Flat Holonomies on Automata Networks
Gene Itkis, Leonid A. Levin
Abstract
We consider asynchronous networks of identical finite (independent of network's size or topology) automata. Our automata drive any network from any initial configuration of states, to a coherent one in which it can carry efficiently any computations implementable on synchronous properly initialized networks of the same size. A useful data structure on such networks is a partial orientation of its edges. It needs to be flat, i.e. have null holonomy (no excess of up or down edges in any cycle). It also needs to be centered, i.e. have a unique node with no down edges. There are (interdependent) self-stabilizing asynchronous finite automata protocols assuring flat centered orientation. Such protocols may vary in assorted efficiency parameters and it is desirable to have each replaceable with any alternative, responsible for a simple limited task. We describe an efficient reduction of any computational task to any such set of protocols compliant with our interface conditions.
Create a lesson
Related papers
Consensus with Stochastic Broadcast
Pierre Fraigniaud, Boaz Patt-Shamir, Sergio Rajsbaum
Decoupled I/O-Dominant Pipelines for Large-Scale Whole-Slide Image Embedding Extraction
Mayanka Chandrashekar, Xi Zhang, Ethan Seefried et al.
Sintr: Safe Interactive Transactions in the Presence of Byzantine Clients
Austin T. Li, Daniel H. Lee, Lorenzo Alvisi et al.
Benchmarking Confidential Computing Performance on NVIDIA Blackwell GPUs
Daniyal Khan, Amean Asad, Ansgar Grunseid
Optimizing API Gateway Placement in Multi-Cloud Kubernetes
Vinoth Punniyamoorthy, Murali Shankar Dulam, Aswathnarayan Muthukrishnan Kirubakaran et al.
VPP: Virtual Pipeline Parallelism for Efficient Chunked Prefill in Long-Context LLM Inference
Yan Shi, Xiaochao Wang, Jingchun Gao et al.