Real-world three-phase microgrids face two interconnected challenges: 1. time-varying uncertainty from renewable generation and demand, and 2. persistent phase imbalances caused by uneven distributed energy resources DERs, load asymmetries, and grid faults. Conventional energy management systems fail to address these challenges holistically and static optimization methods lack adaptability to real-time fluctuations, while balanced three-phase models ignore critical asymmetries that degrade voltage stability and efficiency. This work introduces a dynamic rolling horizon optimization framework specifically designed for unbalanced three-phase microgrids. Unlike traditional two-stage stochastic approaches that fix decisions for the entire horizon, the rolling horizon algorithm iteratively updates decisions in response to real-time data. By solving a sequence of shorter optimization windows, each incorporating the latest system state and forecasts, the method achieves three key advantages: Adaptive Uncertainty Handling by continuously re plans operations to mitigate forecast errors. Phase Imbalance Correction by dynamically adjusts power flows across phases to minimize voltage deviations and losses caused by asymmetries, and computational Tractability, i.e., shorter optimization windows, combined with the mathematical mhodel, enable better decision making holding accuracy. For comparison purposes, we derive three optimization models: a nonlinear nonconvex model for high-fidelity offline planning, a convex quadratic approximation for day-ahead scheduling, and a linearized model to important for theoretical reasons such as decomposition algorithms.
In this paper, we study the optimal timing for pilot and data transmissions to maximize effective throughput, also known as goodput, over a wireless fading channel. The receiver utilizes the received pilot signal and its Age of Information (AoI), termed the Age of Channel State Information (AoCSI), to estimate the channel state. Based on this estimation, the transmitter selects an appropriate modulation and coding scheme (MCS) to maximize goodput while ensuring compliance with a predefined block error probability constraint. Furthermore, we design an optimal pilot scheduling policy that determines whether to transmit a pilot or data at each time step, with the objective of maximizing the long-term average goodput. This problem involves a non-monotonic AoI metric optimization challenge, as the goodput function is non-monotonic with respect to AoCSI. The numerical results illustrate the performance gains achieved by the proposed policy under various SNR levels and mobility speeds.
During the scheduling process, the traffic network operator (TNO) and the distribution network operator (DNO) act noncooperatively. Under the TNO management, the distribution of charging loads may exacerbate the local supply demand imbalance in the power distribution network (PDN), which negatively impacts the economic operation of the PDN. This paper proposes a profitsharing mechanism based on the principle of incentive compatibility for coordinating the traffic network (TN) and the PDN to minimize the operation cost of PDN. Under this mechanism, the scheduling process of the power traffic system is divided into two stages. At the prescheduling stage, the TNO allocates traffic flow and charging loads without considering the operation of the PDN, after which the DNO schedules and obtains the original cost. At the rescheduling stage, the DNO shares part of benefits of the optimal operation to the TNO to redispatch the EV charging to obtain a more effective charging plan, thus minimize the overall cost of PDN. Then, a bilevel model is developed to simulate the operation of the power traffic system with the proposed sharing scheme and identify the best sharing ratio. Finally, numerical results demonstrate that the PDN can achieve the minimum total cost and simultaneously the TN can also benefit from the proposed profit sharing mechanism.
Virtual energy stations (VESs) work as retailers to provide electricity and natural gas sale services for integrated energy systems (IESs), and guide IESs energy consumption behaviors to tackle the varying market prices via integrated demand response (IDR). However, IES customers are risk averse and show low enthusiasm in responding to the IDR incentive signals. To address this problem, exergy is utilized to unify different energies and allowed to be virtually stored and withdrawn for arbitrage by IESs. The whole incentive mechanism operating process is innovatively characterized by a virtual exergy battery. Peer to peer (P2P) exergy trading based on shared exergy storage is also developed to reduce the energy cost of IESs without any extra transmission fee. In this way, IES can reduce the economic loss risk caused by the market price fluctuation via the different time (time dimension), multiple energy conversion (energy dimension), and P2P exergy trading (space dimension) arbitrage. Moreover, the optimal scheduling of VES and IESs is modeled by a bilevel optimization model. The consensus based alternating direction method of multipliers (CADMM) algorithm is utilized to solve this problem in a distributed way. Simulation results validate the effectiveness of the proposed incentive mechanism and show that the shared exergy storage can enhance the benefits of different type IESs by 18.96%, 3.49%, and 3.15 %, respectively.
