Search

In 1996, in his last paper, Erd\H{o}s asked the following question that he formulated together with Faudree: is there a positive $c$ such that any $(n+1)$-regular graph $G$ on $2n$ vertices contains at least $c 2^{2n}$ distinct vertex-subsets $S$ that are cyclic, meaning that there is a cycle in $G$ using precisely the vertices in $S$. We answer this question in the affirmative in a strong form by proving the following exact result: if $n$ is sufficiently large and $G$ minimises the number of cyclic subsets then $G$ is obtained from the complete bipartite graph $K_{n-1,n+1}$ by adding a $2$-factor (a spanning collection of vertex-disjoint cycles) within the part of size $n+1$. In particular, for $n$ large, this implies that the optimal $c$ in the problem is precisely $1/2$., Comment: 17 pages

We study an old question in combinatorial group theory which can be traced back to a conjecture of Graham from 1971. Given a group $\Gamma$, and some subset $S\subseteq \Gamma$, is it possible to permute $S$ as $s_1, s_2, \ldots, s_d$ so that the partial products $\prod_{1 \leq i \leq t} s_i$, $t\in [d]$ are all distinct? Most of the progress towards this problem has been in the case when $\Gamma$ is a cyclic group. We show that for any group $\Gamma$ and any $S \subseteq \Gamma$, there is a permutation of $S$ where all but a vanishing proportion of the partial products are distinct, thereby establishing the first asymptotic version of Graham's conjecture under no restrictions on $\Gamma$ or $S$. To do so, we explore a natural connection between Graham's problem and the following very natural question attributed to Schrijver. Given a $d$-regular graph $G$ properly edge-coloured with $d$ colours, is it always possible to find a rainbow path with $d-1$ edges? We settle this question asymptotically by showing one can find a rainbow path of length $d - o(d)$. While this has immediate applications to Graham's question for example when $\Gamma = \mathbb{F}_2^k$, our general result above requires a more involved result we obtain for the natural directed analogue of Schrijver's question., Comment: 25 pages

This paper presents performance measurements of a new readout electronics system based on silicon photomultipliers for the PHENIX electromagnetic calorimeter. Installation of the lead-scintillator Shashlik style calorimeter into the SeaQuest/SpinQuest spectrometer has been proposed to broaden the experiment's dark sector search program, an upgrade known as DarkQuest. The calorimeter and electronics system were subjected to testing and calibration at the Fermilab Test Beam Facility. Detailed studies of the energy response and resolution, as well as particle identification capabilities, were performed. The background rate in the actual experimental environment was also examined. The system is found to be well-suited for a dark sector search program on the Fermilab 120 GeV proton beamline., Comment: Prepared for submission to Nuclear Instrumentation and Methods

We consider a cell-free massive multiple-input multiple-output (mMIMO) network, where unmanned aerial vehicles (UAVs) equipped with multiple antennas serve as distributed UAV-access points (UAV-APs). These UAV-APs provide seamless coverage by jointly serving user equipments (UEs) with out predefined cell boundaries. However, high-capacity wireless networks face significant challenges due to fronthaul limitations in UAV-assisted architectures. This letter proposes a novel UAV-based cell-free mMIMO framework that leverages distributed UAV-APs to serve UEs while addressing the capacity constraints of wireless fronthaul links. We evaluate functional split Options 7.2 and 8 for the fronthaul links, aiming to maximize the minimum signal-to-interference-plus-noise ratio (SINR) among the UEs and minimize the power consumption by optimizing the transmit powers of UAV-APs and selectively activating them. Our analysis compares sub-6 GHz and millimeter wave (mmWave) bands for the fronthaul, showing that mmWave achieves superior SINR with lower power consumption, particularly under Option 8. Additionally, we determine the minimum fronthaul bandwidth required to activate a single UAV-AP under different split options.

Entanglement distribution is central to the modular scaling of quantum processors and establishing quantum networks. Color centers with telecom-band transitions and long spin coherence times are suitable candidates for long-distance entanglement distribution. However, high-bandwidth memory-enhanced quantum communication is limited by high-yield, scalable creation of efficient spin-photon interfaces. Here, we develop a silicon photonics platform consisting of arrays of bus-coupled cavities. The coupling to a common bus waveguide enables simultaneous access to individually addressable cavity-enhanced T center arrays. We demonstrate frequency-multiplexed operation of two T centers in separate photonic crystal cavities. In addition, we investigate the cavity enhancement of a T center through hybridized modes formed between physically distant cavities. Our results show that bus-coupled arrays of cavity-enhanced color centers could enable efficient on-chip and long-distance entanglement distribution.

