Finding Galaxies in the Shadows of Quasars with Gaussian Processes
Authors: Roman Garnett, Shirley Ho, Jeff Schneider
ICML 2015 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We demonstrate our method s effectiveness with a large-scale validation experiment on over 100 000 spectra, with excellent performance. |
| Researcher Affiliation | Academia | Roman Garnett GARNETT@WUSTL.EDU Washington University in St. Louis, St. Louis, MO 63130, United States Shirley Ho SHIRLEYH@ANDREW.CMU.EDU Jeff Schneider JEFF.SCHNEIDER@CS.CMU.EDU Carnegie Mellon University, Pittsburgh, PA 15213, United States |
| Pseudocode | No | No structured pseudocode or algorithm blocks were found in the paper. |
| Open Source Code | No | We plan to make the parameters of our learned GP model, using the full DR10 release, available for future research efforts. |
| Open Datasets | Yes | We used the QSO spectra from the BOSS DR9 Lymanforest sample (Lee et al., 2013) to train our GP model. Later, we will use the spectra from the corresponding DR10 sample (Lee et al., 2014) to evaluate our proposed DLA finding approach. |
| Dataset Splits | No | The paper mentions using DR9 for training and DR10 for evaluation, but does not specify a distinct validation set with explicit percentages or counts for model selection or hyperparameter tuning. |
| Hardware Specification | No | The paper only mentions that processing takes 'under a second on a standard desktop machine,' without providing specific hardware details like GPU/CPU models, processor types, or memory amounts. |
| Software Dependencies | No | No specific software dependencies with version numbers were listed in the paper. |
| Experiment Setup | Yes | We took k D 10, noting that the first 10 principal components of the flux matrix Y explain approximately 99.9% of the total variance. The chosen grid was this set of wavelengths with a linearly equal spacing of D 0:25 Å. The constant 0 serves to ensure that QSOs with very small redshift have sufficient data for estimating the prior; here we took 0 D 0:1. |