstract: The eigenvector-eigenvalue identities are expanded to include general mixing parameters. Some simple relations are obtained and they reveal an intricate texture of connections between the eigenvalues and the mixing parameters. Permutation symmetry ($S_{3}\times S_{3}$) plays an indispensable role in our analysis. It is the guiding principle for the understanding of our results -- all of them are t… ▽ More The eigenvector-eigenvalue identities are expanded to include general mixing parameters. Some simple relations are obtained and they reveal an intricate texture of connections between the eigenvalues and the mixing parameters. Permutation symmetry ($S_{3}\times S_{3}$) plays an indispensable role in our analysis. It is the guiding principle for the understanding of our results -- all of them are tensor equations under permutation. △ Less

Submitted 29 May, 2022; v1 submitted 24 May, 2022; originally announced May 2022.

Comments: 11 pages

Journal ref: Eur. Phys. J. C 82, 712 (2022)

NFL: Robust Learned Index via Distribution Transformation

Authors: Shangyu Wu, Yufei Cui, Jinghuan Yu, Xuan Sun, Tei-Wei Kuo, Chun Jason Xue

Abstract: Recent works on learned index open a new direction for the indexing field. The key insight of the learned index is to approximate the mapping between keys and positions with piece-wise linear functions. Such methods require partitioning key space for a better approximation. Although lots of heuristics are proposed to improve the approximation quality, the bottleneck is that the segmentation overhe… ▽ More Recent works on learned index open a new direction for the indexing field. The key insight of the learned index is to approximate the mapping between keys and positions with piece-wise linear functions. Such methods require partitioning key space for a better approximation. Although lots of heuristics are proposed to improve the approximation quality, the bottleneck is that the segmentation overheads could hinder the overall performance. This paper tackles the approximation problem by applying a \textit{distribution transformation} to the keys before constructing the learned index. A two-stage Normalizing-Flow-based Learned index framework (NFL) is proposed, which first transforms the original complex key distribution into a near-uniform distribution, then builds a learned index leveraging the transformed keys. For effective distribution transformation, we propose a Numerical Normalizing Flow (Numerical NF). Based on the characteristics of the transformed keys, we propose a robust After-Flow Learned Index (AFLI). To validate the performance, comprehensive evaluations are conducted on both synthetic and real-world workloads, which shows that the proposed NFL produces the highest throughput and the lowest tail latency compared to the state-of-the-art learned indexes. △ Less

Submitted 24 May, 2022; originally announced May 2022.

Effects of zigzag edge states on the thermoelectric properties of finite graphene nanoribbons

Authors: David Ming Ting Kuo

Abstract: Thermoelectric properties of finite graphene nanoribbons (GNRs) coupled to metallic electrodes are theoretically studied in the framework of tight-binding model and Green's function approach. When the zigzag sides are coupled to the electrodes, the electron transport through the localized edge states can occur only if the channel length between electrodes is smaller than the decay length of these… ▽ More Thermoelectric properties of finite graphene nanoribbons (GNRs) coupled to metallic electrodes are theoretically studied in the framework of tight-binding model and Green's function approach. When the zigzag sides are coupled to the electrodes, the electron transport through the localized edge states can occur only if the channel length between electrodes is smaller than the decay length of these localized zigzag edge states. When the armchair edges are coupled to the electrodes, there is an interesting thermoelectric behavior associated with the mid-gap states when the GNR is in the semiconducting phase. Here we show that the thermoelectric behavior of zigzag edge states of GNRs with armchair sides connected to electrodes is similar to that of two parallel quantum dots with similar orbital degeneracy. Furthermore, it is demonstrated that the electrical conductance and power factor given by the zigzag edge states are quite robust against the defect scattering. △ Less

Submitted 8 July, 2022; v1 submitted 9 April, 2022; originally announced April 2022.