Gradient descent for matrix factorization is known to exhibit an implicit bias toward approximately low-rank solutions. While existing theories often assume the boundedness of iterates, empirically the bias persists even with unbounded sequences. We thus hypothesize that implicit bias is driven by divergent dynamics markedly different from the convergent dynamics for data fitting. Using this perspective, we introduce a new factorization model: $X\approx UDV^\top$, where $U$ and $V$ are constrained within norm balls, while $D$ is a diagonal factor allowing the model to span the entire search space. Our experiments reveal that this model exhibits a strong implicit bias regardless of initialization and step size, yielding truly (rather than approximately) low-rank solutions. Furthermore, drawing parallels between matrix factorization and neural networks, we propose a novel neural network model featuring constrained layers and diagonal components. This model achieves strong performance across various regression and classification tasks while finding low-rank solutions, resulting in efficient and lightweight networks.

Alzheimer's disease (AD) is a neurodegenerative disorder affecting millions worldwide, necessitating early and accurate diagnosis for optimal patient management. In recent years, advancements in deep learning have shown remarkable potential in medical image analysis. Methods In this study, we present "ViTranZheimer," an AD diagnosis approach which leverages video vision transformers to analyze 3D brain MRI data. By treating the 3D MRI volumes as videos, we exploit the temporal dependencies between slices to capture intricate structural relationships. The video vision transformer's self-attention mechanisms enable the model to learn long-range dependencies and identify subtle patterns that may indicate AD progression. Our proposed deep learning framework seeks to enhance the accuracy and sensitivity of AD diagnosis, empowering clinicians with a tool for early detection and intervention. We validate the performance of the video vision transformer using the ADNI dataset and conduct comparative analyses with other relevant models. Results The proposed ViTranZheimer model is compared with two hybrid models, CNN-BiLSTM and ViT-BiLSTM. CNN-BiLSTM is the combination of a convolutional neural network (CNN) and a bidirectional long-short-term memory network (BiLSTM), while ViT-BiLSTM is the combination of a vision transformer (ViT) with BiLSTM. The accuracy levels achieved in the ViTranZheimer, CNN-BiLSTM, and ViT-BiLSTM models are 98.6%, 96.479%, and 97.465%, respectively. ViTranZheimer demonstrated the highest accuracy at 98.6%, outperforming other models in this evaluation metric, indicating its superior performance in this specific evaluation metric. Conclusion This research advances the understanding of applying deep learning techniques in neuroimaging and Alzheimer's disease research, paving the way for earlier and less invasive clinical diagnosis.

Integrated sensing and communication (ISAC) boosts network efficiency by using existing resources for diverse sensing applications. In this work, we propose a cell-free massive MIMO (multiple-input multiple-output)-ISAC framework to detect unauthorized drones while simultaneously ensuring communication requirements. We develop a detector to identify passive aerial targets by analyzing signals from distributed access points (APs). In addition to the precision of the sensing, timeliness of the sensing information is also crucial due to the risk of drones leaving the area before the sensing procedure is finished. We introduce the age of sensing (AoS) and sensing coverage as our sensing performance metrics and propose a joint sensing blocklength and power optimization algorithm to minimize AoS and maximize sensing coverage while meeting communication requirements. Moreover, we propose an adaptive weight selection algorithm based on concave-convex procedure to balance the inherent trade-off between AoS and sensing coverage. Our numerical results show that increasing the communication requirements would significantly reduce both the sensing coverage and the timeliness of the sensing. Furthermore, the proposed adaptive weight selection algorithm can provide high sensing coverage and reduce the AoS by 45% compared to the fixed weights, demonstrating efficient utilization of both power and sensing blocklength.

The T center in silicon has recently emerged as a promising candidate for scalable quantum technologies, due to its telecommunications band optical transition and microwave addressable ground state spin. The immense promise of the T center is driven by its silicon host material; silicon is by far the most mature, manufacturable semiconductor material for integrated photonic and electronic devices. Here, we present the first study of T-centers in an electrical device. We study an ensemble of T centers coupled to a buried lateral P-I-N diode in silicon, observing the T-center's optical response to static and dynamic electric fields. We utilize the defect's optical response as a probe of device nonlinearity, observing a phase transition of the carrier density into a stable oscillatory regime characteristic of negative differential resistance. These findings provide fundamental insight into the physics of the T-center for improved quantum device performance and open a promising new direction for defect-based local quantum sensing in semiconductor devices.

