paper_name
stringlengths
4
421
paper_url
stringlengths
21
200
paper_authors
listlengths
0
125
paper_abstract
stringlengths
0
43.4k
paper_code
stringlengths
1
139
conf
stringlengths
6
18
CAGE: Curvature-Aware Gradient Estimation For Accurate Quantization-Aware Training
https://proceedings.mlsys.org/paper_files/paper/2026/hash/c0f1eed3f3ca4b2e19ab639a8426270b-Abstract-Conference.html
[ "Soroush Tabesh", "Mher Safaryan", "Andrei Panferov", "Alexandra Volkova", "Dan Alistarh" ]
Despite significant work on low-bit quantization-aware training (QAT), there is still an accuracy gap between such techniques and native training. To address this, we introduce CAGE (Curvature-Aware Gradient Estimation), a new QAT method that augments the straight-through estimator (STE) gradient with a curvature-aware...
#
MLSys2026
From Tokens to Layers: Redefining Stall-Free Scheduling for MoE Serving with Layered Prefill
https://proceedings.mlsys.org/paper_files/paper/2026/hash/c0f460c6d63599ea870ba9db63dc96a9-Abstract-Conference.html
[ "Gunjun Lee", "Jiwon Kim", "Jaiyoung Park", "Younjoo Lee", "Jung Ho Ahn" ]
Large Language Model (LLM) inference in production must meet stringent service-level objectives for both time-to-first-token (TTFT) and time-between-token (TBT) while maximizing throughput under fixed compute, memory, and interconnect budgets. Modern serving systems adopt stall-free scheduling techniques such as chunke...
#
MLSys2026
AgenticCache: Cache-Driven Asynchronous Planning for Embodied AI Agents
https://proceedings.mlsys.org/paper_files/paper/2026/hash/c66a9db149261435664284a20b6f1d42-Abstract-Conference.html
[ "Hojoon Kim", "Yuheng Wu", "Thierry Tambe" ]
Embodied AI agents increasingly rely on large language models (LLMs) for planning, yet per-step LLM calls impose severe latency and cost. In this paper, we show that embodied tasks exhibit strong plan locality, where the next plan is largely predictable from the current one. Building on this, we introduce AgenticCache,...
https://github.com/hojoonleokim/MLSys26_AgenticCache
MLSys2026
BLASST: Dynamic BLocked Attention Sparsity via Softmax Thresholding
https://proceedings.mlsys.org/paper_files/paper/2026/hash/c6ee784cbe46d854843e4c883a3321ef-Abstract-Conference.html
[ "Jiayi Yuan", "Cameron Shinn", "Kai Xu", "Jingze Cui", "George Klimiashvili", "Guangxuan Xiao", "Perkz Zheng", "Bo Li", "Yuxin Zhou", "Zhouhai Ye", "Weijie You", "Tian Zheng", "Dominic Brown", "Pengbo Wang", "Markus Hoehnerbach", "Richard Cai", "Julien Demouth", "John D. Owens", ...
The growing demand for long-context inference capabilities in Large Language Models (LLMs) has intensified the computational and memory bottlenecks inherent to the self-attention mechanism. To address this challenge, we introduce BLASST, a drop-in, dynamic sparse attention mechanism that accelerates inference by using ...
#
MLSys2026
Locality-Aware Beam Scheduling for Efficient Test-Time Compute with a Consumer-grade GPU
https://proceedings.mlsys.org/paper_files/paper/2026/hash/c74b624843218d9b6713fcf299d6d5e4-Abstract-Conference.html
[ "Hsing-Ti Wang", "Hung-Tso Shiao", "Chia-Lin Yang" ]
Large Language Models (LLMs) are central to modern NLP applications, yet their deployment on consumer-grade GPUs is limited by limited memory capacity and bandwidth. In typical single-batch inference on local devices, the key–value (KV) cache occupies only a small fraction of total memory, so prior studies have largely...
#
MLSys2026
PLayer-FL: A Principled Approach to Personalized Layer-wise Cross-Silo Federated Learning
https://proceedings.mlsys.org/paper_files/paper/2026/hash/ca0aa7883e7dff43b6d1e44bb8c75127-Abstract-Conference.html
[ "Ahmed Elhussein", "Florent Pollet", "Gamze Gürsoy" ]
Federated learning (FL) with non-IID data often degrades client performance below local training baselines. Partial FL addresses this by federating only early layers that learn transferable features, but existing methods rely on ad-hoc, architecture-specific heuristics. We first conduct a systematic analysis of layer-w...
#
MLSys2026
Beat the long tail: Distribution-Aware Speculative Decoding for RL Training
https://proceedings.mlsys.org/paper_files/paper/2026/hash/cbc4ab80cd77aa0eb87da062fbcddb46-Abstract-Conference.html
[ "ZELEI SHAO", "Vikranth Srivatsa", "Sanjana Srivastava", "Qingyang Wu", "Alpay Ariyak", "Xiaoxia wu", "Ameen Patel", "Jue Wang", "Percy Liang", "Tri Dao", "Ce Zhang", "Yiying Zhang", "Ben Athiwaratkun", "Chenfeng Xu", "Junxiong Wang" ]
Reinforcement learning(RL) post-training has become essential for aligning large language models (LLMs), yet its efficiency is increasingly constrained by the rollout phase, where long trajectories are generated token by token. We identify a major bottleneck:the long-tail distribution of rollout lengths, where a small ...
#
MLSys2026
VeriMoA: A Mixture-of-Agents Framework for Spec-to-HDL Generation
https://proceedings.mlsys.org/paper_files/paper/2026/hash/cc8c6b9d89f7a898a29f58869b238e46-Abstract-Conference.html
[ "Heng Ping", "Arijit Bhattacharjee", "Peiyu Zhang", "Shixuan Li", "Wei Yang", "Anzhe Cheng", "Xiaole Zhang", "Jesse Thomason", "Ali Jannesari", "Nesreen Ahmed", "Paul Bogdan" ]
Automation of Register Transfer Level (RTL) design can help developers meet increasing computational demands. Large Language Models (LLMs) show promise for Hardware Description Language (HDL) generation, but face challenges due to limited parametric knowledge and domain-specific constraints. While prompt engineering an...
#
MLSys2026
AXLearn: Modular, Hardware-Agnostic Large Model Training
https://proceedings.mlsys.org/paper_files/paper/2026/hash/d1f9e4a9f109b6e8b75ed362736f22ec-Abstract-Conference.html
[ "Mark Lee", "Chang Lan", "Tom Gunter", "John Peebles", "Hanzhi Zhou", "Xuan Kelvin Zou", "Sneha Bangalore", "Chung-Cheng Chiu", "Nan Du", "Xianzhi Du", "Philipp Dufter", "Liang He", "Ruixuan Hou", "Haoshuo Huang", "Dongseong Hwang", "Xiang Kong", "Jinhao Lei", "Tao Lei", "Ethan L...
AXLearn is a production system which facilitates scalable and high-performance training of large deep learning models. Compared to other state-of-the-art deep learning systems, AXLearn has a unique focus on modularity and support for hardware-agnostic training. AXLearn's internal interfaces between software components ...
