A primal-dual method for conic constrained distributed optimization problems

Authors: Necdet Serhat Aybat, Erfan Yazdandoost Hamedani

NeurIPS 2016 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We tested DPDA-S and DPDA-D on a primal linear SVM problem where the data is distributed among the computing nodes in N. Relative suboptimality and relative consensus violation, i.e., max(i,j) E [w i bi] [w j bj] / [w b ] , and absolute feasibility violation are plotted against iteration counter in Fig. 3, where [w b ] denotes the optimal solution to the central problem.
Researcher Affiliation Academia Necdet Serhat Aybat Department of Industrial Engineering Penn State University University Park, PA 16802 nsa10@psu.edu Erfan Yazdandoost Hamedani Department of Industrial Engineering Penn State University University Park, PA 16802 evy5047@psu.edu
Pseudocode Yes Figure 1: Distributed Primal Dual Algorithm for Static G (DPDA-S) and Figure 2: Distributed Primal Dual Algorithm for Dynamic Gt (DPDA-D)
Open Source Code No The paper mentions implementing the proposed algorithms and presents numerical results, but it does not provide any specific links or explicit statements about the public availability of its source code.
Open Datasets No The dataset used is synthetically generated: '{xℓ}ℓ S is generated from two-dimensional multivariate Gaussian distribution with covariance matrix Σ = [1, 0; 0, 2] and with mean vector either m1 = [ 1, 1]T or m2 = [1, 1]T with equal probability.' No link or citation to a publicly available dataset is provided.
Dataset Splits No The paper specifies the partition of data into 'Stest' and 'Strain' with sizes '|Stest| = 600' and '|Strain| = 300'. However, it does not explicitly mention or provide details for a separate validation dataset split.
Hardware Specification No The paper describes experimental setups but does not provide specific details about the hardware (e.g., GPU/CPU models, memory) used for running the experiments.
Software Dependencies No The paper does not provide specific software dependency details, such as library names with version numbers, required to replicate the experiments.
Experiment Setup Yes The experiment was performed for C = 2, |N| = 10, s = 900 such that |Stest| = 600, |Si| = 30 for i N, i.e., |Strain| = 300, and qk = k.