Many real-world multiagent learning problems involve safety concerns. In these setups, typical safe reinforcement learning algorithms constrain agents' behavior, limiting exploration -- a crucial component for discovering effective cooperative multiagent behaviors. Moreover, the multiagent literature typically models individual constraints for each agent and has yet to investigate the benefits of using joint team constraints. In this work, we analyze these team constraints from a theoretical and practical perspective and propose entropic exploration for constrained multiagent reinforcement learning (E2C) to address the exploration issue. E2C leverages observation entropy maximization to incentivize exploration and facilitate learning safe and effective cooperative behaviors. Experiments across increasingly complex domains show that E2C agents match or surpass common unconstrained and constrained baselines in task performance while reducing unsafe behaviors by up to $50\%$., Comment: 10 pages, 6 figures

Cell-free massive MIMO improves the fairness among the user equipments (UEs) in the network by distributing many cooperating access points (APs) around the region while connecting them to a centralized cloud-computing unit that coordinates joint transmission/reception. However, the fiber cable deployment for the fronthaul transport network and activating all available antennas at each AP lead to increased deployment cost and power consumption for fronthaul signaling and processing. To overcome these challenges, in this work, we consider wireless fronthaul connections and propose a joint antenna activation and power allocation algorithm to minimize the end-to-end (from radio to cloud) power while satisfying the quality-of-service requirements of the UEs under wireless fronthaul capacity limitations. The results demonstrate that the proposed methodology of deactivating antennas at each AP reduces the power consumption by 50% and 84% compared to the benchmarks based on shutting down APs and minimizing only the transmit power, respectively., Comment: Presented in Asilomar Signals, Systems and Computers 2024

Integrated sensing and communication (ISAC) is a promising technology for future mobile networks, enabling sensing applications to be performed by existing communication networks, consequently improving the system efficiency. Millimeter wave (mmWave) signals provide high sensing resolution and high data rate but suffer from sensitivity to blockage. Cell-free massive multiple-input multiple-output (MIMO), with a large number of distributed access points (APs), can overcome this challenge by providing macro diversity against changing blockages and can save energy consumption by deactivating unfavorable APs. Thus, in this work, we propose a joint dynamic AP mode selection and power allocation scheme for mmWave cell-free massive MIMO-ISAC, where APs are assigned either as ISAC transmitters, sensing receivers, or shut down. Due to the large size of the original problem, we propose three different sub-optimal algorithms that minimize the number of active APs while guaranteeing the sensing and communication constraints. Numerical results demonstrate that assigning ISAC transmitters only satisfying communication constraints, followed up by sensing receiver assignment only for sensing constraint achieves the best performance-complexity balance., Comment: Presented in Asilomar Conference on Signals, Systems, and Computers 2024

The discovery of the kagome metal CsV$_3$Sb$_5$ sparked broad interest, due to the coexistence of a charge density wave (CDW) phase and possible unconventional superconductivity in the material. In this study, we use low-energy electron diffraction (LEED) with a $\mu$m-sized electron beam to explore the periodic lattice distortion at the antimony-terminated surface in the CDW phase. We recorded high-quality backscattering diffraction patterns in ultrahigh vacuum from multiple cleaved samples. Unexpectedly, we did not find superstructure reflexes at intensity levels predicted from dynamical LEED calculations for the reported $2 \times 2 \times 2$ bulk structure. Our results suggest that in CsV$_3$Sb$_5$ the periodic lattice distortion accompanying the CDW is less pronounced at Sb-terminated surfaces than in the bulk., Comment: 9 pages, 6 figures

We introduce an extension of the Difference of Convex Algorithm (DCA) in the form of a block coordinate approach for problems with separable structure. For $n$ coordinate-blocks and $k$ iterations, our main result proves a non-asymptotic convergence rate of $O(n/k)$ for the proposed method. Furthermore, leveraging the connection between DCA and Expectation Maximization (EM), we propose a block coordinate EM algorithm., Comment: 21 Pages, 3 figure