This paper studies the robust co-planning of proactive network hardening and mobile hydrogen energy resources (MHERs) scheduling, which is to enhance the resilience of power distribution network (PDN) against the disastrous events. A decision-dependent robust optimization model is formulated with min-max resilience constraint and discrete recourse structure, which helps achieve the load survivability target considering endogenous uncertainties. Different from the traditional model with a fixed uncertainty set, we adopt a dynamic representation that explicitly captures the endogenous uncertainties of network contingency as well as the available hydrogen storage levels of MHERs, which induces a decision-dependent uncertainty (DDU) set. Also, the multi-period adaptive routing and energy scheduling of MHERs are modeled as a mixed-integer recourse problem for further decreasing the resilience cost. Then, a nested parametric column-and-constraint generation (N-PC&CG) algorithm is customized and developed to solve this challenging formulation. By leveraging the structural property of the DDU set as well as the combination of discrete recourse decisions and the corresponding extreme points, we derive a strengthened solution scheme with nontrivial enhancement strategies to realize efficient and exact computation. Numerical results on 14-bus test system and 56-bus real-world distribution network demonstrate the resilience benefits and economical feasibility of the proposed method under different damage severity levels. Moreover, the enhanced N-PC&CG shows a superior solution capability to support prompt decisions for resilient planning with DDU models.
With the accelerated development of Industry 4.0, intelligent manufacturing systems increasingly require efficient task allocation and scheduling in multi-robot systems. However, existing methods rely on domain expertise and face challenges in adapting to dynamic production constraints. Additionally, enterprises have high privacy requirements for production scheduling data, which prevents the use of cloud-based large language models (LLMs) for solution development. To address these challenges, there is an urgent need for an automated modeling solution that meets data privacy requirements. This study proposes a knowledge-augmented mixed integer linear programming (MILP) automated formulation framework, integrating local LLMs with domain-specific knowledge bases to generate executable code from natural language descriptions automatically. The framework employs a knowledge-guided DeepSeek-R1-Distill-Qwen-32B model to extract complex spatiotemporal constraints (82% average accuracy) and leverages a supervised fine-tuned Qwen2.5-Coder-7B-Instruct model for efficient MILP code generation (90% average accuracy). Experimental results demonstrate that the framework successfully achieves automatic modeling in the aircraft skin manufacturing case while ensuring data privacy and computational efficiency. This research provides a low-barrier and highly reliable technical path for modeling in complex industrial scenarios.
Bridging different modalities lies at the heart of cross-modality generation. While conventional approaches treat the text modality as a conditioning signal that gradually guides the denoising process from Gaussian noise to the target image modality, we explore a much simpler paradigm-directly evolving between text and image modalities through flow matching. This requires projecting both modalities into a shared latent space, which poses a significant challenge due to their inherently different representations: text is highly semantic and encoded as 1D tokens, whereas images are spatially redundant and represented as 2D latent embeddings. To address this, we introduce FlowTok, a minimal framework that seamlessly flows across text and images by encoding images into a compact 1D token representation. Compared to prior methods, this design reduces the latent space size by 3.3x at an image resolution of 256, eliminating the need for complex conditioning mechanisms or noise scheduling. Moreover, FlowTok naturally extends to image-to-text generation under the same formulation. With its streamlined architecture centered around compact 1D tokens, FlowTok is highly memory-efficient, requires significantly fewer training resources, and achieves much faster sampling speeds-all while delivering performance comparable to state-of-the-art models. Code will be available at https://github.com/bytedance/1d-tokenizer.
The growing demand for computational power is driven by advancements in deep learning, the increasing need for big data processing, and the requirements of scientific simulations for academic and research purposes. Developing countries like Nepal often struggle with the resources needed to invest in new and better hardware for these purposes. However, optimizing and building on existing technology can still meet these computing demands effectively. To address these needs, we built a cluster using four NVIDIA GeForce GTX 1650 GPUs. The cluster consists of four nodes: one master node that controls and manages the entire cluster, and three compute nodes dedicated to processing tasks. The master node is equipped with all necessary software for package management, resource scheduling, and deployment, such as Anaconda and Slurm. In addition, a Network File Storage (NFS) system was integrated to provide the additional storage required by the cluster. Given that the cluster is accessible via ssh by a public domain address, which poses significant cybersecurity risks, we implemented fail2ban to mitigate brute force attacks and enhance security. Despite the continuous challenges encountered during the design and implementation process, this project demonstrates how powerful computational clusters can be built to handle resource-intensive tasks in various demanding fields.