Comments: 9 pages and 13 Figures, corrected typos in Fig. A.1 and Fig. A.2

A Fast Transformer-based General-Purpose Lossless Compressor

Authors: Yu Mao, Yufei Cui, Tei-Wei Kuo, Chun Jason Xue

Abstract: Deep-learning-based compressor has received interests recently due to much improved compression ratio. However, modern approaches suffer from long execution time. To ease this problem, this paper targets on cutting down the execution time of deep-learning-based compressors. Building history-dependencies sequentially (e.g., recurrent neural networks) is responsible for long inference latency. Inste… ▽ More Deep-learning-based compressor has received interests recently due to much improved compression ratio. However, modern approaches suffer from long execution time. To ease this problem, this paper targets on cutting down the execution time of deep-learning-based compressors. Building history-dependencies sequentially (e.g., recurrent neural networks) is responsible for long inference latency. Instead, we introduce transformer into deep learning compressors to build history-dependencies in parallel. However, existing transformer is too heavy in computation and incompatible to compression tasks. This paper proposes a fast general-purpose lossless compressor, TRACE, by designing a compression-friendly structure based on a single-layer transformer. We first design a new metric to advise the selection part of compression model structures. Byte-grouping and Shared-ffn schemes are further proposed to fully utilize the capacity of the single-layer transformer. These features allow TRACE to achieve competitive compression ratio and a much faster speed. In addition, we further accelerate the compression procedure by designing a controller to reduce the parameter updating overhead. Experiments show that TRACE achieves an overall $\sim$3x speedup while keeps a comparable compression ratio to the state-of-the-art compressors. The source code for TRACE and links to the datasets are available at https://github.com/mynotwo/A-Fast-Transformer-based-General-Purpose-LosslessCompressor. △ Less

Submitted 1 April, 2022; v1 submitted 30 March, 2022; originally announced March 2022.

Balancing SRPT and FCFS via Starvation Mitigation

Authors: Tung-Wei Kuo

Abstract: The assessment of a job's Quality of Service (QoS) often revolves around its flow time, also referred to as response time. This study delves into two fundamental objectives for scheduling jobs: the average flow time and the maximum flow time. While the Shortest Remaining Processing Time (SRPT) algorithm minimizes average flow time, it can result in job starvation, causing certain jobs to experienc… ▽ More The assessment of a job's Quality of Service (QoS) often revolves around its flow time, also referred to as response time. This study delves into two fundamental objectives for scheduling jobs: the average flow time and the maximum flow time. While the Shortest Remaining Processing Time (SRPT) algorithm minimizes average flow time, it can result in job starvation, causing certain jobs to experience disproportionately long and unfair flow times. In contrast, the First-Come-First-Served (FCFS) algorithm minimizes the maximum flow time but may compromise the average flow time. To strike a balance between these two objectives, a common approach is to minimize the $\ell_2$ norm of flow time. In this study, we provide theoretical and experimental evidence that a judicious combination of SRPT and FCFS yields superior performance compared to standard heuristics for minimizing the $\ell_2$ norm of flow time. △ Less

Submitted 14 August, 2023; v1 submitted 29 December, 2021; originally announced December 2021.

Comments: New experimental results and proof sketch

SEOFP-NET: Compression and Acceleration of Deep Neural Networks for Speech Enhancement Using Sign-Exponent-Only Floating-Points

Authors: Yu-Chen Lin, Cheng Yu, Yi-Te Hsu, Szu-Wei Fu, Yu Tsao, Tei-Wei Kuo

Abstract: Numerous compression and acceleration strategies have achieved outstanding results on classification tasks in various fields, such as computer vision and speech signal processing. Nevertheless, the same strategies have yielded ungratified performance on regression tasks because the nature between these and classification tasks differs. In this paper, a novel sign-exponent-only floating-point netwo… ▽ More Numerous compression and acceleration strategies have achieved outstanding results on classification tasks in various fields, such as computer vision and speech signal processing. Nevertheless, the same strategies have yielded ungratified performance on regression tasks because the nature between these and classification tasks differs. In this paper, a novel sign-exponent-only floating-point network (SEOFP-NET) technique is proposed to compress the model size and accelerate the inference time for speech enhancement, a regression task of speech signal processing. The proposed method compressed the sizes of deep neural network (DNN)-based speech enhancement models by quantizing the fraction bits of single-precision floating-point parameters during training. Before inference implementation, all parameters in the trained SEOFP-NET model are slightly adjusted to accelerate the inference time by replacing the floating-point multiplier with an integer-adder. For generalization, the SEOFP-NET technique is introduced to different speech enhancement tasks in speech signal processing with different model architectures under various corpora. The experimental results indicate that the size of SEOFP-NET models can be significantly compressed by up to 81.249% without noticeably downgrading their speech enhancement performance, and the inference time can be accelerated to 1.212x compared with the baseline models. The results also verify that the proposed SEOFP-NET can cooperate with other efficiency strategies to achieve a synergy effect for model compression. In addition, the just noticeable difference (JND) was applied to the user study experiment to statistically analyze the effect of speech enhancement on listening. The results indicate that the listeners cannot facilely differentiate between the enhanced speech signals processed by the baseline model and the proposed SEOFP-NET. △ Less

Submitted 8 November, 2021; originally announced November 2021.