#
MLSys2026
Beyond the Buzz: A Pragmatic Take on Inference Disaggregation
https://proceedings.mlsys.org/paper_files/paper/2026/hash/d49cee5f3a79d97d719df255689d83d7-Abstract-Conference.html
[ "Tiyasa Mitra", "Ritika Borkar", "Nidhi Bhatia", "Shivam Raj", "Hongkuan Zhou", "Yan Ru Pei", "Vishwanath Venkatesan", "Kyle Kranen", "Ramon Matas", "Dheevatsa Mudigere", "Ritchie Zhao", "Maximilian Golub", "Arpan Dutta", "Suresh Nambi", "Sailaja Madduri", "Dharmesh Jani", "Brian Pha...
As inference scales to multi-node deployments, prefill-decode disaggregation — splitting inference into distinct phases — offers a promising path to improving the throughput-interactivity Pareto frontier. Despite growing enthusiasm and a surge of open-source efforts, large-scale deployment of disaggregated serving rema...
#
MLSys2026
HexiScale: Facilitating Large Language Model Training over Heterogeneous Hardware
https://proceedings.mlsys.org/paper_files/paper/2026/hash/d5a655b8b373737b4f2aea8f78e5e754-Abstract-Conference.html
[ "Ran Yan", "YOUHE JIANG", "Xiaonan Nie", "Fangcheng Fu", "Bin CUI", "Binhang Yuan" ]
Training large language models (LLMs) is a computationally intensive task, which is typically conducted in data centers with homogeneous high-performance GPUs. In this paper, we explore an alternative approach by deploying training computations across heterogeneous GPUs to enable better flexibility and efficiency for h...
#
MLSys2026
veScale-FSDP: Flexible and High-Performance FSDP at Scale
https://proceedings.mlsys.org/paper_files/paper/2026/hash/d5cd70b708f726737e2ebace18c3f71b-Abstract-Conference.html
[ "Zezhou Wang", "Youjie Li", "Zhiqi Lin", "Jiacheng Yang", "Cong Xie", "Guanyu Feng", "Zheng Zhong", "Ziyue Huang", "Hongyu Zhu", "Zhi Zhang", "Yanghua Peng", "Xin Liu" ]
Fully Sharded Data Parallel (FSDP), also known as Zero Redundancy Optimizer (ZeRO), is widely used for large-scale model training, because of its memory efficiency and minimal intrusion on model code. However, existing FSDP systems rely on fixed element-wise or row-wise sharding formats that conflict with block-struct...
#
MLSys2026
Cost-aware Duration Prediction for Software Upgrades in Datacenters
https://proceedings.mlsys.org/paper_files/paper/2026/hash/d84c0dd9b1bfeee361f3268dcaebf849-Abstract-Conference.html
[ "Yi Ding", "Aijia Gao", "Thibaud Ryden", "Michal Sedlak", "Essam Ewaisha", "Igor Marnat", "Henry Hoffmann" ]
Software upgrades are critical to maintaining server reliability in datacenters. While job duration prediction and scheduling have been extensively studied, the unique challenges posed by software upgrades remain largely under-explored. This paper presents the first in-depth investigation into software upgrade scheduli...
#
MLSys2026
GhostServe: A Lightweight Checkpointing System in the Shadow for Fault-Tolerant LLM Serving
https://proceedings.mlsys.org/paper_files/paper/2026/hash/d99e8e80a6c41e148db686918dd7eab3-Abstract-Conference.html
[ "Shakya Jayakody", "Youpeng Zhao", "Chinmay Dhanraj Nehate", "Jun Wang" ]
The rise of million-token, agent-based applications has placed unprecedented demands on large language model (LLM) inference services. The long-running nature of these tasks increases their susceptibility to hardware and software faults, leading to costly job failures, wasted resources, and degraded user experience. T...
#
MLSys2026
Learning from Less: Measuring the Effectiveness of RLVR in Low Data and Compute Regimes
https://proceedings.mlsys.org/paper_files/paper/2026/hash/da85790fb1cb4f11f431648455c561b5-Abstract-Conference.html
[ "Justin Bauer", "Thomas Walshe", "Derek Pham", "Harit Vishwakarma", "Armin Parchami", "Frederic Sala", "Paroma Varma" ]
Fine-tuning Large Language Models (LLMs) typically relies on large quantities of high-quality annotated data, or questions with well-defined ground truth answers in the case of Reinforcement Learning with Verifiable Rewards (RLVR). While previous work has explored the benefits to model reasoning capabilities by scaling...
#
MLSys2026
db-SP: Accelerating Sparse Attention for Visual Generative Models with Dual-Balanced Sequence Parallelism
https://proceedings.mlsys.org/paper_files/paper/2026/hash/db988b089d8d97d0f159c15ed0be6a71-Abstract-Conference.html
[ "Siqi Chen", "Ke Hong", "Tianchen Zhao", "Ruiqi Xie", "Zhenhua Zhu", "Xudong Zhang", "Yu Wang" ]
Scaling Diffusion Transformer (DiT) inference via sequence parallelism is critical for reducing latency in visual generation, but is severely hampered by workload imbalance when applied to models employing block-wise sparse attention. The imbalance stems from the inherent variation in sparsity across attention heads an...
#
MLSys2026
Charon: A Unified and Fine-Grained Simulator for Large-Scale LLM Training and Inference
https://proceedings.mlsys.org/paper_files/paper/2026/hash/dbc8ce0fdfcd55172d73fb05dbae07fc-Abstract-Conference.html
[ "Mengtian Yang", "Zhekun Zhang", "Mingheng Wu", "Jianwen Yan", "Hanshi Sun", "Li-Wen Chang" ]
Deploying large-scale LLM training and inference with optimal performance is exceptionally challenging due to a complex design space of parallelism strategies, system optimizations, and hardware configurations. Accurate and rapid performance simulation is critical for guiding optimization efforts and system studies by ...
#
MLSys2026
ReSpec: Towards Optimizing Speculative Decoding in Reinforcement Learning Systems
https://proceedings.mlsys.org/paper_files/paper/2026/hash/dcb38c6ad7911842ab31081be9540b89-Abstract-Conference.html
[ "Qiaoling Chen", "Zijun Liu", "Peng Sun", "Shenggui Li", "Guoteng Wang", "Ziming Liu", "Yonggang Wen", "Siyuan Feng", "Tianwei Zhang" ]
Adapting large language models (LLMs) via reinforcement learning (RL) is often bottlenecked by the generation stage, which can consume over 75% of the training time. Speculative decoding (SD) accelerates autoregressive generation in serving systems, but its behavior under RL training remains largely unexplored. We iden...
