Pablo Villanueva Domingo
PhD in Physics | Cosmologist | Deep Learning Scientist at Computer Vision Center - Universitat Autònoma de Barcelona
PhD in Physics | Cosmologist | Deep Learning Scientist at Computer Vision Center - Universitat Autònoma de Barcelona
Graph Neural Networks to infer cosmological parameters and power spectrum on galaxy catalogues, making use of PyTorch Geometric. More details in the paper 2204.13713.
Graph Neural Networks to predict the mass of a dark matter halo from its hosted galaxies in hydrodynamic simulations, using of PyTorch Geometric. More details in the papers 2111.08683, 2111.14874.
Convolutional Neural Network to extract the underlying matter density field from a 21 cm map, employing the U-Net architecture in PyTorch. More details in the paper 2006.14305, ApJ 907 44, 2021.
Denoising Diffusion Probabilistic Model (DDPM) in Pytorch trained in the CAMELS simulations to generate cosmological density maps.
Integration of the data-driven realistic traffic generation InvertedAI API in the autonomous driving CARLA simulator.
Deep learning model to account for vehicle-terrain interaction and replace terramechanics computations in Chrono simulator.
Pytorch model to learn semantic segmentation of driving images generated with the CARLA simulator.
Generate wide lens fisheye camera images from cubemap renders.
Virtual assistant with a retrieval-augmented generation (RAG) system to query information and ask questions about the CAMELS simulations, using Langchain with a MistralAI LLM and deployed with Gradio.
A MistralAI powered chatbot which talks in the way of ancient times, using Langchain and deployed with Gradio.
Train a recurrent neural network (LSTM) with PyTorch to generate poems imitating the style of an author.
Train reinforcement learning algorithms with the Mujoco’s Humanoid environment using StableBaselines3.
Implementation of a DCGAN architecture in Tensorflow for generating pictures.
Compute the center of mass of a particle system using several models of Graph Neural Networks with PyTorch Geometric.
Exploratory data analysis and classification of stars from the Hipparcos catalog using pandas, seaborn and sklearn.
Classify glitches in gravitational waves events detected by LIGO using machine learning models with sklearn.
Python library to compute 21 cm forest observables, including subhalo contributions and non-standard cosmologies. More details in the paper 2209.01305. Read the documentation here and install with pip install hayashi.
Code to study bounds on the primordial black holes abundance with neutrino fluxes from emitted Hawking evaporation. More details in the paper 2203.14979.
Python script to plot the current observational upper limits on the cosmological 21 cm power spectrum.
Python code to predict the number of absorption features in the 21 cm forest within primordial black holes scenarios. More details in the paper 2104.10695.
Python code to predict the thermal evolution of the universe and the 21 cm global signal, taking into account the overdensities. More details in the paper 1912.09488, Phys. Rev. D 101, 083502.
Compute the Fisher matrix from simulations of the 21 cm power spectrum in warm dark matter cosmologies.
Fortran code to identify Baryonic Acoustic Oscillations in the LOWZ SDSS galaxy survey via wavelets. Data extracted with a SQL query.
Run Monte Carlo Markov Chains using the Metropolis-Hastings algorithm with Supernovae Ia data to perform cosmological parameter inference.
Visualize the relations between different musical bands and musicians within a musical scene, using Networkx and ipysigma. Source code here.
A summary of my own activity biking, hiking and running registered in Strava, scrapped through the Strava API and visualized with Folium and Plotly. Source code here.
Website for a machine learning company project, rendered using Jekyll.
My own webpage you are seeing, rendered using Jekyll.
Simple first-person shooter developed in Godot.
Videogame based on the classic Metroid programmed in Python and Pygame.
Python script to generate a height map from a given power spectrum.
Implementation of cellular automata as simple Convolutional Neural Networks using PyTorch.
Generate procedural geographical maps from gaussian fields, Perlin noise or other random fields. Try an interactive demo here.
Scrape publication data from the INSPIRE-HEP database using its API to create a graph of collaborations employing Networkx.
A Processing script to generate convolutional neural network diagrams.
Simple N-body simulator of a pandemic spread in Python, following the SEIR model.
Generate pictures of fractal universes for different models of hierarchical clustering of galaxies.
Script for solving the wave equation numerically in Fortran.
Implementation of Conway’s Game of Life using Python and Pygame.
Elementary Wolfram’s one-dimensional cellular automata following binary rules.