Online Optimization with Memory and Competitive Control

Authors: Guanya Shi, Yiheng Lin, Soon-Jo Chung, Yisong Yue, Adam Wierman

NeurIPS 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental In this section we use simple numerical examples to illustrate the contrast between the best linear controller in hindsight and the optimal offline controller. We also test our algorithm, Optimistic ROBO, and then numerically illustrate that Optimistic ROBD can obtain near-optimal cost and outperform the offline optimal linear controller.
Researcher Affiliation Collaboration Guanya Shi* CMS, Caltech gshi@caltech.edu Yiheng Lin* IIIS, Tsinghua University linyh16@mails.tsinghua.edu.cn Soon-Jo Chung CMS, GALCIT, JPL, Caltech sjchung@caltech.edu Yisong Yue CMS, Caltech yyue@caltech.edu Adam Wierman CMS, Caltech adamw@caltech.edu
Pseudocode Yes Algorithm 1: Regularized OBD (ROBD), Goel et al. [24]... Algorithm 2: Optimistic ROBD...
Open Source Code No The paper does not include an unambiguous statement about releasing source code for the methodology described, nor does it provide a direct link to a code repository.
Open Datasets No The paper describes generating synthetic data for its numerical examples (e.g., 'wt U( 1, 1) i.i.d.', 'wt+1 = wt + ψt where ψt U( 0.2, 0.2) i.i.d.') but does not refer to or provide access information for any publicly available or open dataset.
Dataset Splits No The paper uses numerical examples with generated data but does not specify exact split percentages, sample counts, or refer to predefined splits for training, validation, or testing.
Hardware Specification No The paper does not provide any specific hardware details such as GPU/CPU models, processor types, or memory amounts used for running its experiments.
Software Dependencies No The paper does not list specific software dependencies with version numbers, such as programming languages or libraries.
Experiment Setup No The paper describes the setup of its numerical examples (e.g., 1-d and 2-d system dynamics, objective functions) but does not provide specific hyperparameter values, training configurations, or detailed system-level settings for the algorithms.