#
MLSys2026
fabric-lib: RDMA Point-to-Point Communication for LLM Systems
https://proceedings.mlsys.org/paper_files/paper/2026/hash/dea9b4b6f55ae611c54065d6fc750755-Abstract-Conference.html
[ "Nandor Licker", "Kevin Hu", "Vladimir Zaytsev", "Lequn Chen" ]
Emerging Large Language Model (LLM) system patterns, such as disaggregated inference, Mixture-of-Experts (MoE) routing, and asynchronous reinforcement fine-tuning, require flexible point-to-point communication beyond simple collectives. Existing implementations are locked to specific Network Interface Controllers (NICs...
https://github.com/perplexityai/pplx-garden/
MLSys2026
Unleashing Scalable Context Parallelism for Foundation Models Pre-Training via FCP
https://proceedings.mlsys.org/paper_files/paper/2026/hash/e01c431bbb83153632c0dcfaf8ccda0a-Abstract-Conference.html
[ "Yilong Zhao", "Xiaonan Nie", "Kan Zhu", "Shuang Ma", "Zhichao Lai", "Hongxiang Hao", "Yang Zhou", "Baris Kasikci", "Ion Stoica" ]
Context parallelism (CP) has been widely adopted to support the growing context length in foundation model pretraining. However, existing designs fail to handle the large variation in sequence length from training datasets, resulting in suboptimal performance. These methods often over-shard short sequences, leading to ...
#
MLSys2026
Hawkeye: Reproducing GPU-Level Non-Determinism
https://proceedings.mlsys.org/paper_files/paper/2026/hash/e217c271a57c365a246b0ad39e668ba8-Abstract-Conference.html
[ "Erez Badash", "Dan Boneh", "Ilan Komargodski", "Megha Srivastava" ]
We present Hawkeye, a system for analyzing and reproducing GPU-level arithmetic operations. Using our framework, anyone can re-execute on a CPU the exact matrix multiplication operations underlying a machine learning model training or inference workflow that was executed on an NVIDIA GPU, without any precision loss. Th...
#
MLSys2026
LEANN: A Low-Storage Overhead Vector Index
https://proceedings.mlsys.org/paper_files/paper/2026/hash/e27ea0cd50b798ff8942caf9203f0992-Abstract-Conference.html
[ "Yichuan Wang", "Zhifei Li", "Shu Liu", "Yongji Wu", "Ziming Mao", "Yilong Zhao", "Xiao Yan", "Zhiying Xu", "Yang Zhou", "Ion Stoica", "Sewon Min", "Matei Zaharia", "Joseph E. Gonzalez" ]
Embedding-based vector search underpins many important applications, such as recommendation and retrieval-augmented generation (RAG). It relies on vector indices to enable efficient search. However, these indices require storing high-dimensional embeddings and large index metadata, whose total size can be several times...
#
MLSys2026
Kitty: Accurate and Efficient 2-bit KV Cache Quantization with Dynamic Channel-wise Precision Boost
https://proceedings.mlsys.org/paper_files/paper/2026/hash/e4d8d1b5120be349d3fff8878650cf45-Abstract-Conference.html
[ "Haojun Xia", "Xiaoxia wu", "Jisen Li", "Tsai-chuan Wu", "Junxiong Wang", "Jue Wang", "Chenxi Li", "Aman Singhal", "Alay Dilipbhai Shah", "Alpay Ariyak", "Donglin Zhuang", "Zhongzhu Zhou", "Ben Athiwaratkun", "Zhen Zheng", "Shuaiwen Song" ]
The KV cache is a dominant memory bottleneck for LLM inference. While 4-bit KV quantization preserves accuracy, 2-bit often degrades it, especially on long-context reasoning. We close this gap via an algorithm–system co-design for mixed-precision KV caching: Kitty. On the algorithm side, extensive experiments show that...
https://github.com/Summer-Summer/Kitty
MLSys2026
DriftBench: Measuring and Predicting Infrastructure Drift in LLM Serving Systems
https://proceedings.mlsys.org/paper_files/paper/2026/hash/ea0b5818ae9255ee1fb1e3b4442d2ffe-Abstract-Conference.html
[ "Gianluigi Vitale" ]
Production LLM deployments lack systematic methods to assess output consistency risks when infrastructure changes. We present DriftBench, a measurement and prediction framework comprising 236,985 prompt-response pairs across 105 configurations spanning 5 models, 4 GPU platforms, 3 frameworks, 3 precisions. We develop t...
#
MLSys2026
IntAttention: A Fully Integer Attention Pipeline for Efficient Edge Inference
https://proceedings.mlsys.org/paper_files/paper/2026/hash/ea5ffdf7da91256ecd2770f9fd2dade9-Abstract-Conference.html
[ "Wanli Zhong", "Haibo Feng", "Zirui Zhou", "Hanyang Peng", "Shiqi Yu" ]
Deploying Transformer models on edge devices is limited by latency and energy budgets. While INT8 quantization effectively accelerates the primary matrix multiplications, it exposes the softmax-related path as the dominant bottleneck. This stage incurs a costly dequantize -> softmax -> requantize detour, which can acco...
https://github.com/WanliZhong/IntAttention
MLSys2026
BEAM: Joint Resource–Power Optimization for Energy-Efficient LLM Inference under SLO contraints
https://proceedings.mlsys.org/paper_files/paper/2026/hash/eb3c42ddfa16d8421fdba13528107cc1-Abstract-Conference.html
[ "Hyunjae Lee", "Sangjin Choi", "Seungjae Lim", "Youngjin Kwon" ]
Large Language Model (LLM) serving is rapidly becoming one of the most power-intensive workloads in modern datacenters. Unlike training, where throughput dominates, inference must satisfy strict per-request latency targets such as Time-to-First-Token (TTFT) and Time-Between-Tokens (TBT). Once an SLO is met, the remaini...
#
MLSys2026
BOute: Cost-Efficient LLM Serving with Heterogeneous LLMs and GPUs via Multi-Objective Bayesian Optimization
https://proceedings.mlsys.org/paper_files/paper/2026/hash/ed1d3d4c64dc1b95332a8cde3f2a0bdf-Abstract-Conference.html
[ "YOUHE JIANG", "Fangcheng Fu", "Eiko Yoneki" ]
The rapid growth of large language model (LLM) deployments has made cost-efficient serving systems essential. Recent efforts to enhance system cost-efficiency adopt two main perspectives: (\textbf{\underline{i}}) An \textit{algorithmic} perspective that exploits heterogeneous model capabilities to route simpler queries...
#
MLSys2026
Practical Adversarial Multi-Armed Bandits with Sublinear Runtime
https://proceedings.mlsys.org/paper_files/paper/2026/hash/efbd571f139d26604e53fe2760e2c073-Abstract-Conference.html
[ "Kasper Overgaard Mortensen", "Ama Bembua Bainson", "Mathias Ravn Tversted", "Kristoffer Strube Græm", "Renata Borovica-Gajic", "Andrea Paudice", "Davide Mottin", "Panagiotis Karras" ]
We study the Multi-Armed Bandit problem in nonstationary adversarial environments, where the identity of the optimal arm can change over time due to shifts in the loss sequence. Motivated by applications such as physical design tuning in database systems, we focus on settings with a very large number of arms and seek p...
#
MLSys2026
ADR: AN AGENTIC DETECTION SYSTEMFORENTERPRISE AGENTIC AI SECURITY
https://proceedings.mlsys.org/paper_files/paper/2026/hash/f03cb785864596fa5901f1359d23fd81-Abstract-Conference.html
[ "Chenning Li", "Pan Hu", "Justin Xu", "Baris Ozbas", "Olivia Liu", "Caroline Van", "Manxue Li", "Wei Zhou", "Mohammad Alizadeh", "Pengyu Zhang", "KK Sriramadhesikan", "Ming Zhang" ]
We present the Agentic AI Detection and Response (ADR) system, the first large-scale, production-proven enterprise framework for securing AI agents operating through the Model Context Protocol (MCP). We identify three persistent challenges in this domain: (1) limited observability– existing Endpoint Detection and Respo...
