Search

Preferred axion models are minimal realizations of the Peccei-Quinn solution to the strong CP problem while providing a dark matter candidate. These models invoke new heavy quarks that interact strongly with the Standard Model bringing them into thermal equilibrium in the early Universe. We show that for a number of these models, the heavy quarks will decay after axions have decoupled from the Standard Model thermal bath. As a consequence, any axion products in the decay form a component of dark radiation. This provides the potential to differentiate between preferred axion models through measurements of the number of relativistic degrees of freedom. The most sensitive of which comes from the Planck collaboration's measurements of the Cosmic Microwave Background. We find that existing constraints allow us to rule out regions of parameter space for 40% of preferred axion models., Comment: 22 pages, 3 figures, 1 table

We propose practical ways of differentiating the various (Breit-Wigner, theoretical, and energy-dependent) resonance schemes of unstable particles at lepton colliders. First, the energy-dependent scheme can be distinguished from the other two by fitting the $Z$ lineshape scan and forward-backward asymmetries at LEP and future lepton colliders with the $Z$ mass $m_Z$, decay width $\Gamma_Z$, and coupling strength as fitting parameters. Although the Breit-Wigner and theoretical schemes work equally well, the scheme conversion requires the decay width $\Gamma_Z$ to scale inversely with $m_Z$ rather than the usual linear dependence from theoretical calculation. These contradicting behaviors can be used to distinguish the Breit-Wigner and theoretical schemes by the precision $Z$ measurements with single parameter ($m_Z$) fit at future lepton colliders. For the $WW$ threshold scan, its combination with the precise Fermi constant provides another way of distinguishing the Breit-Wigner and theoretical schemes., Comment: 6 pages, 1 table

In the early universe, strong large scale random flows prevail, and when these flows collide it produces sheets and filaments for the conventional particle dark matter model. However, for wave dark matter, the sheets and filaments are parallelly modulated by fine-scale and large-amplitude interference fringes. These long fine fringes tend to be unstable initially and subsequently deform and collapse to form halos of various sizes, perhaps down to proto-globular cluster scales. This work aims to investigate the instability of such fringes. Two colliding streams of quantum waves result in a standing density wave. By contrast, two colliding beams in classical collisionless fluid systems often give rise to beam instability, characterized by unstable oscillations with a complex frequency. Will the quantum standing density wave be subject to a similar instability? In fact, such unstable oscillations have never been observed in conventional quantum mechanical systems. This study discovers that periodic fringes can indeed undergo unstable oscillations or beam instability. The presence of any local density null in the background density that fails the Madelung formulation is shown to be the necessary condition for unstable oscillations. We establish a phase diagram to distinguish between beam instability, Jeans instability, and stable oscillation through theoretical calculations and computer simulations. Roughly speaking, Jeans instability almost always occurs for long-wave density perturbations; beam instability on the other hand always occurs when the density wave is aligned with the beam direction and moreover the period of the density wave is twice the period of background density; lastly, both instabilities diminish for a very low background density. We have engaged in Blocks theorem extensively in the analysis., Comment: 34 pages, 4 figures, 1 table

As wireless networks evolve towards open architectures like O-RAN, testing, and integration platforms are crucial to address challenges like interoperability. This paper describes ARA-O-RAN, a novel O-RAN testbed established through the NSF Platforms for Advanced Wireless Research (PAWR) ARA platform. ARA provides an at-scale rural wireless living lab focused on technologies for digital agriculture and rural communities. As an O-RAN Alliance certified Open Testing and Integration Centre (OTIC), ARA launched ARA-O-RAN -- the first public O-RAN testbed tailored to rural and agriculture use cases, together with the end-to-end, whole-stack programmability. ARA-O-RAN uniquely combines support for outdoor testing across a university campus, surrounding farmlands, and rural communities with a 50-node indoor sandbox. The testbed facilitates vital R\&D to implement open architectures that can meet rural connectivity needs. The paper outlines ARA-O-RAN's hardware system design, software architecture, and enabled research experiments. It also discusses plans aligned with national spectrum policy and rural spectrum innovation. ARA-O-RAN exemplifies the value of purpose-built wireless testbeds in accelerating impactful wireless research.

