Constrained Optimization via Exact Augmented Lagrangian and Randomized Iterative Sketching
Authors: Ilgee Hong, Sen Na, Michael W. Mahoney, Mladen Kolar
ICML 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We implement Ada Sketch-Newton and benchmark it against constrained nonlinear problems in CUTEst test set, constrained logistic regression with datasets from LIBSVM, and a PDE-constrained problem. |
| Researcher Affiliation | Academia | 1Department of Statistics, University of Chicago 2Department of Statistics, University of California, Berkeley 3International Computer Science Institute 4Lawrence Berkeley National Laboratory 5Booth School of Business, University of Chicago. |
| Pseudocode | Yes | Algorithm 1 Ada Sketch-Newton Method 1: Input: initial iterate z0; sequence {θk} (0, 1]; scalars η1,0, η2,0, ξB > 0, δ0 (0, 1), β (0, 0.5), ν > 1;... |
| Open Source Code | Yes | Our code for the implementation is available at https://github.com/Ilgee Hong/Ada Sketch-Newton. |
| Open Datasets | Yes | We benchmark Ada Sketch-Newton (Algorithm 1) on nonlinear problems in CUTEst collection set (Gould et al., 2014), on constrained logistic regression with data from LIBSVM (Chang & Lin, 2011), and on a PDE-constrained problem (Hinterm uller et al., 2002). |
| Dataset Splits | No | The paper uses various datasets for benchmarking (CUTEst, LIBSVM, PDE-constrained problem) but does not explicitly specify how these datasets are split into training, validation, and test sets. It mentions using the 'CUTEst test set' and providing 'details of the datasets' in Table 1 for LIBSVM (which only lists dimension and number of data points), without detailing specific splits or cross-validation setups. |
| Hardware Specification | No | The paper describes the experiments conducted and the methods used but does not provide any specific details about the hardware used, such as CPU or GPU models, memory, or cloud computing resources. |
| Software Dependencies | No | The paper mentions using 'GMRES' as a deterministic solver but does not specify any version numbers for GMRES or any other software dependencies. |
| Experiment Setup | Yes | The parameters of each algorithm are specified as follows. (We test the sensitivity to parameters for Algorithm 1 in Section 5.4). Alg. 1: η2,0 = δ0 = ξB = β = 0.1, η1,0 = θk = 1, ν = 1.5. |