Neuron importance assessment is crucial for understanding the inner workings of artificial neural networks (ANNs) and improving their interpretability and efficiency. This paper introduces a novel approach to neuron significance assessment inspired by frequency tagging, a technique from neuroscience. By applying sinusoidal contrast modulation to image inputs and analyzing resulting neuron activations, this method enables fine-grained analysis of a network's decision-making processes. Experiments conducted with a convolutional neural network for image classification reveal notable harmonics and intermodulations in neuron-specific responses under part-based frequency tagging. These findings suggest that ANNs exhibit behavior akin to biological brains in tuning to flickering frequencies, thereby opening avenues for neuron/filter importance assessment through frequency tagging. The proposed method holds promise for applications in network pruning, and model interpretability, contributing to the advancement of explainable artificial intelligence and addressing the lack of transparency in neural networks. Future research directions include developing novel loss functions to encourage biologically plausible behavior in ANNs., Comment: 10 pages, 5 figures

Integrated sensing and communications (ISAC) allows networks to perform sensing alongside data transmission. While most ISAC studies focus on single-target, multi-user scenarios, multi-target sensing is scarcely researched. This letter examines the monostatic sensing performance of a multi-target massive MIMO system, aiming to minimize the sum of Cram\'er-Rao lower bounds (CRLBs) for target direction-of-arrival estimates while meeting user equipment (UE) rate requirements. We propose several precoding schemes, comparing sensing performance and complexity, and find that sensing-focused precoding with power allocation for communication achieves near-optimal performance with 20 times less complexity than joint precoding. Additionally, time-sharing between communication and sensing outperforms simple time division, highlighting the benefits of resource-sharing for ISAC.

Solving non-convex, NP-hard optimization problems is crucial for training machine learning models, including neural networks. However, non-convexity often leads to black-box machine learning models with unclear inner workings. While convex formulations have been used for verifying neural network robustness, their application to training neural networks remains less explored. In response to this challenge, we reformulate the problem of training infinite-width two-layer ReLU networks as a convex completely positive program in a finite-dimensional (lifted) space. Despite the convexity, solving this problem remains NP-hard due to the complete positivity constraint. To overcome this challenge, we introduce a semidefinite relaxation that can be solved in polynomial time. We then experimentally evaluate the tightness of this relaxation, demonstrating its competitive performance in test accuracy across a range of classification tasks.

Resonance is a well-known phenomenon that happens in systems with second order dynamics. In this paper we address the fundamental question of making a network robust to signal being periodically pumped into it at or near a resonant frequency by an adversarial agent with the aim of saturating the network with the signal. Towards this goal, we develop the notion of network vulnerability, which is measured by the expected resonance amplitude on the network under a stochastically modeled adversarial attack. Assuming a second order dynamics model based on the network graph Laplacian matrix and a known stochastic model for the adversarial attack, we propose two methods for minimizing the network vulnerability through optimization of the spectrum of the network graph. We provide extensive numerical results analyzing the effects of both methods., Comment: 13 pages, 18 figures

While the language modeling objective has been shown to be deeply connected with compression, it is surprising that modern LLMs are not employed in practical text compression systems. In this paper, we provide an in-depth analysis of neural network and transformer-based compression techniques to answer this question. We compare traditional text compression systems with neural network and LLM-based text compression methods. Although LLM-based systems significantly outperform conventional compression methods, they are highly impractical. Specifically, LLMZip, a recent text compression system using Llama3-8B requires 9.5 days to compress just 10 MB of text, although with huge improvements in compression ratios. To overcome this, we present FineZip - a novel LLM-based text compression system that combines ideas of online memorization and dynamic context to reduce the compression time immensely. FineZip can compress the above corpus in approximately 4 hours compared to 9.5 days, a 54 times improvement over LLMZip and comparable performance. FineZip outperforms traditional algorithmic compression methods with a large margin, improving compression ratios by approximately 50\%. With this work, we take the first step towards making lossless text compression with LLMs a reality. While FineZip presents a significant step in that direction, LLMs are still not a viable solution for large-scale text compression. We hope our work paves the way for future research and innovation to solve this problem.