In speech separation, time-domain approaches have successfully replaced the time-frequency domain with latent sequence feature from a learnable encoder. Conventionally, the feature is separated into speaker-specific ones at the final stage of the network. Instead, we propose a more intuitive strategy that separates features earlier by expanding the feature sequence to the number of speakers as an extra dimension. To achieve this, an asymmetric strategy is presented in which the encoder and decoder are partitioned to perform distinct processing in separation tasks. The encoder analyzes features, and the output of the encoder is split into the number of speakers to be separated. The separated sequences are then reconstructed by the weight-shared decoder, which also performs cross-speaker processing. Without relying on speaker information, the weight-shared network in the decoder directly learns to discriminate features using a separation objective. In addition, to improve performance, traditional methods have extended the sequence length, leading to the adoption of dual-path models, which handle the much longer sequence effectively by segmenting it into chunks. To address this, we introduce global and local Transformer blocks that can directly handle long sequences more efficiently without chunking and dual-path processing. The experimental results demonstrated that this asymmetric structure is effective and that the combination of proposed global and local Transformer can sufficiently replace the role of inter- and intra-chunk processing in dual-path structure. Finally, the presented model combining both of these achieved state-of-the-art performance with much less computation in various benchmark datasets., Comment: Accepted to NeurIPS 2024; Project Page: https://dmlguq456.github.io/SepReformer_Demo

This paper introduces a novel benchmark as part of the AIS 2024 Real-Time Image Super-Resolution (RTSR) Challenge, which aims to upscale compressed images from 540p to 4K resolution (4x factor) in real-time on commercial GPUs. For this, we use a diverse test set containing a variety of 4K images ranging from digital art to gaming and photography. The images are compressed using the modern AVIF codec, instead of JPEG. All the proposed methods improve PSNR fidelity over Lanczos interpolation, and process images under 10ms. Out of the 160 participants, 25 teams submitted their code and models. The solutions present novel designs tailored for memory-efficiency and runtime on edge devices. This survey describes the best solutions for real-time SR of compressed high-resolution images., Comment: CVPR 2024, AI for Streaming (AIS) Workshop

We introduce HyperCLOVA X, a family of large language models (LLMs) tailored to the Korean language and culture, along with competitive capabilities in English, math, and coding. HyperCLOVA X was trained on a balanced mix of Korean, English, and code data, followed by instruction-tuning with high-quality human-annotated datasets while abiding by strict safety guidelines reflecting our commitment to responsible AI. The model is evaluated across various benchmarks, including comprehensive reasoning, knowledge, commonsense, factuality, coding, math, chatting, instruction-following, and harmlessness, in both Korean and English. HyperCLOVA X exhibits strong reasoning capabilities in Korean backed by a deep understanding of the language and cultural nuances. Further analysis of the inherent bilingual nature and its extension to multilingualism highlights the model's cross-lingual proficiency and strong generalization ability to untargeted languages, including machine translation between several language pairs and cross-lingual inference tasks. We believe that HyperCLOVA X can provide helpful guidance for regions or countries in developing their sovereign LLMs., Comment: 44 pages; updated authors list and fixed author names

The analysis of large population economies with incomplete information often entails the integration of a continuum of random variables. We showcase the usefulness of the integral notion \`a la Pettis (1938) to study such models. We present several results on Pettis integrals, including convenient sufficient conditions for Pettis integrability and Fubini-like exchangeability formulae, illustrated through a running example. Building on these foundations, we conduct a unified analysis of Bayesian games with arbitrarily many heterogeneous agents. We provide a sufficient condition on payoff structures, under which the equilibrium uniqueness is guaranteed across all signal structures. Our condition is parsimonious, as it turns out necessary when strategic interactions are undirected. We further identify the moment restrictions, imposed on the equilibrium action-state joint distribution, which have crucial implications for information designer's problem of persuading a population of strategically interacting agents. To attain these results, we introduce and develop novel mathematical tools, built on the theory of integral kernels and reproducing kernel Hilbert spaces in functional analysis.