#
MLSys2026
TriInfer: Hybrid EPD Disaggregation for Efficient Multimodal Large Language Model Inference
https://proceedings.mlsys.org/paper_files/paper/2026/hash/f068c65585985c25c17f221390774ec7-Abstract-Conference.html
[ "Xianzhe Dong", "Tongxuan Liu", "Yuting Zeng", "Weizhe Huang", "Xiaoyang Zhao", "Siyu Wu", "Liangyu Liu", "Yang Liu", "Yu Wu", "Hailong Yang", "Ke Zhang", "Jing Li" ]
Existing MLLM inference systems are typically designed based on the architecture of language models, coupling image processing and language processing. This design struggles to accommodate the heterogeneous demands of different stages in terms of computational resources, memory access patterns, and service-level object...
https://github.com/dongxianzhe/triinfer
MLSys2026
FreeScale: Distributed Training for Sequence Recommendation Models with Minimal Scaling Cost
https://proceedings.mlsys.org/paper_files/paper/2026/hash/f189e7580acad0fc7fd45405817ddee3-Abstract-Conference.html
[ "Chenhao Feng", "Haoli Zhang", "Shakhzod Ali-zade", "Yanli Zhao", "Liang Luo", "Jennifer Cao", "Lisen Deng", "Siqiao Chen", "Chenyu Zhao", "Tristan Rice", "Daniel Johnson", "Min Si", "Tiantu Xu", "Yi Zhang", "Evgenii Kolpakov", "Siqi Yan", "Chuanhao Zhuge", "Min Ni", "Bi Xue", ...
Modern industrial Deep Learning Recommendation Models typically extract user preferences through the analysis of sequential interaction histories, subsequently generating predictions based on these derived interests. The inherent heterogeneity in data characteristics frequently result in substantial under-utilization o...
#
MLSys2026
FLoRIST: Singular Value Thresholding for Efficient and Accurate Federated Fine-Tuning of Large Language Models
https://proceedings.mlsys.org/paper_files/paper/2026/hash/f38cb4cf9a5eaa92b3cfa481832719c6-Abstract-Conference.html
[ "Hariharan Ramesh", "Jyotikrishna Dass" ]
Integrating Low-Rank Adaptation (LoRA) into federated learning offers a promising solution for parameter-efficient fine-tuning of Large Language Models (LLMs) without sharing local data. However, several methods designed for federated LoRA present significant challenges in balancing communication efficiency, model accu...
#
MLSys2026
Dataflow Is All You Need
https://proceedings.mlsys.org/paper_files/paper/2026/hash/f502981cbe221d857ad409450a7917c3-Abstract-Conference.html
[ "Darshan Gandhi", "Pushkar Nandkar", "David Koeplinger", "Nasim Farahini", "Romy Tsoupidi", "Samuel Rydh", "Matheen Musaddiq", "Tuowen Zhao", "Reid Goodbar", "Nathan Sheeley", "Leon Zhang", "Matthew Shaffer", "John Long", "Han Wang", "Angela Wang", "Arjun Sabnis", "Joshua Brot", "Y...
The autoregressive decode phase of token generation is often a major performance bottleneck in modern AI workflows due to its memory-bandwidth-bound characteristics, which are amplified by powerful open-source models with large context windows and techniques such as chain-of-thought reasoning. Popular GPU architectures...
#
MLSys2026
DreamDDP: Accelerating Low-Bandwidth Geo-Distributed LLM Training with Layer-wise Partial Synchronization
https://proceedings.mlsys.org/paper_files/paper/2026/hash/f5c530d3f287c43dccfca298698e9254-Abstract-Conference.html
[ "zhenheng Tang", "Zichen TANG", "Junlin Huang", "Xinglin Pan", "Rudan Yan", "Yuxin Wang", "Amelie Chi Zhou", "Shaohuai Shi", "Xiaowen Chu", "Bo Li" ]
Scaling up training large language models (LLMs) in computing and data perspectives motivates distributed training across different geo-distributed data centers. Communication in geo-distributed data parallel training (DDP) with stochastic gradient descent (S-SGD) is the main bottleneck in low-bandwidth environments. R...
#
MLSys2026
Using Span Queries to Optimize Cache and Attention Locality
https://proceedings.mlsys.org/paper_files/paper/2026/hash/f5d77f1e501e0496377d8b68c8e81a48-Abstract-Conference.html
[ "Paul Castro", "Nick Mitchell", "Nathan Ordonez", "Thomas Parnell", "Mudhakar Srivatsa", "Antoni Viros-i-Martin" ]
Clients are evolving beyond chat completion, and now include a variety of innovative inference-time scaling and deep reasoning techniques. At the same time, inference servers remain heavily optimized for chat completion and thus largely use linear KV cache strategies. Prior work has shown that large improvements to KV ...
#
MLSys2026
Privatar: Scalable Privacy-preserving Multi-user VR via Secure Offloading
https://proceedings.mlsys.org/paper_files/paper/2026/hash/fb1d9c3fc2161e12aa71cdcab74b9d2c-Abstract-Conference.html
[ "Jianming Tong", "Hanshen Xiao", "Krishna Kumar Nair", "Hao Kang", "Ashish Sirasao", "Ziqi Zhang", "G. Edward Suh", "Tushar Krishna" ]
Multi-user virtual reality (VR) applications such as football and concert experiences rely on real-time avatar reconstruction to enable immersive interaction. However, rendering avatars for numerous participants on each headset incurs prohibitive computational overhead, fundamentally limiting scalability. This work int...
https://github.com/georgia-tech-synergy-lab/Privatar
MLSys2026
Machine Learning Fleet Efficiency: Improving TPU Systems at Scale with ML Productivity Goodput
https://proceedings.mlsys.org/paper_files/paper/2026/hash/fbe2b2f74a2ece8070d8fb073717bda6-Abstract-Conference.html
[ "Arissa Wongpanich", "Tayo Oguntebi", "Jose Baiocchi Paredes", "Yu Wang", "Phitchaya Phothilimthana", "Ritwika Mitra", "Zongwei Zhou", "Naveen Kumar", "Vijay Janapa Reddi" ]
Machine learning (ML) infrastructures operating at warehouse scale present unique performance characterization challenges beyond traditional high-performance computing metrics. This paper introduces a systematic framework for analyzing ML fleet efficiency, demonstrated on Google's production TPU infrastructure comprisi...
#
MLSys2026
ParallelKittens: Systematic and Practical Simplification of Multi-GPU AI Kernels
https://proceedings.mlsys.org/paper_files/paper/2026/hash/ff997469ac66cf893c4183efeb22212a-Abstract-Conference.html
[ "Stuart H. Sul", "Simran Arora", "Benjamin F. Spector", "Christopher Ré" ]
Inter-GPU communication has become a major bottleneck for modern AI workloads as models scale and improvements in hardware compute throughput outpace improvements in interconnect bandwidth. Existing systems mitigate this through compute-communication overlap but often fail to meet theoretical peak performance across he...
