Accelerated Mirror Descent in Continuous and Discrete Time

Authors: Walid Krichene, Alexandre Bayen, Peter L. Bartlett

NeurIPS 2015 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We test these methods on numerical examples in Section 5 and comment on their performance. The results are given in Figure 1.
Researcher Affiliation Academia Walid Krichene UC Berkeley walid@eecs.berkeley.edu Alexandre M. Bayen UC Berkeley bayen@berkeley.edu Peter L. Bartlett UC Berkeley and QUT bartlett@berkeley.edu
Pseudocode Yes Algorithm 1 Accelerated mirror descent with distance generating function ψ , regularizer R, step size s, and parameter r 3 and Algorithm 2 Accelerated mirror descent with restart
Open Source Code No The paper does not provide any explicit statements about releasing source code, nor does it include links to a code repository.
Open Datasets No The paper describes generating data for two different objective functions: a simple quadratic f(x) = x x , Q(x x ) , for a random positive semi-definite matrix Q, and a log-sum-exp function... where each entry in ai Rn and bi R is iid normal. No publicly available dataset is mentioned or linked.
Dataset Splits No The paper does not discuss dataset splits (training, validation, or test) as it focuses on optimizing generated objective functions rather than using pre-existing datasets with fixed splits.
Hardware Specification No The paper does not provide any specific details about the hardware (e.g., CPU, GPU models, memory) used for running the experiments.
Software Dependencies No The paper does not provide specific software dependencies with version numbers (e.g., Python, PyTorch, TensorFlow versions or specific solver versions).
Experiment Setup Yes We test the accelerated mirror descent method in Algorithm 1, on simplex-constrained problems in Rn, n = 100... (c) Effect of the parameter r. r = 3 r = 10 r = 30 r = 90. The restart conditions are the following: (i) gradient restart: x(k+1) x(k), f(x(k)) > 0, and (ii) speed restart: x(k+1) x(k) < x(k) x(k 1) .