Scheduling deep learning (DL) models to train on powerful clusters with accelerators like GPUs and TPUs, presently falls short, either lacking fine-grained heterogeneity awareness or leaving resources substantially under-utilized. To fill this gap, we propose a novel design of a task-level heterogeneity-aware scheduler, Hadar, based on an optimization framework that can boost resource utilization. Hadar leverages the performance traits of DL jobs on a heterogeneous DL cluster, characterizes the task-level performance heterogeneity in the optimization problem, and makes scheduling decisions across both spatial and temporal dimensions. It involves the primal-dual framework employing a dual subroutine, to solve the optimization problem and guide the scheduling design. Our trace-driven simulation with representative DL model training workloads demonstrates that Hadar accelerates the total time duration by 1.20x when compared with its state-of-the-art heterogeneity-aware counterpart, Gavel. Further, our Hadar scheduler is enhanced to HadarE by forking each job into multiple copies to let a job train concurrently on heterogeneous GPUs resided on separate available nodes (i.e., machines or servers) for resource utilization enhancement. HadarE is evaluated extensively on physical DL clusters for comparison with Hadar and Gavel. With substantial enhancement in cluster resource utilization (by 1.45x), HadarE exhibits considerable speed-ups in DL model training, reducing the total time duration by 50% (or 80%) on an Amazon's AWS (or our lab) cluster, while producing trained DL models with consistently better inference quality than those trained by Hadar.
Transformer-based large language models (LLMs) have demonstrated exceptional capabilities in sequence modeling and text generation, with improvements scaling proportionally with model size. However, the limitations of GPU memory have restricted LLM training accessibility for many researchers. Existing heterogeneous training methods significantly expand the scale of trainable models but introduce substantial communication overheads and CPU workloads. In this work, we propose AutoHete, an automatic and efficient heterogeneous training system compatible with both single-GPU and multi-GPU environments. AutoHete dynamically adjusts activation checkpointing, parameter offloading, and optimizer offloading based on the specific hardware configuration and LLM training needs. Additionally, we design a priority-based scheduling mechanism that maximizes the overlap between operations across training iterations, enhancing throughput. Compared to state-of-the-art heterogeneous training systems, AutoHete delivers a 1.32x~1.91x throughput improvement across various model sizes and training configurations.
We propose a diffusion-based inverse rendering framework that decomposes a single RGB image into geometry, material, and lighting. Inverse rendering is inherently ill-posed, making it difficult to predict a single accurate solution. To address this challenge, recent generative model-based methods aim to present a range of possible solutions. However, finding a single accurate solution and generating diverse solutions can be conflicting. In this paper, we propose a channel-wise noise scheduling approach that allows a single diffusion model architecture to achieve two conflicting objectives. The resulting two diffusion models, trained with different channel-wise noise schedules, can predict a single highly accurate solution and present multiple possible solutions. The experimental results demonstrate the superiority of our two models in terms of both diversity and accuracy, which translates to enhanced performance in downstream applications such as object insertion and material editing.
Generating co-speech gestures in real time requires both temporal coherence and efficient sampling. We introduce Accelerated Rolling Diffusion, a novel framework for streaming gesture generation that extends rolling diffusion models with structured progressive noise scheduling, enabling seamless long-sequence motion synthesis while preserving realism and diversity. We further propose Rolling Diffusion Ladder Acceleration (RDLA), a new approach that restructures the noise schedule into a stepwise ladder, allowing multiple frames to be denoised simultaneously. This significantly improves sampling efficiency while maintaining motion consistency, achieving up to a 2x speedup with high visual fidelity and temporal coherence. We evaluate our approach on ZEGGS and BEAT, strong benchmarks for real-world applicability. Our framework is universally applicable to any diffusion-based gesture generation model, transforming it into a streaming approach. Applied to three state-of-the-art methods, it consistently outperforms them, demonstrating its effectiveness as a generalizable and efficient solution for real-time, high-fidelity co-speech gesture synthesis.
Speculative inference is a promising paradigm employing small speculative models (SSMs) as drafters to generate draft tokens, which are subsequently verified in parallel by the target large language model (LLM). This approach enhances the efficiency of inference serving by reducing LLM inference latency and costs while preserving generation quality. However, existing speculative methods face critical challenges, including inefficient resource utilization and limited draft acceptance, which constrain their scalability and overall effectiveness. To overcome these obstacles, we present CoSine, a novel speculative inference system that decouples sequential speculative decoding from parallel verification, enabling efficient collaboration among multiple nodes. Specifically, CoSine routes inference requests to specialized drafters based on their expertise and incorporates a confidence-based token fusion mechanism to synthesize outputs from cooperating drafters, ensuring high-quality draft generation. Additionally, CoSine dynamically orchestrates the execution of speculative decoding and verification in a pipelined manner, employing batch scheduling to selectively group requests and adaptive speculation control to minimize idle periods. By optimizing parallel workflows through heterogeneous node collaboration, CoSine balances draft generation and verification throughput in real-time, thereby maximizing resource utilization. Experimental results demonstrate that CoSine achieves superior performance compared to state-of-the-art speculative approaches. Notably, with equivalent resource costs, CoSine achieves up to a 23.2% decrease in latency and a 32.5% increase in throughput compared to baseline methods.