#
MLSys2026
41st Annual Symposium on Logic in Computer Science, LICS 2026, Lisbon, Portugal, July 20-23, 2026
https://www.dagstuhl.de/dagpub/978-3-95977-434-5
[ "Claudia Faggian", "Joost-Pieter Katoen" ]
#
LICS2026
Front Matter, Table of Contents, Preface, Conference Organization
https://doi.org/10.4230/LIPIcs.LICS.2026.0
[]
Front Matter, Table of Contents, Preface, Conference Organization
#
LICS2026
A Rational Defense of Reasonable Reflection (Invited Paper)
https://doi.org/10.4230/LIPIcs.LICS.2026.1
[ "Nada Amin" ]
Computational reflection is the ability of a system to inspect and modify itself. It has resisted static reasoning, especially in reflective languages, where semantics can change during program execution. We argue that reflection should be revisited in light of progress in verification and large language models. By req...
#
LICS2026
Formal Verification of Security Protocols: 25 Years of ProVerif (Invited Paper)
https://doi.org/10.4230/LIPIcs.LICS.2026.2
[ "Stéphanie Delaune" ]
Cryptographic protocols are essential to secure communication but remain difficult to design correctly, motivating the need for rigorous verification methods. This paper provides a brief overview of symbolic verification techniques, focusing on the evolution and impact of the tool ProVerif over the past 25 years. We di...
#
LICS2026
Complete ω-Regular Supermartingale Certificates
https://doi.org/10.4230/LIPIcs.LICS.2026.3
[ "Alessandro Abate", "Mirco Giacobbe", "Sergey Ichtchenko", "Diptarko Roy" ]
Controlled commands - computations whose execution depends on a separate input - play a central role in reversible Boolean circuits and quantum circuits. However, existing formalisms typically treat control only implicitly, entangled with other aspects of computation. From a semantic perspective, control is most natura...
#
LICS2026
A Convenient Fibration for Dependently-Typed Probability Theory
https://doi.org/10.4230/LIPIcs.LICS.2026.4
[ "Danel Ahman", "Ohad Kammar", "Rasmus Ejlers Møgelberg" ]
We describe semantic structures relevant for interpreting dependent types for statistical and probabilistic modelling. Our development extends the theory of quasi-Borel spaces (qbses) of Staton et. al, which support simply-typed, higher-order probability theory with continuous distributions. It is well-known that qbses...
#
LICS2026
A Complexity Bound for Determinisation of Min-Plus Weighted Automata
https://doi.org/10.4230/LIPIcs.LICS.2026.5
[ "Shaull Almagor", "Guy Arbel", "Sarai Sheinvald" ]
The determinisation problem for min-plus (tropical) weighted automata was recently shown to be decidable. However, the proof is purely existential, relying on several non-constructive arguments. Our contribution in this work is twofold: first, we present the first complexity bound for this problem, showing it is primit...
#
LICS2026
Induction and Recursion Principles in a Higher-Order Quantitative Logic for Probability
https://doi.org/10.4230/LIPIcs.LICS.2026.6
[ "Giorgio Bacci", "Rasmus Ejlers Møgelberg" ]
Quantitative logic reasons about the degree to which formulas are satisfied. This paper studies the fundamental reasoning principles of higher-order quantitative logic and their application to reasoning about probabilistic programs and processes. We construct an affine calculus for $1$-bounded complete metric spaces an...
#
LICS2026
On Variable-Bounded Non-Linear Expansions of Presburger Arithmetic
https://doi.org/10.4230/LIPIcs.LICS.2026.7
[ "Piotr Bacik", "Joris Nieuwveld", "Joël Ouaknine", "Mihir Vahanwala", "Madhavan Venkatesh", "Emil Rugaard Wieser" ]
We consider expansions of Presburger arithmetic with families of monadic polynomial predicates. (Examples of such predicates are the set of perfect squares, or the set of integers of the form $2n^3-5n+3$, etc.) Although the full attendant first-order theories are well known to be undecidable, very little is known when ...
#
LICS2026
On the Subspace Orbit Problem and the Simultaneous Skolem Problem
https://doi.org/10.4230/LIPIcs.LICS.2026.8
[ "Piotr Bacik", "Anton Varonka" ]
The Orbit Problem asks whether the orbit of a point under a matrix reaches a given target set. When the target is a single point, the problem was shown to be decidable in polynomial time by Kannan and Lipton. This decidability result was later extended by Chonev et al. to targets of dimension 3 (in arbitrary ambient di...
#
LICS2026
A Machine-Independent, Log-Sensitive Space-Cost Measure for the Weak Lambda-Calculus
https://doi.org/10.4230/LIPIcs.LICS.2026.9
[ "Thibaut Balabonski" ]
#
LICS2026
Hypersequent Calculi Have Ackermann Complexity
https://doi.org/10.4230/LIPIcs.LICS.2026.10
[ "A. R. Balasubramanian", "Vitor Greati", "Revantha Ramanayake" ]
For substructural logics with contraction or weakening admitting cut-free sequent calculi, proof search was analyzed using well-quasi-orders on ℕ^d (Dickson’s lemma), yielding Ackermann upper bounds via controlled bad-sequence arguments. For hypersequent calculi, that argument lifted the ordering to the powerset, since...
#
LICS2026
The Complexity of Nested Reset Counter Systems
https://doi.org/10.4230/LIPIcs.LICS.2026.11
[ "A. R. Balasubramanian", "Franzisco Schmidt" ]
Nested counter systems (NCS) are a generalization of counter systems to higher-order counters. Here, a higher-order counter is allowed to have other (lower-order) counters as elements, instead of just a number. Such systems can be viewed as working on trees, where the height of the tree naturally corresponds to the hig...
#
LICS2026
Unbounded Data Nesting for Loops in Higher-Order Programs
https://doi.org/10.4230/LIPIcs.LICS.2026.12
[ "Adriana Baldacchino", "Andrzej S. Murawski" ]
We study contextual interactions in an ML-like language equipped with general references and continuations, focusing on the reachability and approximation problems. Previous work addressed higher-order programs with first-order references in the absence of loops using automata over nested data; however, extending these...
#
LICS2026
Towards Infinite PCSP: A Dichotomy for Monochromatic Cliques
https://doi.org/10.4230/LIPIcs.LICS.2026.13
[ "Demian Banakh", "Alexey Barsukov", "Tamio-Vesa Nakajima" ]
The logic MMSNP is a well-studied fragment of Existential Second-Order logic that, from a computational perspective, captures finite-domain Constraint Satisfaction Problems (CSPs) modulo polynomial-time reductions. At the same time, MMSNP contains many problems that are expressible as $\omega$-categorical CSPs but not ...
#
LICS2026
Quantum Control and General Recursion Beyond the Unitary Case
https://doi.org/10.4230/LIPIcs.LICS.2026.14
[ "Kathleen Barsse", "Romain Péchoux", "Simon Perdrix" ]
Coherent control, aka quantum control, is a central concept in quantum computing that is attracting increasing attention from both the quantum foundations and quantum software communities. Defining coherent control in the presence of recursion and measurement has long been known to be a major challenge. In particular, ...
