arXiv subjects

Christian Fidler

Publications and source records attributed to Christian Fidler.

At least 19 recordsLinked to original sources

arXiv2026

Effective Field Theory of Large Scale Structure and Newtonian Motion Gauges

The simplest flavor of the Effective Field Theory of Large Scale Structure is based on Newtonian equations and describes the nonlinear matter density and velocity using Einstein-de-Sitter kernels. Even in the presence of massive neutrinos, this has been argued to be sufficient for the analysis of data from Stage-III galaxy surveys. In this paper, we show that there exists a simple way to extend the validity range of this framework to more complex problems with a scale-dependent growth factor, while incorporating linear general relativistic (GR) corrections as well. For a given cosmology, an Einstein-Boltzmann code can find the exact gauge transformation that brings the full linear equations of motion of the clustering matter components into a form where they are identical to Newtonian equations for a self-gravitating fluid with scale-independent growth. Non-linear clustering can be consistently computed in this gauge, and the results can be transformed back to the initial gauge in order to incorporate GR and scale-dependent-growth effects. Redshift-space distortions can also be accounted for with a similar strategy. Our method does not incur any additional computational cost. As a showcase, we apply this method to cosmologies with massive neutrinos. For the real-space one-loop power spectrum, we find that the largest deviation between the accurate and standard methods remains below 0.7% for M_nu<0.30 eV. However, in redshift space, it reaches 1.7% for the one-loop quadrupole spectrum at k=0.3 h/Mpc and z=0, with the largest contribution coming from the effect of the cosmological constant on the growth of the velocity field. Our method could be applied to a much wider range of models with more significant scale-dependent growth, as long as a self-consistency condition evaluated by the Einstein-Boltzmann code (on the smallness of a gauge transformation field) is fulfilled.

astro-ph.CO↗

arXiv2025

CLAPP: The CLASS LLM Agent for Pair Programming

We introduce CLAPP (CLASS LLM Agent for Pair Programming), an interactive AI assistant designed to support researchers working with the Einstein-Boltzmann solver CLASS. CLAPP leverages large language models (LLMs) and domain-specific retrieval to provide conversational coding support for CLASS-answering questions, generating code, debugging errors, and producing plots. Its architecture combines multi-agent LLM orchestration, semantic search across CLASS documentation, and a live Python execution environment. Deployed as a user-friendly web application, CLAPP lowers the entry barrier for scientists unfamiliar with AI tools and enables more productive human-AI collaboration in computational and numerical cosmology. The app is available at https://classclapp.streamlit.app

astro-ph.IM↗

arXiv2025

Open Source Planning & Control System with Language Agents for Autonomous Scientific Discovery

We present a multi-agent system for automation of scientific research tasks, cmbagent (https://github.com/CMBAgents/cmbagent). The system is formed by about 30 Large Language Model (LLM) agents and implements a Planning & Control strategy to orchestrate the agentic workflow, with no human-in-the-loop at any point. Each agent specializes in a different task (performing retrieval on scientific papers and codebases, writing code, interpreting results, critiquing the output of other agents) and the system is able to execute code locally. We successfully apply cmbagent to carry out a PhD level cosmology task (the measurement of cosmological parameters using supernova data) and evaluate its performance on two benchmark sets, finding superior performance over state-of-the-art LLMs. The source code is available on GitHub, demonstration videos are also available, and the system is deployed on HuggingFace and will be available on the cloud.

cs.AI↗

arXiv2025

OL\'E -- Online Learning Emulation in Cosmology

In this work, we present OL\'E, a new online learning emulator for use in cosmological inference. The emulator relies on Gaussian Processes and Principal Component Analysis for efficient data compression and fast evaluation. Moreover, OL\'E features an automatic error estimation for optimal active sampling and online learning. All training data is computed on-the-fly, making the emulator applicable to any cosmological model or dataset. We illustrate the emulator's performance on an array of cosmological models and data sets, showing significant improvements in efficiency over similar emulators without degrading accuracy compared to standard theory codes. We find that OL\'E is able to considerably speed up the inference process, increasing the efficiency by a factor of $30-350$, including data acquisition and training. Typically the runtime of the likelihood code becomes the computational bottleneck. Furthermore, OL\'E emulators are differentiable; we demonstrate that, together with the differentiable likelihoods available in the $\texttt{candl}$ library, we can construct a gradient-based sampling method which yields an additional improvement factor of 4. OL\'E can be easily interfaced with the popular samplers $\texttt{MontePython}$ and $\texttt{Cobaya}$, and the Einstein-Boltzmann solvers $\texttt{CLASS}$ and $\texttt{CAMB}$. OL\'E is publicly available at https://github.com/svenguenther/OLE .

