Embedded Files
My work involves spectral analysis and characterization of high redshift active galaxies and quasars. I utilize, spectral fitting, SED and photoionization modelling to understand the physics at play in peculiar sources. These sources are usually detected by application of machine learning algorithms to large spectroscopic surveys like SDSS and DESI. I am particularly interested in QSOs and high redshift primordial galaxies.
Here are the projects I have undertaken during my journey so far !
Projects
Spectroscopic quasar anomaly detection (SQuAD I) : Rest frame UV spectra from SDSS DR16
Spectroscopic quasar anomaly detection (SQuAD I) : Rest frame UV spectra from SDSS DR16
With Prof. Vivek M at the Indian Institute of Astrophysics (IIA). Explores the results of applying anomaly detection algorithm (hierarchical K-means clustering) to SDSS DR16 quasar catalog, in order identify, group & characterize 10 classes of anomalous quasars from redshift 1.88 to 2.47. A paper presenting the findings and catalog has been published in A&A (see publications).
Anomalous narrow line seyfert 1 galaxies from SDSS DR17 (SQuAD II)
Anomalous narrow line seyfert 1 galaxies from SDSS DR17 (SQuAD II)
With Rachana (IISc), Prof. Vivek M (IIA) and Dr. Suvendu Rakshit (ARIES). Presents the results of multi-wavelength analysis of three classes of peculiar NLSy1 galaxies, red, blue and emission line dominated. BPT analysis, WISE color analysis, orientation analysis using [OIII} EW, redsfhit versus luminosity distribution and other multi-approach diagnosis were performed to characterize the three classes. Paper submitted to ApJ along with catalog. (see publications).
Parallel galaxy simulations and analysis: SKIRT and GASOLINE
Parallel galaxy simulations and analysis: SKIRT and GASOLINE
With Prof. James Wadsley at McMaster University, Hamilton, ON, Canada. The project involved utilization of SKIRT dust radiative transfer code to mimic the SED of three NGC galaxies simulated using GASOLINE code and match the generated data to archival JWST and HST data. A poster was presented summarizing the results (publications).
Projectinos
Using deep learnig to predict the bulge to total light ratio of galaxies
Using deep learnig to predict the bulge to total light ratio of galaxies
With Dr. Ana Paulino Afonso, Institute of Astrophysics and Space Science, Portugal. Developed CNN based algorithms to predict Bulge to Total light (B/T) ratio of high redshift (z>5) galaxies using nearly 180 thousand galaxies from SDSS DR17 for training. Solved the problem of CNN (or other image based) models failing to predict the type of galaxy with increasing
redshifts due to significant distortions in the shape and light of the galaxies. Studied the ability of B/T ratio as an indicator of the galactic morphological type.
A group project for the fulfilment of course requirement for DSE 313 at the Indian Institute of Science Education and Research, Bhopal under Dr. Vaibhav Kumar during Fall 2021 semester. Developed CNN models to classify SDSS galaxy images achieving an accuracy of 82.7% for a three-fold classification problem.
A group project for the fulfilment of course requirement for DSE 302 at the Indian Institute of Science Education and Research, Bhopal under Dr. Tanmay Basu during Fall 2022 semester. Used Kepler telescope data to engineer features that were used to predict the semi-major axis of exoplanet orbitals attaining an accuracy of 92.8%.
This project was for the fulfilment of course requirement for DSE 309 at the Indian Institute of Science Education and Research, Bhopal under Dr. Parthiban Srinivasan during Fall 2021 semester. We developed an ontology in Owlready2 for stellar classification. The Harvard classification scheme was used for spectral differentiation using temperature, colors and emission lines.
Page updated
Google Sites
Report abuse