#
LICS2026
Complete Relational Logic for Infinite-Dimensional Quantum Programs with Unbounded Assertions
https://doi.org/10.4230/LIPIcs.LICS.2026.15
[ "Gilles Barthe", "Minbo Gao", "Jam Kabeer Ali Khan", "Matthijs Muis", "Ivan Renison", "Keiya Sakabe", "Michael Walter", "Yingte Xu", "Tianshi Yu", "Li Zhou" ]
We present sound and complete relational program logics for infinite-dimensional quantum and classical-quantum programs. The logics model assertions as self-adjoint unbounded linear relations, which simultaneously support quantitative and qualitative reasoning. Our main theoretical results include new convergence theor...
#
LICS2026
A Computer Formalisation of the Serre Finiteness Theorem
https://doi.org/10.4230/LIPIcs.LICS.2026.16
[ "Reid Barton", "Axel Ljungström", "Owen Milner", "Anders Mörtberg" ]
#
LICS2026
The Algebra of Iterative Constructions
https://doi.org/10.4230/LIPIcs.LICS.2026.17
[ "Kevin Batz", "Benjamin Lucien Kaminski", "Lucas Kehrer", "Gerwin Klein", "Todd Schmid", "Henning Urbat" ]
Fixed points are a recurring theme in computer science and are often constructed as limits of suitably seeded fixed point iterations. We present the algebra of iterative constructions (AIC) - a purely algebraic approach to reasoning about fixed point iterations of continuous endomaps on complete lattices. AIC allows de...
#
LICS2026
Minimization of Streaming Transducers
https://doi.org/10.4230/LIPIcs.LICS.2026.18
[ "Christian Bianchini", "Gabriele Puppis" ]
We provide general criteria for the existence of minimal models of streaming transducers, namely devices that read an input word and produce an output value by iteratively updating an internal memory. This abstract model subsumes classical (sub)sequential transducers (Schützenberger), streaming string-to-string transdu...
#
LICS2026
Reachability in VASS Extended with Integer Counters
https://doi.org/10.4230/LIPIcs.LICS.2026.19
[ "Clotilde Bizière", "Wojciech Czerwinski", "Roland Guttenberg", "Jérôme Leroux", "Vincent Michielini", "Lukasz Orlikowski", "Antoni Puch", "Henry Sinclair-Banks" ]
We consider a variant of VASS extended with integer counters, denoted VASS+ℤ. These are automata equipped with ℕ- and ℤ-counters; the ℕ-counters are required to remain nonnegative and the ℤ-counters do not have this restriction. We study the complexity of the reachability problem for VASS+ℤ when the number of ℕ-counter...
#
LICS2026
On the Computational Power of Extensional ESO
https://doi.org/10.4230/LIPIcs.LICS.2026.20
[ "Manuel Bodirsky", "Santiago Guzmán-Pro" ]
Extensional ESO is a fragment of existential second-order logic (ESO) that captures the following family of problems. Given a fixed ESO sentence Ψ and an input structure A the task is to decide whether there is an extension B of A that satisfies the first-order part of Ψ, i.e., a structure B such that R^A ⊆ R^B for eve...
#
LICS2026
Automata for MSO over Infinite Trees with Quantification over Borel Sets of Branches
https://doi.org/10.4230/LIPIcs.LICS.2026.21
[ "Mikolaj Bojanczyk", "Antonio Casares", "Sven Manthe", "Pawel Parys" ]
Rabin’s Tree Theorem says that the {mso} theory of the infinite binary tree 2^* is decidable. Shelah showed that MSO logic becomes undecidable if this tree is extended to 2^{≤ω}, i.e. by allowing quantification over sets of infinite branches. A longstanding open problem is whether the decidability can be recovered in 2...
#
LICS2026
Low Rank MSO
https://doi.org/10.4230/LIPIcs.LICS.2026.22
[ "Mikolaj Bojanczyk", "Michal Pilipczuk", "Wojciech Przybyszewski", "Marek Sokolowski", "Giannos Stamoulis" ]
We introduce a new logic for describing properties of graphs, which we call low rank MSO. This is the fragment of monadic second-order logic in which set quantification is restricted to vertex sets of bounded cutrank. We prove the following statements about the expressive power of low rank MSO. - Over any class of grap...
#
LICS2026
Dicey Games: Shared Sources of Randomness in Distributed Systems
https://doi.org/10.4230/LIPIcs.LICS.2026.23
[ "Léonard Brice", "Thomas A. Henzinger", "K. S. Thejaswini" ]
Consider a 4-player version of Matching Pennies where a team of three players competes against the Devil. Each player simultaneously says "Heads" or "Tails". The team wins if all four choices match; otherwise the Devil wins. If all team players randomise independently, they win with probability 1/8; if all players shar...
#
LICS2026
Layered Automata: A Canonical Model for Automata over Infinite Words
https://doi.org/10.4230/LIPIcs.LICS.2026.24
[ "Antonio Casares", "Christof Löding", "Igor Walukiewicz" ]
We introduce layered automata, a subclass of alternating parity automata that generalises deterministic automata. Assuming a consistency property, these automata are history deterministic and 0-1 probabilistic. We show that every omega-regular language is recognised by a unique minimal consistent layered automaton, and...
#
LICS2026
Lazy Intermediate Representations for Algebraic Effects
https://doi.org/10.4230/LIPIcs.LICS.2026.25
[ "Simon Castellan", "Hugo Paquet" ]
A lazy program interpreter postpones computation until the result is actually needed. This is typically more efficient than an eager (or call-by-value) interpreter, but a concern is that the semantics is not generally preserved. We propose a new semantic analysis of lazy evaluation that relies on a subtle combination o...
#
LICS2026
The Size of Interpolants in Modal Logics
https://doi.org/10.4230/LIPIcs.LICS.2026.26
[ "Balder ten Cate", "Louwe B. Kuijer", "Frank Wolter" ]
We start a systematic investigation of the size of Craig interpolants, uniform interpolants, and strongest implicates for (quasi-)normal modal logics. Our main upper bound states that for tabular modal logics, the computation of strongest implicates can be reduced in polynomial time to uniform interpolant computation i...
#
LICS2026
Eliminating Reversals from Cubical Type Theories
https://doi.org/10.4230/LIPIcs.LICS.2026.27
[ "Evan Cavallo", "Christian Sattler" ]
Cubical type theories are designed around an abstract unit interval from which types of paths, used to represent equalities, are defined. Varying the operations available on this interval yields different type theories. A reversal is an involutive operator on the interval that swaps its two endpoints. We show that for ...
#
LICS2026
A Complete Equational Theory for Real-Clifford+CH Quantum Circuits
https://doi.org/10.4230/LIPIcs.LICS.2026.28
[ "Alexandre Clément" ]
We introduce a complete equational theory for the fragment of quantum circuits generated by the real Clifford gates plus the two-qubit controlled-Hadamard gate. That is, we give a simple set of equalities between circuits of this fragment, and prove that any other true equation can be derived from these. This is the fi...
#
LICS2026
Commutative Algebras of Series
https://doi.org/10.4230/LIPIcs.LICS.2026.29
[ "Lorenzo Clemente" ]
We consider a large family of product operations of formal power series in noncommuting indeterminates, the classes of automata they define, and the respective equivalence problems. A P-product of series is defined coinductively by a polynomial product rule P, which gives a recursive recipe to build the product of two ...