Speech Recovery for Real-World Self-powered Intermittent Devices

Authors: Yu-Chen Lin, Tsun-An Hsieh, Kuo-Hsuan Hung, Cheng Yu, Harinath Garudadri, Yu Tsao, Tei-Wei Kuo

Abstract: The incompleteness of speech inputs severely degrades the performance of all the related speech signal processing applications. Although many researches have been proposed to address this issue, they controlled the data missing conditions by simulation with self-defined masking lengths or sizes. Besides, the masking definitions are different among all these experimental settings. This paper presen… ▽ More The incompleteness of speech inputs severely degrades the performance of all the related speech signal processing applications. Although many researches have been proposed to address this issue, they controlled the data missing conditions by simulation with self-defined masking lengths or sizes. Besides, the masking definitions are different among all these experimental settings. This paper presents a novel intermittent speech recovery (ISR) system for real-world self-powered intermittent devices. Three contributive stages: interpolation, enhancement, and combination are applied to the ISR system for speech reconstruction. The experimental results show that our recovery system increases speech quality by up to 591.7%, while increasing speech intelligibility by up to 80.5%. Most importantly, the proposed ISR system improves the WER scores by up to 52.6%. The promising results not only confirm the effectiveness of the reconstruction but also encourage the utilization of these battery-free wearable/IoT devices. △ Less

Submitted 24 January, 2022; v1 submitted 9 June, 2021; originally announced June 2021.

Proof of a conjecture of Dahmen and Beukers on counting integral Lamé equations with finite monodromy

Authors: Zhijie Chen, Ting-Jung Kuo, Chang-Shou Lin

Abstract: In this paper, we prove Dahmen and Beukers' conjecture that the number of integral Lamé equations with index $n$ modulo scalar equivalence with the monodromy group dihedral $D_{N}$ of order $2N$ is given by \[L_{n}(N)=\frac{1}{2}\left( \frac{n(n+1)Ψ(N)}{24}-\left( a_{n}% φ(N)+b_{n}φ(\tfrac{N}{2}) \right) \right) +\frac{2}% {3}\varepsilon_{n}(N).\] Our main tool is the new pre-modular form… ▽ More In this paper, we prove Dahmen and Beukers' conjecture that the number of integral Lamé equations with index $n$ modulo scalar equivalence with the monodromy group dihedral $D_{N}$ of order $2N$ is given by \[L_{n}(N)=\frac{1}{2}\left( \frac{n(n+1)Ψ(N)}{24}-\left( a_{n}% φ(N)+b_{n}φ(\tfrac{N}{2}) \right) \right) +\frac{2}% {3}\varepsilon_{n}(N).\] Our main tool is the new pre-modular form $Z_{r,s}^{(n)}(τ)$ of weight $n(n+1)/2$ introduced by Lin and Wang \cite{LW2} and the associated modular form $M_{n,N}(τ):=\prod_{(r,s)}Z_{r,s}^{(n)}(τ)$ of weight $Ψ(N)n(n+1)/{2}$, where the product runs over all $N$-torsion points $(r,s)$ of exact order $N$. We show that this conjecture is equivalent to the precise formula of the vanishing order of $M_{n,N}(τ)$ at infinity: \[v_{\infty}(M_{n,N}(τ))=a_{n}φ(N)+b_{n}φ( N/2).\] This formula is extremely hard to prove because the explicit expression of $Z_{r,s}^{(n)}(τ)$ is not known for general $n$. Here we succeed to prove it by using certain Painlevé VI equations. Our result also indicates that this conjecture is intimately connected with the problem of counting pole numbers of algebraic solutions of certain Painlevé VI equations. The main results of this paper has been announced in \cite{Lin-CDM}. △ Less

Submitted 10 May, 2021; originally announced May 2021.

Comments: 53 pages

High thermoelectric figure of merit of quantum dot array quantum wires

Authors: David M T Kuo

Abstract: How to design silicon-based quantum wires with figure of merit ($ZT$) larger than three is under hot pursuit due to the advantage of low cost and the availability of matured fabrication technique. Quantum wires consisting of finite three dimensional quantum dot (QD) arrays coupled to electrodes are proposed to realize high efficient thermoelectric devices with optimized power factors. The transmis… ▽ More How to design silicon-based quantum wires with figure of merit ($ZT$) larger than three is under hot pursuit due to the advantage of low cost and the availability of matured fabrication technique. Quantum wires consisting of finite three dimensional quantum dot (QD) arrays coupled to electrodes are proposed to realize high efficient thermoelectric devices with optimized power factors. The transmission coefficient of 3D QD arrays can exhibit 3D, 2D, 1D and 0D topological distribution functions by tailoring the interdot coupling strengths. Such topological effects on the thermoelectric properties are revealed. The 1D topological distribution function shows the maximum power factor and the best $ZT$ value. We have demonstrated that 3D silicon QD array nanowires with diameters below $20~nm$ and length $250~nm$ show high potential to achieve $ZT\ge 3$ near room temperature. △ Less

Submitted 2 June, 2021; v1 submitted 29 April, 2021; originally announced April 2021.