In XR downlink transmission, energy-efficient power scheduling (EEPS) is essential for conserving power resource while delivering large data packets within hard-latency constraints. Traditional constrained reinforcement learning (CRL) algorithms show promise in EEPS but still struggle with non-convex stochastic constraints, non-stationary data traffic, and sparse delayed packet dropout feedback (rewards) in XR. To overcome these challenges, this paper models the EEPS in XR as a dynamic parameter-constrained Markov decision process (DP-CMDP) with a varying transition function linked to the non-stationary data traffic and solves it by a proposed context-aware constrained reinforcement learning (CACRL) algorithm, which consists of a context inference (CI) module and a CRL module. The CI module trains an encoder and multiple potential networks to characterize the current transition function and reshape the packet dropout rewards according to the context, transforming the original DP-CMDP into a general CMDP with immediate dense rewards. The CRL module employs a policy network to make EEPS decisions under this CMDP and optimizes the policy using a constrained stochastic successive convex approximation (CSSCA) method, which is better suited for non-convex stochastic constraints. Finally, theoretical analyses provide deep insights into the CADAC algorithm, while extensive simulations demonstrate that it outperforms advanced baselines in both power conservation and satisfying packet dropout constraints.
As the artificial intelligence community advances into the era of large models with billions of parameters, distributed training and inference have become essential. While various parallelism strategies-data, model, sequence, and pipeline-have been successfully implemented for popular neural networks on main-stream hardware, optimizing the distributed deployment schedule requires extensive expertise and manual effort. Further more, while existing frameworks with most simple chain-like structures, they struggle with complex non-linear architectures. Mixture-of-experts and multi-modal models feature intricate MIMO and branch-rich topologies that require fine-grained operator-level parallelization beyond the capabilities of existing frameworks. We propose formulating parallelism planning as a scheduling optimization problem using mixed-integer programming. We propose a bi-level solution framework balancing optimality with computational efficiency, automatically generating effective distributed plans that capture both the heterogeneous structure of modern neural networks and the underlying hardware constraints. In experiments comparing against expert-designed strategies like DeepSeek's DualPipe, our framework achieves comparable or superior performance, reducing computational bubbles by half under the same memory constraints. The framework's versatility extends beyond throughput optimization to incorporate hardware utilization maximization, memory capacity constraints, and other considerations or potential strategies. Such capabilities position our solution as both a valuable research tool for exploring optimal parallelization strategies and a practical industrial solution for large-scale AI deployment.
Large Language Models have revolutionized natural language processing, yet serving them efficiently in data centers remains challenging due to mixed workloads comprising latency-sensitive (LS) and best-effort (BE) jobs. Existing inference systems employ iteration-level first-come-first-served scheduling, causing head-of-line blocking when BE jobs delay LS jobs. We introduce QLLM, a novel inference system designed for Mixture of Experts (MoE) models, featuring a fine-grained, priority-aware preemptive scheduler. QLLM enables expert-level preemption, deferring BE job execution while minimizing LS time-to-first-token (TTFT). Our approach removes iteration-level scheduling constraints, enabling the scheduler to preempt jobs at any layer based on priority. Evaluations on an Nvidia A100 GPU show that QLLM significantly improves performance. It reduces LS TTFT by an average of $65.5\times$ and meets the SLO at up to $7$ requests/sec, whereas the baseline fails to do so under the tested workload. Additionally, it cuts LS turnaround time by up to $12.8\times$ without impacting throughput. QLLM is modular, extensible, and seamlessly integrates with Hugging Face MoE models.