Translating Natural Language Queries into Structured Query Language (Text-to-SQL or NLQ-to-SQL) is a critical task extensively studied by both the natural language processing and database communities, aimed at providing a natural language interface to databases (NLIDB) and lowering the barrier for non-experts. Despite recent advancements made through the use of Large Language Models (LLMs), significant challenges remain. These include handling complex database schemas, resolving ambiguity in user queries, and generating SQL queries with intricate structures that accurately reflect the user's intent. In this work, we introduce E-SQL, a novel pipeline specifically designed to address these challenges through direct schema linking and candidate predicate augmentation. E-SQL enhances the natural language query by incorporating relevant database items (i.e., tables, columns, and values) and conditions directly into the question and SQL construction plan, bridging the gap between the query and the database structure. The pipeline leverages candidate predicate augmentation to mitigate erroneous or incomplete predicates in generated SQLs. Comprehensive evaluations on the BIRD benchmark illustrate that E-SQL achieves competitive performance, particularly excelling in complex queries with a 66.29% execution accuracy on the test set. A further observation from our experiments reveals that incorporating schema filtering into the translation pipeline does not have a positive impact on performance when the most advanced proprietary LLMs are used. Additionally, our experiments with small LLMs highlight the importance and positive impact of enriched questions on their performance. Without fine-tuning, single-prompt SQL generation using enriched questions with DeepSeek Coder 7B Instruct 1.5v achieves 56.45% execution accuracy on the BIRD development set.

Wireless communication is critical for the non-invasive in-situ monitoring of vital signs in deep tissues. Wireless ultrasonic links demonstrated so far solved the shortcomings of electromagnetic wave-based communication methods. However, they either require highly customized and complex designs for the implant electronics or rely on physical changes in implanted metamaterials. Here, we report a wireless, passive, frequency-based, and electronics-free ultrasonic communication method for deep tissue sensor implants. The device consists of a piezoelectric crystal used as the ultrasonic antenna, considerably simplifying the implant design, and can be integrated with any capacitive implantable sensor. We demonstrate the applicability of the passive communication method using a commercial pressure sensor at a depth of 5 cm with decent precision and accuracy.

The evolution of superconducting quantum processors is driven by the need to reduce errors and scale for fault-tolerant computation. Reducing physical qubit error rates requires further advances in the microscopic modeling and control of decoherence mechanisms in superconducting qubits. Piezoelectric interactions contribute to decoherence by mediating energy exchange between microwave photons and acoustic phonons. Centrosymmetric materials like silicon and sapphire do not display piezoelectricity and are the preferred substrates for superconducting qubits. However, the broken centrosymmetry at material interfaces may lead to piezoelectric losses in qubits. While this loss mechanism was predicted two decades ago, interface piezoelectricity has not been experimentally observed in superconducting devices. Here, we report the observation of interface piezoelectricity at an aluminum-silicon junction and show that it constitutes an important loss channel for superconducting devices. We fabricate aluminum interdigital surface acoustic wave transducers on silicon and demonstrate piezoelectric transduction from room temperature to millikelvin temperatures. We find an effective electromechanical coupling factor of $K^2\approx 2 \times 10^{-5}\%$ comparable to weakly piezoelectric substrates. We model the impact of the measured interface piezoelectric response on superconducting qubits and find that the piezoelectric surface loss channel limits qubit quality factors to $Q\sim10^4-10^8$ for designs with different surface participation ratios and electromechanical mode matching. These results identify electromechanical surface losses as a significant dissipation channel for superconducting qubits, and show the need for heterostructure and phononic engineering to minimize errors in next-generation superconducting qubits.

A seminal result of Koml\'os, S\'ark\"ozy, and Szemer\'edi states that any n-vertex graph G with minimum degree at least (1/2 + {\alpha})n contains every n-vertex tree T of bounded degree. Recently, Pham, Sah, Sawhney, and Simkin extended this result to show that such graphs G in fact support an optimally spread distribution on copies of a given T, which implies, using the recent breakthroughs on the Kahn-Kalai conjecture, the robustness result that T is a subgraph of sparse random subgraphs of G as well. Pham, Sah, Sawhney, and Simkin construct their optimally spread distribution by following closely the original proof of the Koml\'os-S\'ark\"ozy-Szemer\'edi theorem which uses the blow-up lemma and the Szemer\'edi regularity lemma. We give an alternative, regularity-free construction that instead uses the Koml\'os-S\'ark\"ozy-Szemer\'edi theorem (which has a regularity-free proof due to Kathapurkar and Montgomery) as a black-box. Our proof is based on the simple and general insight that, if G has linear minimum degree, almost all constant sized subgraphs of G inherit the same minimum degree condition that G has., Comment: 12 pages