Comments: 5 pages, 6 figures

Variational Nested Dropout

Authors: Yufei Cui, Yu Mao, Ziquan Liu, Qiao Li, Antoni B. Chan, Xue Liu, Tei-Wei Kuo, Chun Jason Xue

Abstract: Nested dropout is a variant of dropout operation that is able to order network parameters or features based on the pre-defined importance during training. It has been explored for: I. Constructing nested nets: the nested nets are neural networks whose architectures can be adjusted instantly during testing time, e.g., based on computational constraints. The nested dropout implicitly ranks the netwo… ▽ More Nested dropout is a variant of dropout operation that is able to order network parameters or features based on the pre-defined importance during training. It has been explored for: I. Constructing nested nets: the nested nets are neural networks whose architectures can be adjusted instantly during testing time, e.g., based on computational constraints. The nested dropout implicitly ranks the network parameters, generating a set of sub-networks such that any smaller sub-network forms the basis of a larger one. II. Learning ordered representation: the nested dropout applied to the latent representation of a generative model (e.g., auto-encoder) ranks the features, enforcing explicit order of the dense representation over dimensions. However, the dropout rate is fixed as a hyper-parameter during the whole training process. For nested nets, when network parameters are removed, the performance decays in a human-specified trajectory rather than in a trajectory learned from data. For generative models, the importance of features is specified as a constant vector, restraining the flexibility of representation learning. To address the problem, we focus on the probabilistic counterpart of the nested dropout. We propose a variational nested dropout (VND) operation that draws samples of multi-dimensional ordered masks at a low cost, providing useful gradients to the parameters of nested dropout. Based on this approach, we design a Bayesian nested neural network that learns the order knowledge of the parameter distributions. We further exploit the VND under different generative models for learning ordered latent distributions. In experiments, we show that the proposed approach outperforms the nested network in terms of accuracy, calibration, and out-of-domain detection in classification tasks. It also outperforms the related generative models on data generation tasks. △ Less

Submitted 17 June, 2022; v1 submitted 27 January, 2021; originally announced January 2021.

Comments: 20 pages, 17 figures

Linear magnetoresistance with a universal energy scale in a strong-coupling superconductor

Authors: W. Zhang, Y. J. Hu, C. N. Kuo, S. T. Kuo, Yue-Wen Fang, Kwing To Lai, X. Y. Liu, K. Y. Yip, D. Sun, F. F. Balakirev, C. S. Lue, Hanghui Chen, Swee K. Goh

Abstract: The recent discovery of a nonsaturating linear magnetoresistance in several correlated electron systems near a quantum critical point has revealed an interesting interplay between the linear magnetoresistance and the zero-field linear-in-temperature resistivity. These studies suggest a possible role of quantum criticality on the observed linear magnetoresistance. Here, we report our discovery of a… ▽ More The recent discovery of a nonsaturating linear magnetoresistance in several correlated electron systems near a quantum critical point has revealed an interesting interplay between the linear magnetoresistance and the zero-field linear-in-temperature resistivity. These studies suggest a possible role of quantum criticality on the observed linear magnetoresistance. Here, we report our discovery of a nonsaturating, linear magnetoresistance in Mo$_8$Ga$_{41}$, a nearly isotropic strong electron-phonon coupling superconductor with a linear-in-temperature resistivity from the transition temperature to $\sim$55 K. The growth of the resistivity in field is comparable to that in temperature, provided that both quantities are measured in the energy unit. Our datasets are remarkably similar to magnetoresistance data of the optimally doped La$_{2-x}$Sr$_x$CuO$_4$, despite the clearly different crystal and electronic structures, and the apparent absence of quantum critical physics in Mo$_8$Ga$_{41}$. A new empirical scaling formula is developed, which is able to capture the key features of the low-temperature magnetoresistance data of Mo$_8$Ga$_{41}$, as well as the data of La$_{2-x}$Sr$_x$CuO$_4$. △ Less

Submitted 7 October, 2020; originally announced October 2020.

