Stochastic Discrete Clenshaw-Curtis Quadrature
Authors: Nico Piatkowski, Katharina Morik
ICML 2016 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Experiments confirm that the new randomized algorithm is highly accurate if the parameter norm is small, and is otherwise comparable to methods with unbounded error. 5. Numerical Evaluation For our experiments, we implemented DCCQ (Alg. 1) and SDCCQ (Alg. 2) and execute them on a machine with 40 E5-2697 Xeon CPU cores. |
| Researcher Affiliation | Academia | Nico Piatkowski NICO.PIATKOWSKI@TU-DORTMUND.DE Artificial Intelligence Group, TU Dortmund, Germany Katharina Morik KATHARINA.MORIK@TU-DORTMUND.DE Artificial Intelligence Group, TU Dortmund, Germany |
| Pseudocode | Yes | Algorithm 1 DCCQ Input: θ Rd, k N, φ Output: Approximate partition function ˆZk(θ) and Algorithm 2 SDCCQ Input: θ Rd, k N, m Nk, φ Output: Approximate partition function Zm k (θ) |
| Open Source Code | Yes | Our C++ source code and the precomputed Qφ(i) values are available at http://sfb876.tu-dortmund.de/sdccq. |
| Open Datasets | No | The paper describes how parameters and data were generated for experiments (e.g., 'random Gaussian parameters', 'Ising grid models with randomly generated parameter vectors'), but does not provide concrete access information (link, DOI, formal citation) for a publicly available or open dataset. |
| Dataset Splits | No | The paper mentions 'averaged over 5-folds of SDCCQ' but does not provide specific details on how data was split into training, validation, or test sets (e.g., percentages, sample counts, or references to predefined splits). |
| Hardware Specification | Yes | For our experiments, we implemented DCCQ (Alg. 1) and SDCCQ (Alg. 2) and execute them on a machine with 40 E5-2697 Xeon CPU cores. |
| Software Dependencies | No | The paper mentions 'Our C++ source code' but does not provide specific ancillary software or library names with version numbers (e.g., 'PyTorch 1.9', 'TensorFlow 2.x'). |
| Experiment Setup | Yes | SDCCQ has been run with mi = 103, i and k {1, 2, 4, 8}, where, due to space restrictions, the plot shows the average over all SDCCQ runs. Specifically, we have n = 10 10 binary variables x { 1, 1}n with weights θvu=xy = wvu whenever x = y and θvu=xy = wvu otherwise. In the attractive setting, the wvu are drawn from [0, ω]; in the mixed setting, from [ ω, ω]. Moreover, vertex weights θv=1 = θv= 1 = wv are sampled from [ κ, κ] with κ {0.1, 1.0}. |