Our research investigates the challenges Deep Reinforcement Learning (DRL) faces in complex, Partially Observable Markov Decision Processes (POMDP) such as autonomous driving (AD), and proposes a solution for vision-based navigation in these environments. Partial observability reduces RL performance significantly, and this can be mitigated by augmenting sensor information and data fusion to reflect a more Markovian environment. However, this necessitates an increasingly complex perception module, whose training via RL is complicated due to inherent limitations. As the neural network architecture becomes more complex, the reward function's effectiveness as an error signal diminishes since the only source of supervision is the reward, which is often noisy, sparse, and delayed. Task-irrelevant elements in images, such as the sky or certain objects, pose additional complexities. Our research adopts an offline-trained encoder to leverage large video datasets through self-supervised learning to learn generalizable representations. Then, we train a head network on top of these representations through DRL to learn to control an ego vehicle in the CARLA AD simulator. This study presents a broad investigation of the impact of different learning schemes for offline-training of encoders on the performance of DRL agents in challenging AD tasks. Furthermore, we show that the features learned by watching BDD100K driving videos can be directly transferred to achieve lane following and collision avoidance in CARLA simulator, in a zero-shot learning fashion. Finally, we explore the impact of various architectural decisions for the RL networks to utilize the transferred representations efficiently. Therefore, in this work, we introduce and validate an optimal way for obtaining suitable representations of the environment, and transferring them to RL networks.

Federated learning (FL) has gained a lot of attention in recent years for building privacy-preserving collaborative learning systems. However, FL algorithms for constrained machine learning problems are still limited, particularly when the projection step is costly. To this end, we propose a Federated Frank-Wolfe Algorithm (FedFW). FedFW features data privacy, low per-iteration cost, and communication of sparse signals. In the deterministic setting, FedFW achieves an $\varepsilon$-suboptimal solution within $O(\varepsilon^{-2})$ iterations for smooth and convex objectives, and $O(\varepsilon^{-3})$ iterations for smooth but non-convex objectives. Furthermore, we present a stochastic variant of FedFW and show that it finds a solution within $O(\varepsilon^{-3})$ iterations in the convex setting. We demonstrate the empirical performance of FedFW on several machine learning tasks., Comment: European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases

Prior work on the Image Quality Transfer on Diffusion MRI (dMRI) has shown significant improvement over traditional interpolation methods. However, the difficulty in obtaining ultra-high resolution Diffusion MRI scans poses a problem in training neural networks to obtain high-resolution dMRI scans. Here we hypothesise that the inclusion of structural MRI images, which can be acquired at much higher resolutions, can be used as a guide to obtaining a more accurate high-resolution dMRI output. To test our hypothesis, we have constructed a novel framework that incorporates structural MRI scans together with dMRI to obtain high-resolution dMRI scans. We set up tests which evaluate the validity of our claim through various configurations and compare the performance of our approach against a unimodal approach. Our results show that the inclusion of structural MRI scans do lead to an improvement in high-resolution image prediction when T1w data is incorporated into the model input.

We present foundation language models developed to power Apple Intelligence features, including a ~3 billion parameter model designed to run efficiently on devices and a large server-based language model designed for Private Cloud Compute. These models are designed to perform a wide range of tasks efficiently, accurately, and responsibly. This report describes the model architecture, the data used to train the model, the training process, how the models are optimized for inference, and the evaluation results. We highlight our focus on Responsible AI and how the principles are applied throughout the model development.

In the design of a metasurface-assisted system for indoor environments, it is essential to take into account not only the performance gains and coverage extension provided by the metasurface but also the operating costs brought by its reconfigurability, such as powering and cabling. These costs can present challenges, particularly in indoor dense spaces (IDSs). A self-sustainable metasurface (SSM), which retains reconfigurability unlike a static metasurface (SMS), achieves a lower operating cost than a reconfigurable intelligent surface (RIS) by being self-sustainable through power harvesting. In this paper, in order to find a better trade-off between metasurface gain, coverage, and operating cost, the design and performance of an SSM-assisted indoor mmWave communication system are investigated. We first simplify the design of the SSM-assisted system by considering the use of SSMs in a preset-based manner and the formation of coverage groups by associating SSMs with the closest user equipments (UEs). We propose a two-stage iterative algorithm to maximize the minimum data rate in the system by jointly deciding the association between the UEs and the SSMs, the phase-shifts of the SSMs, and allocating time resources for each UE. The non-convexities that exist in the proposed optimization problem are tackled using the feasible point pursuit successive convex approximation method and the concave-convex procedure. To understand the best scenario for using SSM, the resulting performance is compared with that achieved with RIS and SMS. Our numerical results indicate that SSMs are best utilized in a small environment where self-sustainability is easier to achieve when the budget for operating costs is tight., Comment: 13 Pages, 16 Figures, submitted to IEEE Transaction on Wireless Communication, under review