Comments: 7 pages, 3 figures

Journal ref: Phys. Rev. B 102, 241113 (2020)

Thermoelectric properties of finite two dimensional triangular lattices coupled to electrodes

Authors: David M T Kuo

Abstract: Novel intrinsic two-dimensional materials have attracted many researchers' attention. The unusual transport and optical properties of these materials originate mainly from triangular lattices (TLs). Therefore, the application of energy harvesting calls for a study of the thermoelectric properties of 2D TLs coupled to electrodes. The transmission coefficient of 2D TLs is calculated by using the Gre… ▽ More Novel intrinsic two-dimensional materials have attracted many researchers' attention. The unusual transport and optical properties of these materials originate mainly from triangular lattices (TLs). Therefore, the application of energy harvesting calls for a study of the thermoelectric properties of 2D TLs coupled to electrodes. The transmission coefficient of 2D TLs is calculated by using the Green's function technique to treat ballistic transports. Especially important among our findings is the electron-hole asymmetric behavior of the power factor ($PF$). Specifically, the maximum $PF$ of electrons is significantly larger than that of holes. At room temperature, the maximum $PF$ of electrons is dictated by the position of the chemical potential of electrodes near the band edge of TLs. The enhancement of $PF$ with increasing electronic states results from the enhancement of electrical conductance and constant Seebeck coefficient. When the band gap is ten times larger than the thermal energy, it is appropriate to make one-band model predictions for thermoelectric optimization. △ Less

Submitted 24 September, 2020; originally announced September 2020.

Comments: 6 pages, 7 figures

The geometry of generalized Lame equation, III: One-to-one of the Riemann-Hilbert correspondence

Authors: Zhijie Chen, Ting-Jung Kuo, Chang-Shou Lin

Abstract: In this paper, the third in a series, we continue to study the generalized Lamé equation H$(n_0,n_1,n_2,n_3;B)$ with the Darboux-Treibich-Verdier potential \begin{equation*} y^{\prime \prime }(z)=\bigg[ \sum_{k=0}^{3}n_{k}(n_{k}+1)\wp(z+\tfrac{ ω_{k}}{2}|τ)+B\bigg] y(z),\quad n_{k}\in \mathbb{Z}_{\geq0} \end{equation*} and a related linear ODE with additional singularities $\pm p$ from the monodro… ▽ More In this paper, the third in a series, we continue to study the generalized Lamé equation H$(n_0,n_1,n_2,n_3;B)$ with the Darboux-Treibich-Verdier potential \begin{equation*} y^{\prime \prime }(z)=\bigg[ \sum_{k=0}^{3}n_{k}(n_{k}+1)\wp(z+\tfrac{ ω_{k}}{2}|τ)+B\bigg] y(z),\quad n_{k}\in \mathbb{Z}_{\geq0} \end{equation*} and a related linear ODE with additional singularities $\pm p$ from the monodromy aspect.We establish the uniqueness of these ODEs with respect to the global monodromy data. Surprisingly, our result shows that the Riemann-Hilbert correspondence from the set \[\{\text{H}(n_0,n_1,n_2,n_3;B)|B\in\mathbb{C}\}\cup \{\text{H}(n_0+2,n_1,n_2,n_3;B) | B\in\mathbb{C}\}\] to the set of group representations $ρ:π_1(E_τ)\to SL(2,\mathbb{C})$ is one-to-one. We emphasize that this result is not trivial at all. There is an example that for $τ=\frac12+i\frac{\sqrt{3}}{2}$, there are $B_1,B_2$ such that the monodromy representations of H$(1,0,0,0;B_1)$ and H$(4,0,0,0;B_2)$ are {\bf the same}, namely the Riemann-Hilbert correspondence from the set \[\{\text{H}(n_0,n_1,n_2,n_3;B)|B\in\mathbb{C}\}\cup \{\text{H}(n_0+3,n_1,n_2,n_3;B) | B\in\mathbb{C}\}\] to the set of group representations is {\bf not} necessarily one-to-one. This example shows that our result is completely different from the classical one concerning linear ODEs defined on $\mathbb{CP}^1$ with finite singularities. △ Less

Submitted 2 September, 2020; originally announced September 2020.

Comments: 41 pages

Thermoelectric properties of finite two-dimensional quantum dot arrays with band-like electronic states

Authors: David M T Kuo

Abstract: The thermal power ($PF=S^2G_e$) depends on the Seebeck coefficient ($S$) and electron conductance ($G_e$). The enhancement of $G_e$ will unavoidably suppress $S$ because they are closely related. As a consequence, the optimization of $PF$ is extremely difficult. Here, we theoretically investigated the thermoelectric properties of two-dimensional quantum dot (QD) arrays with carriers injected from… ▽ More The thermal power ($PF=S^2G_e$) depends on the Seebeck coefficient ($S$) and electron conductance ($G_e$). The enhancement of $G_e$ will unavoidably suppress $S$ because they are closely related. As a consequence, the optimization of $PF$ is extremely difficult. Here, we theoretically investigated the thermoelectric properties of two-dimensional quantum dot (QD) arrays with carriers injected from electrodes. The Lorenz number of 2D QD arrays in the resonant tunneling procedure satisfies the Wiedemann-Franz law, which confirms the formation of minibands. When the miniband center is far away from the Fermi level of the electrodes, the electron transport is in the thermionic-assisted tunneling procedure (TATP). In this regime, $G_e$ in band-like situation and $S$ in atom-like situation can happen simultaneously. We have demonstrated that the enhancement of $G_e$ with an increasing number of electronic states will not suppress $S$ in the TATP. △ Less

Submitted 24 May, 2020; originally announced May 2020.

