lblogan14 / iqnet-v1

Trained iQNet-V1 models with HST and HST+SDSS training spectra

iQNet Model

This repository is reserved for the paper A Deep Learning Approach to Quasar Continuum Prediction. The tutorial of model usage is written in the IPython Notebook. Please cite the paper https://arxiv.org/abs/2006.04814 if you use any data, model, or code in this repository.

We're currently working on the 2nd version of iQNet implemented with new upgrades so this repository is named as the vallina 1st version of iQNet. To better adapt to other test spectra, users should retrain the model with their own dataset to reduce prediction error in the predicted intrinsic continua.

We provided two iQNet models, one trained with HST Cosmic Origin Spectrograph QSO spectra and the other one trained with both HST and SDSS QSO spectra. The discrepany between those predicted continua are merely the difference in the training spectra.

The figure below shows some of the test results from iQNet-V1.