Silicon nanomechanical resonators display ultra-long lifetimes at cryogenic temperatures and microwave frequencies. Achieving quantum control of single-phonons in these devices has so far relied on nonlinearities enabled by coupling to ancillary qubits. In this work, we propose using atomic forces to realize a silicon nanomechanical qubit without coupling to an ancillary qubit. The proposed qubit operates at 60 MHz with a single-phonon level anharmonicity of 5 MHz. We present a circuit quantum acoustodynamics architecture where electromechanical resonators enable dispersive state readout and multi-qubit operations. The combination of strong anharmonicity, ultrahigh mechanical quality factors, and small footprints achievable in this platform could enable quantum-nonlinear phononics for quantum information processing and transduction.

The key challenge of personalized federated learning (PerFL) is to capture the statistical heterogeneity properties of data with inexpensive communications and gain customized performance for participating devices. To address these, we introduced personalized federated learning in multi-tier architecture (PerMFL) to obtain optimized and personalized local models when there are known team structures across devices. We provide theoretical guarantees of PerMFL, which offers linear convergence rates for smooth strongly convex problems and sub-linear convergence rates for smooth non-convex problems. We conduct numerical experiments demonstrating the robust empirical performance of PerMFL, outperforming the state-of-the-art in multiple personalized federated learning tasks.

An $\ell$-lift of a graph $G$ is any graph obtained by replacing every vertex of $G$ with an independent set of size $\ell$, and connecting every pair of two such independent sets that correspond to an edge in $G$ by a matching of size $\ell$. Graph lifts have found numerous interesting applications and connections to a variety of areas over the years. Of particular importance is the random graph model obtained by considering an $\ell$-lift of a graph sampled uniformly at random. This model was first introduced by Amit and Linial in 1999, and has been extensively investigated since. In this paper, we study the size of the largest topological clique in random lifts of complete graphs. In 2006, Drier and Linial raised the conjecture that almost all $\ell$-lifts of the complete graph on $n$ vertices contain a subdivision of a clique of order $\Omega(n)$ as a subgraph provided $\ell$ is at least linear in $n$. We confirm their conjecture in a strong form by showing that for $\ell \ge (1+o(1))n$, one can almost surely find a subdivision of a clique of order $n$. We prove that this is tight by showing that for $\ell \le (1-o(1))n$, almost all $\ell$-lifts do not contain subdivisions of cliques of order $n$. Finally, for $2 \le \ell \ll n$, we show that almost all $\ell$-lifts of $K_n$ contain a subdivision of a clique on $(1-o(1))\sqrt{\frac{2n \ell}{1-1/\ell}}$ vertices and that this is tight up to the lower order term.

We show that a $k$-uniform hypergraph on $n$ vertices has a spanning subgraph homeomorphic to the $(k - 1)$-dimensional sphere provided that $H$ has no isolated vertices and each set of $k - 1$ vertices supported by an edge is contained in at least $n/2 + o(n)$ edges. This gives a topological extension of Dirac's theorem and asymptotically confirms a conjecture of Georgakopoulos, Haslegrave, Montgomery, and Narayanan. Unlike typical results in the area, our proof does not rely on the Absorption Method, the Regularity Lemma or the Blow-up Lemma. Instead, we use a recently introduced framework that is based on covering the vertex set of the host graph with a family of complete blow-ups., Comment: 20 pages

Recent work has shown that 8-bit floating point (FP8) can be used for efficiently training neural networks with reduced computational overhead compared to training in FP32/FP16. In this work, we investigate the use of FP8 training in a federated learning context. This brings not only the usual benefits of FP8 which are desirable for on-device training at the edge, but also reduces client-server communication costs due to significant weight compression. We present a novel method for combining FP8 client training while maintaining a global FP32 server model and provide convergence analysis. Experiments with various machine learning models and datasets show that our method consistently yields communication reductions of at least 2.9x across a variety of tasks and models compared to an FP32 baseline.