In this paper, we first present the character texture generation system \textit{Minecraft-ify}, specified to Minecraft video game toward in-game application. Ours can generate face-focused image for texture mapping tailored to 3D virtual character having cube manifold. While existing projects or works only generate texture, proposed system can inverse the user-provided real image, or generate average/random appearance from learned distribution. Moreover, it can be manipulated with text-guidance using StyleGAN and StyleCLIP. These features provide a more extended user experience with enlarged freedom as a user-friendly AI-tool. Project page can be found at https://gh-bumsookim.github.io/Minecraft-ify/, Comment: 2 pages, 2 figures. Accepted as Spotlight to NeurIPS 2023 Workshop on Machine Learning for Creativity and Design

This paper addresses information design with linear best responses of agents, quadratic objective functions of an information designer, and a payoff state distributed according to a Gaussian distribution. We formulate the problem as semidefinite programming (SDP) and use the duality principle to characterize an optimal information structure. There exists a Gaussian information structure that is optimal among all information structures. A necessary and sufficient condition for optimality is that the realizations of the induced action profile and a state satisfy linear constraints derived from the primal and dual SDP. As a result, an observed action profile typically reveals the true state even if individual agents have only partial knowledge. In symmetric network games, an optimal information structure inherits this symmetry, which facilitates the computation of an optimal information structure.

In speaker verification, ECAPA-TDNN has shown remarkable improvement by utilizing one-dimensional(1D) Res2Net block and squeeze-and-excitation(SE) module, along with multi-layer feature aggregation (MFA). Meanwhile, in vision tasks, ConvNet structures have been modernized by referring to Transformer, resulting in improved performance. In this paper, we present an improved block design for TDNN in speaker verification. Inspired by recent ConvNet structures, we replace the SE-Res2Net block in ECAPA-TDNN with a novel 1D two-step multi-scale ConvNeXt block, which we call TS-ConvNeXt. The TS-ConvNeXt block is constructed using two separated sub-modules: a temporal multi-scale convolution (MSC) and a frame-wise feed-forward network (FFN). This two-step design allows for flexible capturing of inter-frame and intra-frame contexts. Additionally, we introduce global response normalization (GRN) for the FFN modules to enable more selective feature propagation, similar to the SE module in ECAPA-TDNN. Experimental results demonstrate that NeXt-TDNN, with a modernized backbone block, significantly improved performance in speaker verification tasks while reducing parameter size and inference time. We have released our code for future studies., Comment: Accepted by ICASSP 2024

Axions and dark photons are common in many extensions of the Standard Model. The dark axion portal -- an axion coupling to the dark photon and photon -- can significantly modify their phenomenology. We study the cosmological constraints on the dark axion portal from Cosmic Microwave Background (CMB) bounds on the energy density of dark radiation, $\Delta N_\text{eff}$. By computing the axion-photon-dark photon collision terms and solving the Boltzmann equations including their effects, we find that light axions are generally more constrained by $\Delta N_\text{eff}$ than from supernova cooling or collider experiments. However, with dark photons at the MeV scale, a window of parameter space is opened up above the supernova limits and below the experimental exclusion, allowing for axion decay constants as low as $f_a \sim 10^4$ GeV. This region also modifies indirectly the neutrino energy density, thus relaxing the cosmological upper bound on the sum of neutrino masses. Future CMB measurements could detect a signal or close this open window on the dark axion portal., Comment: 31 pages, 13 figures. v2: Added section on BBN constraints and extended discussions, figures, and references. Version accepted for publication in JHEP

This work studies the net sum-rate performance of a distributed reconfigurable intelligent surfaces (RISs)-assisted multi-user multiple-input-single-output (MISO) downlink communication system under imperfect instantaneous-channel state information (I-CSI) to implement precoding at the base station (BS) and statistical-CSI (S-CSI) to design the RISs phase-shifts. Two channel estimation (CE) protocols are considered for I-CSI acquisition: (i) a full CE protocol that estimates all direct and RISs-assisted channels over multiple training sub-phases, and (ii) a low-overhead direct estimation (DE) protocol that estimates the end-to-end channel in a single sub-phase. We derive the asymptotic equivalents of signal-to-interference-plus-noise ratio (SINR) and ergodic net sum-rate under both protocols for given RISs phase-shifts, which are then optimized based on S-CSI. Simulation results reveal that the low-complexity DE protocol yields better net sum-rate than the full CE protocol when used to obtain CSI for precoding. A benchmark full I-CSI based RISs design is also outlined and shown to yield higher SINR but lower net sum-rate than the S-CSI based RISs design.