Comments: 5 pages, 6 figures

Blow up at infinity in the SU(3) Chern-Simons model, part I

Authors: Ting-Jung Kuo, Youngae Lee, Chang-Shou Lin

Abstract: We consider non-topological solutions of a nonlinear elliptic system problem derived from the $SU(3)$ Chern-Simons models in $\mathbb{R}^2$. The existence of non-topological solutions even for radial symmetric case has been a long standing open problem. Recently, [Choe, Kim, Lin (2015, 2016)] showed the existence of radial symmetric non-topological solution when the vortex points collapse. However… ▽ More We consider non-topological solutions of a nonlinear elliptic system problem derived from the $SU(3)$ Chern-Simons models in $\mathbb{R}^2$. The existence of non-topological solutions even for radial symmetric case has been a long standing open problem. Recently, [Choe, Kim, Lin (2015, 2016)] showed the existence of radial symmetric non-topological solution when the vortex points collapse. However, the arguments in [Choe, Kim, Lin (2015, 2016)] cannot work for an arbitrary configuration of vortex points. In this paper, we develop a new approach by using different scalings for different components of the system to construct a family of non-topological solutions, which blows up at infinity. △ Less

Submitted 30 April, 2020; originally announced April 2020.

Efficient attention guided 5G power amplifier digital predistortion

Authors: Alexandru Cioba, Alvin Chua, Da-shan Shiu, Ting-Hsun Kuo, Chia-Sheng Peng

Abstract: We investigate neural network (NN) assisted techniques for compensating the non-linear behaviour and the memory effect of a 5G PA through digital predistortion (DPD). Traditionally, the most prevalent compensation technique computes the compensation element using a Memory Polynomial Model (MPM). Various neural network proposals have been shown to improve on this performance. However, thus far they… ▽ More We investigate neural network (NN) assisted techniques for compensating the non-linear behaviour and the memory effect of a 5G PA through digital predistortion (DPD). Traditionally, the most prevalent compensation technique computes the compensation element using a Memory Polynomial Model (MPM). Various neural network proposals have been shown to improve on this performance. However, thus far they mostly come with prohibitive training or inference costs for real world implementations. In this paper, we propose a DPD architecture that builds upon the practical MPM formulation governed by neural attention. Our approach enables a set of MPM DPD components to individually learn to target different regions of the data space, combining their outputs for a superior overall compensation. Our method produces similar performance to that of higher capacity NN models with minimal complexity. Finally, we view our approach as a framework that can be extended to a wide variety of local compensator types. △ Less

Submitted 30 March, 2020; originally announced March 2020.

Thermoelectric and electron heat rectification properties of quantum dot superlattice nanowire arrays

Authors: David M T Kuo

Abstract: Heat engines made of quantum-dot (QD) superlattice nanowires (SLNWs) offer promising applications in energy harvesting due to the reduction of phonon thermal conductivity. In solid state electrical generators (refrigerators), one needs to generate (remove) large amount of charge current (heat current). Consequently, a high QD SLNW density is required for realistic applications. This study theoreti… ▽ More Heat engines made of quantum-dot (QD) superlattice nanowires (SLNWs) offer promising applications in energy harvesting due to the reduction of phonon thermal conductivity. In solid state electrical generators (refrigerators), one needs to generate (remove) large amount of charge current (heat current). Consequently, a high QD SLNW density is required for realistic applications. This study theoretically investigated the properties of power factor and electron heat rectification for an SLNW array under the transition from a one dimensional system to a two dimensional system. The SLNW arrays show the functionality of heat diodes, which is mainly attributed to a transmission coefficient with a temperature-bias direction dependent characteristic. △ Less

Submitted 29 March, 2020; originally announced March 2020.

Comments: 8 pages, 11 figures

No Need for Recovery: A Simple Two-Step Byzantine Consensus

Authors: Tung-Wei Kuo, Kung Chen

Abstract: In this paper, we give a deterministic two-step Byzantine consensus protocol that achieves safety and liveness. A two-step Byzantine consensus protocol only needs two communication steps to commit in the absence of faults. Most two-step Byzantine consensus protocols exploit optimism and require a recovery protocol in the presence of faults. In this paper, we give a simple two-step Byzantine consen… ▽ More In this paper, we give a deterministic two-step Byzantine consensus protocol that achieves safety and liveness. A two-step Byzantine consensus protocol only needs two communication steps to commit in the absence of faults. Most two-step Byzantine consensus protocols exploit optimism and require a recovery protocol in the presence of faults. In this paper, we give a simple two-step Byzantine consensus protocol that does not need a recovery protocol. △ Less

Submitted 23 November, 2019; originally announced November 2019.