#
LICS2026
The Uniformisation of Monadic Second-Order Logic over Countable Ordinals
https://doi.org/10.4230/LIPIcs.LICS.2026.30
[ "Thomas Colcombet", "Alexander Rabinovich" ]
We study the uniformisation problem for monadic second-order logic (MSO) over countable ordinal chains. Given a formula defining a relation between subsets of the input structure, the question is whether there exists a formula that defines a function selecting, for every set in the domain of the relation, a unique set ...
#
LICS2026
Constructive Higher Sheaf Models with Applications to Synthetic Mathematics
https://doi.org/10.4230/LIPIcs.LICS.2026.31
[ "Thierry Coquand", "Jonas Höfer", "Christian Sattler" ]
There have recently been several developments in synthetic mathematics using extensions of dependent type theory with univalence and higher inductive types: simplicial homotopy type theory, synthetic algebraic geometry and synthetic Stone duality. We provide a foundation of higher sheaf models of type theory in a const...
#
LICS2026
A Categorical Account of the Metropolis-Hastings Algorithm
https://doi.org/10.4230/LIPIcs.LICS.2026.32
[ "Rob Cornish", "Andi Q. Wang" ]
Metropolis-Hastings (MH) is a foundational Markov chain Monte Carlo (MCMC) algorithm. In this paper, we ask whether it is possible to formulate and analyse MH in terms of categorical probability, using a recent involutive framework for MH-type procedures as a concrete case study. We show how basic MCMC concepts such as...
#
LICS2026
Interpreting De Finetti's Theorem in the Category of Integrable Cones
https://doi.org/10.4230/LIPIcs.LICS.2026.33
[ "Raphaëlle Crubillé" ]
We establish a connection between two results in the literature on probabilistic semantics: a formulation of De Finetti’s theorem in the language of category theory due to Jacobs and Staton, and the generic construction of the free exponential of Linear Logic by Melliès et al, that has been instantiated in the model of...
#
LICS2026
On Higher-Order Probabilistic Verification via the Weighted Relational Model of Linear Logic
https://doi.org/10.4230/LIPIcs.LICS.2026.34
[ "Ugo Dal Lago", "Guido Fiorillo", "Paolo Pistone" ]
The problem of determining whether a probabilistic program terminates almost surely (i.e. with probability one) is undecidable, and actually Π⁰₂-complete. For this reason, a growing literature has explored classes of programs for which this and related problems can be shown (semi-)decidable. In this work we consider th...
#
LICS2026
Star Complexity of Parikh Images of Languages over Infinite Alphabets
https://doi.org/10.4230/LIPIcs.LICS.2026.35
[ "Yoav Danieli" ]
It has been conjectured that the Parikh (commutative) image of every language over an infinite alphabet recognized by an automaton with registers is defined by a rational expression. This conjecture is known to hold for all languages recognized by one-register automata. We refine this result by proving that the star-he...
#
LICS2026
Dynamic Planar Graph Isomorphism Is in DynFO
https://doi.org/10.4230/LIPIcs.LICS.2026.36
[ "Samir Datta", "Asif Khan", "Felix Tschirbs", "Nils Vortmeier", "Thomas Zeume" ]
Consider two planar graphs which are subject to edge insertions and deletions. We show that whether the two graphs are isomorphic can be maintained with first-order logic formulas and auxiliary data of polynomial size. This places the dynamic planar graph isomorphism problem into the dynamic descriptive complexity clas...
#
LICS2026
Complexity of Satisfiability in Kochen-Specker Partial Boolean Algebras
https://doi.org/10.4230/LIPIcs.LICS.2026.37
[ "Anuj Dawar", "Nihil Shah" ]
The Kochen-Specker no-go theorem established that hidden-variable theories in quantum mechanics necessarily admit contextuality. This theorem is formally stated in terms of the partial Boolean algebra structure of projectors on a Hilbert space. Each partial Boolean algebra provides a semantics for interpreting proposit...
#
LICS2026
Well-Quasi-Ordered Classes of Bounded Clique-Width
https://doi.org/10.4230/LIPIcs.LICS.2026.38
[ "Maël Dumas", "Aliaume Lopez" ]
We study classes of graphs with bounded clique-width that are well-quasi-ordered by the induced subgraph relation, in the presence of labels on the vertices. We prove that, given a finite presentation of a class of graphs, one can decide whether the class is labelled-well-quasi-ordered. This answers positively to two c...
#
LICS2026
A Naturally-Colored Translation from LTL to Parity and COCOA
https://doi.org/10.4230/LIPIcs.LICS.2026.39
[ "Rüdiger Ehlers", "Ayrat Khalimov" ]
Chains of co-B\"uchi automata (COCOA) have recently been introduced as a new canonical representation of omega-regular languages. The co-B\"uchi automata in a chain assign each omega-word its natural color, which depends only on the language itself and not on the chosen automaton representation. Automata in such a chai...
#
LICS2026
Decidability Results for Fragments of First-Order Logic via a Symbolic Model Property
https://doi.org/10.4230/LIPIcs.LICS.2026.40
[ "Neta Elad", "Sharon Shoham" ]
Recently, symbolic structures were proposed as finite representations of potentially infinite first-order structures, where Linear Integer Arithmetic terms and formulas define the domain and interpretations of a structure. We generalize symbolic structures to use any base theory that admits a standard model. Symbolic s...
#
LICS2026
Definitional Proof Irrelevance Made Accessible
https://doi.org/10.4230/LIPIcs.LICS.2026.41
[ "Thiago Felicissimo", "Yann Leray", "Loïc Pujet", "Nicolas Tabareau", "Éric Tanter", "Théo Winterhalter" ]
A universe of propositions equipped with definitional proof irrelevance constitutes a convenient medium to express properties and proofs in type-theoretic proof assistants such as Lean, Rocq, and Agda. However, allowing accessibility predicates - used to establish semantic termination arguments - to inhabit such a univ...
#
LICS2026
Decidability of Interpretability
https://doi.org/10.4230/LIPIcs.LICS.2026.42
[ "Roman Feller", "Michael Pinsker" ]
The Bodirsky-Pinsker conjecture asserts a P vs. NP-complete dichotomy for the computational complexity of Constraint Satisfaction Problems (CSPs) of first-order reducts of finitely bounded homogeneous structures. Prominently, two structures in the scope of the conjecture have log-space equivalent CSPs if they are pp-bi...
#
LICS2026
Guarded Negation Transitive Closure Logic
https://doi.org/10.4230/LIPIcs.LICS.2026.43
[ "Diego Figueira", "Santiago Figueira", "Yoshiki Nakamura" ]
We study the guarded negation fragment of transitive closure logic (GNTC). We show that the satisfiability problem for GNTC is 2ExpTime-complete, by establishing the following reductions: (i) a polynomial-time reduction from the satisfiability problem for GNTC to the satisfiability problem for the unary negation fragme...
#
LICS2026
The Guarded Fragment with Nested Equivalences
https://doi.org/10.4230/LIPIcs.LICS.2026.44
[ "Oskar Fiuk" ]
The Guarded Fragment (GF) is a well-established decidable fragment of first-order logic. We study an extension of GF with nested equivalence relations, namely a family of distinguished binary predicates E₁, E₂, … interpreted as equivalence relations such that E_{k+1} is coarser than E_k for every k. We show that the eq...