The emergence of Large Language Models (LLMs) in Multi-Agent Systems (MAS) has opened new possibilities for artificial intelligence, yet current implementations face significant challenges in resource management, task coordination, and system efficiency. While existing frameworks demonstrate the potential of LLM-based agents in collaborative problem-solving, they often lack sophisticated mechanisms for parallel execution and dynamic task management. This paper introduces DynTaskMAS, a novel framework that orchestrates asynchronous and parallel operations in LLM-based MAS through dynamic task graphs. The framework features four key innovations: (1) a Dynamic Task Graph Generator that intelligently decomposes complex tasks while maintaining logical dependencies, (2) an Asynchronous Parallel Execution Engine that optimizes resource utilization through efficient task scheduling, (3) a Semantic-Aware Context Management System that enables efficient information sharing among agents, and (4) an Adaptive Workflow Manager that dynamically optimizes system performance. Experimental evaluations demonstrate that DynTaskMAS achieves significant improvements over traditional approaches: a 21-33% reduction in execution time across task complexities (with higher gains for more complex tasks), a 35.4% improvement in resource utilization (from 65% to 88%), and near-linear throughput scaling up to 16 concurrent agents (3.47X improvement for 4X agents). Our framework establishes a foundation for building scalable, high-performance LLM-based multi-agent systems capable of handling complex, dynamic tasks efficiently.
Neural Combinatorial Optimization (NCO) has emerged as a promising approach for NP-hard problems. However, prevailing RL-based methods suffer from low sample efficiency due to sparse rewards and underused solutions. We propose Preference Optimization for Combinatorial Optimization (POCO), a training paradigm that leverages solution preferences via objective values. It introduces: (1) an efficient preference pair construction for better explore and exploit solutions, and (2) a novel loss function that adaptively scales gradients via objective differences, removing reliance on reward models or reference policies. Experiments on Job-Shop Scheduling (JSP), Traveling Salesman (TSP), and Flexible Job-Shop Scheduling (FJSP) show POCO outperforms state-of-the-art neural methods, reducing optimality gaps impressively with efficient inference. POCO is architecture-agnostic, enabling seamless integration with existing NCO models, and establishes preference optimization as a principled framework for combinatorial optimization.
Spatial-frequency scheduling and hybrid precoding in wideband multi-user multi-antenna systems have never been learned jointly due to the challenges arising from the massive user combinations on resource blocks (RBs) and the shared analog precoder among RBs. In this paper, we strive to jointly learn the scheduling and precoding policies with graph neural networks (GNNs), which have emerged as a powerful tool for optimizing resource allocation thanks to their potential in generalizing across problem scales. By reformulating the joint optimization problem into an equivalent functional optimization problem for the scheduling and precoding policies, we propose a GNN-based architecture consisting of two cascaded modules to learn the two policies. We discover a same-parameter same-decision (SPSD) property for wireless policies defined on sets, revealing that a GNN cannot well learn the optimal scheduling policy when users have similar channels. This motivates us to develop a sequence of GNNs to enhance the scheduling module. Furthermore, by analyzing the SPSD property, we find when linear aggregators in GNNs impede size generalization. Based on the observation, we devise a novel attention mechanism for information aggregation in the precoder module. Simulation results demonstrate that the proposed architecture achieves satisfactory spectral efficiency with short inference time and low training complexity, and is generalizable to the numbers of users, RBs, and antennas at the base station and users.
Distributed photovoltaic (DPV) systems are essential for advancing renewable energy applications and achieving energy independence. Accurate DPV power forecasting can optimize power system planning and scheduling while significantly reducing energy loss, thus enhancing overall system efficiency and reliability. However, solar energy's intermittent nature and DPV systems' spatial distribution create significant forecasting challenges. Traditional methods often rely on costly external data, such as numerical weather prediction (NWP) and satellite images, which are difficult to scale for smaller DPV systems. To tackle this issue, this study has introduced an advanced large language model (LLM)-based time series forecasting framework Time-LLM to improve the DPV power forecasting accuracy and generalization ability. By reprogramming, the framework aligns historical power data with natural language modalities, facilitating efficient modeling of time-series data. Then Qwen2.5-3B model is integrated as the backbone LLM to process input data by leveraging its pattern recognition and inference abilities, achieving a balance between efficiency and performance. Finally, by using a flatten and linear projection layer, the LLM's high-dimensional output is transformed into the final forecasts. Experimental results indicate that Time-LLM outperforms leading recent advanced time series forecasting models, such as Transformer-based methods and MLP-based models, achieving superior accuracy in both short-term and long-term forecasting. Time-LLM also demonstrates exceptional adaptability in few-shot and zero-shot learning scenarios. To the best of the authors' knowledge, this study is the first attempt to explore the application of LLMs to DPV power forecasting, which can offer a scalable solution that eliminates reliance on costly external data sources and improve real-world forecasting accuracy.