Moiré lattice-induced formation and tuning of hybrid dipolar excitons in twisted WS$_2$/MoSe$_2$ heterobilayers

Authors: Long Zhang, Zhe Zhang, Fengcheng Wu, Danqing Wang, Rahul Gogna, Shaocong Hou, Kenji Watanabe, Takashi Taniguchi, Krishnamurthy Kulkarni, Thomas Kuo, Stephen R. Forrest, Hui Deng

Abstract: Moiré superlattices formed in van der Waals bilayers have enabled the creation and manipulation of new quantum states, as is exemplified by the discovery of superconducting and correlated insulating states in twisted bilayer graphene near the magic angle. Twisted bilayer semiconductors may lead to tunable exciton lattices and topological states, yet signatures of moiré excitons have been reported… ▽ More Moiré superlattices formed in van der Waals bilayers have enabled the creation and manipulation of new quantum states, as is exemplified by the discovery of superconducting and correlated insulating states in twisted bilayer graphene near the magic angle. Twisted bilayer semiconductors may lead to tunable exciton lattices and topological states, yet signatures of moiré excitons have been reported only in closely angularly-aligned bilayers. Here we report tuning of moiré lattice in WS$_{2}$ /MoSe$_{2}$ bilayers over a wide range of twist angles, leading to the continuous tuning of moiré lattice induced interlayer excitons and their hybridization with optically bright intralayer excitons. A pronounced revival of the hybrid excitons takes place near commensurate twist angles, 21.8°and 38.2°, due to interlayer tunneling between states connected by a moiré reciprocal lattice vector. From the angle dependence, we obtain the effective mass of the interlayer excitons and the electron inter-layer tunneling strength. These findings pave the way for understanding and engineering rich moiré-lattice induced phenomena in angle-twisted semiconductor van dar Waals heterostructures. △ Less

Submitted 22 November, 2019; originally announced November 2019.

Rephasing invariance and permutation symmetry in flavor physics

Authors: T. K. Kuo, S. H. Chiu

Abstract: With some modifications, the arguments for rephasing invariance can be used to establish permutation symmetry for the standard model. The laws of evolution of physical variables, which transform as tensors under permutation, are found to obey the symmetry, explicitly. We also propose to use a set of four mixing parameters, with unique properties, which may serve to characterize flavor mixing. With some modifications, the arguments for rephasing invariance can be used to establish permutation symmetry for the standard model. The laws of evolution of physical variables, which transform as tensors under permutation, are found to obey the symmetry, explicitly. We also propose to use a set of four mixing parameters, with unique properties, which may serve to characterize flavor mixing. △ Less

Submitted 16 September, 2020; v1 submitted 16 October, 2019; originally announced October 2019.

Comments: 15 pages, 2 figures

Journal ref: Eur. Phys. J. C 80, 203 (2020)

Minimum Age TDMA Scheduling

Authors: Tung-Wei Kuo

Abstract: We consider a transmission scheduling problem in which multiple systems receive update information through a shared Time Division Multiple Access (TDMA) channel. To provide timely delivery of update information, the problem asks for a schedule that minimizes the overall age of information. We call this problem the Min-Age problem. This problem is first studied by He \textit{et al.} [IEEE Trans. In… ▽ More We consider a transmission scheduling problem in which multiple systems receive update information through a shared Time Division Multiple Access (TDMA) channel. To provide timely delivery of update information, the problem asks for a schedule that minimizes the overall age of information. We call this problem the Min-Age problem. This problem is first studied by He \textit{et al.} [IEEE Trans. Inform. Theory, 2018], who identified several special cases where the problem can be solved optimally in polynomial time. Our contribution is threefold. First, we introduce a new job scheduling problem called the Min-WCS problem, and we prove that, for any constant $r \geq 1$, every $r$-approximation algorithm for the Min-WCS problem can be transformed into an $r$-approximation algorithm for the Min-Age problem. Second, we give a randomized 2.733-approximation algorithm and a dynamic-programming-based exact algorithm for the Min-WCS problem. Finally, we prove that the Min-Age problem is NP-hard. △ Less

Submitted 30 May, 2019; v1 submitted 26 May, 2019; originally announced May 2019.