#
LICS2026
Monads and Distributive Laws in Substructural Contexts
https://doi.org/10.4230/LIPIcs.LICS.2026.45
[ "Soichiro Fujii", "Yun Chen Tsai", "Yoàv Montacute", "Ichiro Hasuo" ]
We present a categorical theory of monads and distributive laws in substructural contexts. In the study of distributive laws, the roles of (the absence of) structural rules for variable contexts have been recognized; our theory formalizes these substructural situations using Tronin’s verbal categories W, in a uniform a...
#
LICS2026
The Logic of Bunched Implications Is Undecidable
https://doi.org/10.4230/LIPIcs.LICS.2026.46
[ "Nikolaos Galatos", "Peter Jipsen", "Søren Brinck Knudstorp", "Revantha Ramanayake" ]
The logic of bunched implications (BI), introduced by O’Hearn and Pym (1999), has attracted significant attention due to its elegant proof calculus, varied semantics, and close connections to the propositional fragment of separation logic. We show here that provability in BI is undecidable by encoding Wang tilings into...
#
LICS2026
Constructing Small Monadic Decompositions in Presburger Arithmetic
https://doi.org/10.4230/LIPIcs.LICS.2026.47
[ "Moses Ganardi", "Marin Ricros" ]
A monadic decomposition of a formula over a first-order theory is an equivalent Boolean combination of atomic formulas, each containing only one variable. Monadic decomposition is a generic simplification technique that has found applications in various settings such as quantifier elimination, string solving, and const...
#
LICS2026
Graphical Algebraic Geometry: From Ideals and Varieties to Quantum Calculi
https://doi.org/10.4230/LIPIcs.LICS.2026.48
[ "Dichuan Gao", "Razin A. Shaikh", "Aleks Kissinger" ]
We introduce Graphical Algebraic Geometry (GAG), a family of diagrammatic languages extending the Graphical Linear Algebra programme. We construct several languages within this family and prove that they are universal and complete for the corresponding (co)span semantics of commutative algebras and affine varieties. Th...
#
LICS2026
Meta-Mathematics of Algebraic Complexity
https://doi.org/10.4230/LIPIcs.LICS.2026.49
[ "Michal Garlík", "Svyatoslav Gryaznov", "Jiaqi Lu", "Rahul Santhanam", "Iddo Tzameret" ]
#
LICS2026
An Algebraic Approach to Formal System Metatheory
https://doi.org/10.4230/LIPIcs.LICS.2026.50
[ "Francesco Gavazzo" ]
We introduce an algebraic approach to the metatheory of formal systems of term assertions akin to those of structural and natural operational semantics, type theories, rewriting systems, and equational theories. We rest on Term Relation Algebras, viz. pointfree algebras of structurally-defined predicates on terms, to g...
#
LICS2026
LFPL: Revisited and Mechanized
https://doi.org/10.4230/LIPIcs.LICS.2026.51
[ "Nathaniel Glover", "Jan Hoffmann" ]
Hofmann (1999) introduced the functional programming language LFPL to characterize the functions computable in polynomial time using an affine type system. LFPL enables a natural programming style, including nested recursion, and has inspired the development of type systems for automatic cost analysis, linear dependent...
#
LICS2026
The ∞-Category of ∞-Categories in Simplicial Type Theory
https://doi.org/10.4230/LIPIcs.LICS.2026.52
[ "Daniel Gratzer", "Jonathan Weinberger", "Ulrik Buchholtz" ]
Simplicial type theory (STT) was introduced by Riehl and Shulman to leverage homotopy type theory to prove results about (∞,1)-categories. Initial work on simplicial type theory focused on "formal" arguments in higher category theory and, in particular, no non-trivial examples of ∞-category theory were constructible wi...
#
LICS2026
PVASS Reachability Is Decidable
https://doi.org/10.4230/LIPIcs.LICS.2026.53
[ "Roland Guttenberg", "Eren Keskin", "Roland Meyer" ]
Reachability in pushdown vector addition systems with states (PVASS) is among the longest standing open problems in Theoretical Computer Science. We show that the problem is decidable in full generality. Our decision procedure is similar in spirit to the KLMST algorithm for VASS reachability, but works over objects tha...
#
LICS2026
A Categorical Perspective on Constraint Satisfaction: The Wonderland of Adjunctions
https://doi.org/10.4230/LIPIcs.LICS.2026.54
[ "Maximilian Hadek", "Tomás Jakl", "Jakub Oprsal" ]
The so-called algebraic approach to the constraint satisfaction problem (CSP) has been a prevalent method of the study of complexity of these problems since early 2000's. The core of this approach is the notion of polymorphisms which determine the complexity of the problem (up to log-space reductions). In the past few ...
#
LICS2026
Classifying 2-Groups in Homotopy Type Theory
https://doi.org/10.4230/LIPIcs.LICS.2026.55
[ "Perry Hart", "Owen Milner" ]
Under the homotopy hypothesis, higher dimensional groups are defined as pointed homotopy types whose homotopy groups vanish outside a certain range. In particular, a 2-group is a pointed connected homotopy 2-type. Classically, 2-groups have two equivalent algebraic descriptions: one in terms of weak monoidal categories...
#
LICS2026
One Rig to Control Them All
https://doi.org/10.4230/LIPIcs.LICS.2026.56
[ "Chris Heunen", "Robin Kaarsgaard", "Louis Lemonnier" ]
#
LICS2026
Causality in Pure Quantum Computation with Quantum Control
https://doi.org/10.4230/LIPIcs.LICS.2026.57
[ "Kengo Hirata", "Takeshi Tsukada" ]
Indefinite causal order is a characteristic phenomenon in quantum computation, with examples including the quantum SWITCH and the OCB process. Not all such processes are believed to be physically realizable: while some implementations of the quantum SWITCH have been proposed, the OCB process is suspected to be unrealiz...
#
LICS2026
Fat Cell Structures and Generalized Algebraic Theories
https://doi.org/10.4230/LIPIcs.LICS.2026.58
[ "Xu Huang", "Carlo Angiuli" ]
#
LICS2026
Generalized Decidability via Brouwer Trees
https://doi.org/10.4230/LIPIcs.LICS.2026.59
[ "Tom de Jong", "Nicolai Kraus", "Aref Mohammadzadeh", "Fredrik Nordvall Forsberg" ]
In the setting of constructive mathematics, we suggest and study a framework for decidability of properties, which allows for finer distinctions than just "decidable, semidecidable, or undecidable". We work in homotopy type theory and use Brouwer tree ordinals to specify the level of decidability of a property. In this...
#
LICS2026
Computation and Size of Interpolants for Hybrid Modal Logics
https://doi.org/10.4230/LIPIcs.LICS.2026.60
[ "Jean Christoph Jung", "Jedrzej Kolodziejski", "Frank Wolter" ]
Recent research has established complexity results for the problem of deciding the existence of interpolants in logics lacking the Craig interpolation property (CIP). The proof techniques developed so far are non-constructive, and no meaningful bounds on the size of interpolants are known. Hybrid modal logics (or modal...
#
LICS2026