Martín de los Rios

Postdoctoral fellow in astronomy/physics @ IFT/UAM-Madrid.

ORCID iD iconorcid.org/0000-0003-2190-2196

About me

I am a posdoctoral fellow at the IFT (Instituto de Física Teórica) in the UAM (Universidad Autónoma de Madrid) applying machine learning techniques for the direct detection of the dark matter. I have a Phd in Astronomy from the Córdoba National University (UNC). My main interest are the sudy of standard cosmological model, the large scale structure of the universe and the dark matter and dark energy properties. I am also interest on studying different inhomogeneous models of the universe and their implications in the interpretation of the observations. To study all this I like to implement Machine Learning methods, as they are the perfect tool to analyze big datasets.

Publications

Oral Presentations

Codes

  • MeSsI: The MeSsI Algorithm is a software that performs an automatic classiffication between merging and relaxed clusters. This method was calibrated using mock catalogues constructed from the millenium simulation, and perform the classiffication using some machine learning techniques, namely random forest for classiffication and mixture of gaussians for the substructure identification.

    You can also perform an online classification using MeSsI in our website or download the R package 'MeSsI' from my github repository and use it in an interactive way.

    For more details, please read de los Rios et al. 2016 (1509.02524).

  • ROGER: This Software can be used to make a dynamical classification of galaxies taking into account their projected phase-space information. It can be downloaded from my github repository and used as an R package or using the online web interface.

    For more details please read de los Rios et al. 2020 (2010.11959) .

  • COSMIC-KITE: This python software is the results of an auto-encoding analysis of the Cosmic Microwave (CMB) power spectra. The encoder part can be used as a fast tool for estimating the maximum likelihood cosmological parameters from a given power spectrum, while the decoder part can be used as a fast emulator that computes the power spectrum corresponding to a given set of cosmological parameters. This code can be installed as a python library following from my github repository.

    For more details please read de los Rios 2022 (2202.05853) .

  • CADDENA: This is a python library for analysing dark matter direct detection experiment with Bayesian statistic using Machine Learning tools. The main goal of CADDENA is to estimate the full posterior distribution of some interesting parameter given new data. This is done using the TMNRE (Truncated Marginal Neural Ratio Estimation) method which computes the likelihood-to-evidence ratio using neural networks. This is a modular library, so if you have several observations you can combine them without the need of re-training the machine learning models! This code can be installed as a python library following the instructions from my github repository.

    For more details please read Bayesian technique to combine independently-trained Machine-Learning models applied to direct dark matter detection .