On the CR analogue of Frankel conjecture and a smooth representative of the first Kohn-Rossi cohomology group

Authors: Der-Chen Chang, Shu-Cheng Chang, Ting-Jung Kuo, Chien Lin

Abstract: In this note, we first give a criterion of pseudo-Einstein contact forms and then affirm the CR analogue of Frankel conjecture in a closed, spherical, strictly pseudoconvex CR manifold of nonnegative pseudohermitian curvature on the space of smooth representatives of the first Kohn-Rossi cohomology group. Moreover, we obtain the CR Frankel conjecture in a closed, spherical, strictly pseudoconvex C… ▽ More In this note, we first give a criterion of pseudo-Einstein contact forms and then affirm the CR analogue of Frankel conjecture in a closed, spherical, strictly pseudoconvex CR manifold of nonnegative pseudohermitian curvature on the space of smooth representatives of the first Kohn-Rossi cohomology group. Moreover, we obtain the CR Frankel conjecture in a closed, spherical, strictly pseudoconvex CR manifold with the vanishing first Kohn-Rossi cohomology group. In particular, this conjecture holds in a spherical boundary of the Stein manifold. △ Less

Submitted 23 September, 2019; v1 submitted 20 May, 2019; originally announced May 2019.

Comments: 33 pages for revised version where abstract also been revised

MSC Class: 32V05; 32V20 (Primary); 53C56 (Secondary)

Global existence and convergence for the CR Q-curvature flow in a closed strictly pseudoconvex CR 3-manifold

Authors: Shu-Cheng Chang, Ting-Jung Kuo, Takanari Saotome

Abstract: In this note, we affirm the partial answer to the long open Conjecture which states that any closed embeddable strictly pseudoconvex CR $3$-manifold admits a contact form $θ$ with the vanishing CR $Q$-curvature. More precisely, we deform the contact form according to an CR analogue of $Q$%-curvature flow in a closed strictly pseudoconvex CR $3$-manifold $(M,\ J,[θ_{0}])$ of the vanishing first Che… ▽ More In this note, we affirm the partial answer to the long open Conjecture which states that any closed embeddable strictly pseudoconvex CR $3$-manifold admits a contact form $θ$ with the vanishing CR $Q$-curvature. More precisely, we deform the contact form according to an CR analogue of $Q$%-curvature flow in a closed strictly pseudoconvex CR $3$-manifold $(M,\ J,[θ_{0}])$ of the vanishing first Chern class $c_{1}(T_{1,0}M)$. Suppose that $M$ is embeddable and the CR Paneitz operator $P_{0}$ is nonnegative with kernel consisting of the CR pluriharmonic functions. We show that the solution of CR $Q$-curvature flow exists for all time and has smoothly asymptotic convergence on $M\times \lbrack 0,\infty ).$\ As a consequence, we are able to affirm the Conjecture in a closed strictly pseudoconvex CR $3$-manifold of the vanishing first Chern class and vanishing torsion. △ Less

Submitted 5 July, 2019; v1 submitted 5 May, 2019; originally announced May 2019.

Comments: 30 pages. arXiv admin note: text overlap with arXiv:math/0510494

MSC Class: 32V20 (Primary); 53C44 (Secondary)

Superlattice nanowire heat engines with direction-dependent power output and heat current

Authors: David M T Kuo, Yia-Chung Chang

Abstract: Heat engines (HEs) made of low dimensional structures offer promising applications in energy harvesting due to their reduced phonon thermal conductance. Many efforts have been devoted to the design of HEs made of quantum-dot (QD) superlattice nanowire (SLNW), but only SLNWs with uniform energy levels in QDs were considered. Here we propose a HE made of SLNW with staircase-like QD energy levels. It… ▽ More Heat engines (HEs) made of low dimensional structures offer promising applications in energy harvesting due to their reduced phonon thermal conductance. Many efforts have been devoted to the design of HEs made of quantum-dot (QD) superlattice nanowire (SLNW), but only SLNWs with uniform energy levels in QDs were considered. Here we propose a HE made of SLNW with staircase-like QD energy levels. It is demonstrated that the nonlinear Seebeck effect can lead to significant electron transports for such a nanowire with staircase-like energy levels. The asymmetrical alignment of energy levels of quantum dots embedded in nanowires can be controlled to allow resonant electron transport under forward temperature bias, while they are in off-resonant regime under backward bias. Under such a mechanism,the power output and efficiency of such a SLNW are better than SLNWs with uniform QD energy levels. The SLNW HE has direction-dependent power output and heat current. In addition, the HE has the functionality of a heat diode with impressive negative differential thermal conductance under open circuit condition. △ Less

Submitted 4 August, 2019; v1 submitted 15 April, 2019; originally announced April 2019.

Comments: 6 pages and 5 figures