astro-ph.CO↗

arXiv2024

TOPO: Time-Ordered Provable Outputs

We present TOPO (Time-Ordered Provable Outputs), a tool designed to enhance reproducibility and data integrity in astrophysical research, providing a trustless alternative to data analysis blinding. Astrophysical research frequently involves probabilistic algorithms, high computational demands, and stringent data privacy requirements, making it difficult to guarantee the integrity of results. TOPO provides a secure framework for verifying reproducible data analysis while ensuring sensitive information remains hidden. Our approach utilizes deterministic hashing to generate unique digital fingerprints of outputs, and Merkle Trees to store outputs in a time-ordered manner. This enables efficient verification of specific components or the entire dataset while making it computationally infeasible to manipulate results - thereby mitigating the risk of human interference and confirmation bias, key objectives of blinding methods. We demonstrate TOPO's utility in a cosmological context using TOPO-Cobaya, showcasing how blinding and verification can be seamlessly integrated into Markov Chain Monte Carlo calculations, rendering the process cryptographically provable. This method addresses pressing challenges in the reproducibility crisis and offers a robust, verifiable framework for astrophysical data analysis.

astro-ph.IM↗

arXiv2023

GW_CLASS: Cosmological Gravitational Wave Background in the Cosmic Linear Anisotropy Solving System

The anisotropies of the Cosmological Gravitational Wave Background (CGWB) retain information about the primordial mechanisms that source the gravitational waves and about the geometry and the particle content of the universe at early times. In this work, we discuss in detail the computation of the angular power spectra of CGWB anisotropies and of their cross correlation with Cosmic Microwave Background (CMB) anisotropies, assuming different processes for the generation of these primordial signals. We present an efficient implementation of our results in a modified version of CLASS which will be publicly available. By combining our new code GW_CLASS with MontePython, we forecast the combined sensitivity of future gravitational wave interferometers and CMB experiments to the cosmological parameters that characterize the cosmological gravitational wave background.

gr-qc↗

arXiv2022

Fast and robust Bayesian Inference using Gaussian Processes with GPry

We present the GPry algorithm for fast Bayesian inference of general (non-Gaussian) posteriors with a moderate number of parameters. GPry does not need any pre-training, special hardware such as GPUs, and is intended as a drop-in replacement for traditional Monte Carlo methods for Bayesian inference. Our algorithm is based on generating a Gaussian Process surrogate model of the log-posterior, aided by a Support Vector Machine classifier that excludes extreme or non-finite values. An active learning scheme allows us to reduce the number of required posterior evaluations by two orders of magnitude compared to traditional Monte Carlo inference. Our algorithm allows for parallel evaluations of the posterior at optimal locations, further reducing wall-clock times. We significantly improve performance using properties of the posterior in our active learning scheme and for the definition of the GP prior. In particular we account for the expected dynamical range of the posterior in different dimensionalities. We test our model against a number of synthetic and cosmological examples. GPry outperforms traditional Monte Carlo methods when the evaluation time of the likelihood (or the calculation of theoretical observables) is of the order of seconds; for evaluation times of over a minute it can perform inference in days that would take months using traditional methods. GPry is distributed as an open source Python package (pip install gpry) and can also be found at https://github.com/jonaselgammal/GPry.

astro-ph.CO↗

arXiv2022

CosmicNet II: Emulating extended cosmologies with efficient and accurate neural networks

In modern analysis pipelines, Einstein-Boltzmann Solvers (EBSs) are an invaluable tool for obtaining CMB and matter power spectra. To accelerate the computation of these observables, the CosmicNet strategy is to replace the bottleneck of an EBS, which is the integration of a system of differential equations for linear cosmological perturbations, by neural networks. This strategy offers advantages compared to the direct emulation of the final observables, including small networks that are easy to train in high-dimensional parameter spaces, and which do not depend by on primordial spectrum parameters nor observation-related quantities such as selection functions. In this second CosmicNet paper, we present a more efficient set of networks that are already trained for extended cosmologies beyond LCDM, with massive neutrinos, extra relativistic degrees of freedom, spatial curvature, and dynamical dark energy. We release a new branch of the CLASS code, called CLASSNET, which automatically uses networks within a region of trusted accuracy. We demonstrate the accuracy and performance of CLASSNET by presenting parameter inference runs from Planck, BAO and supernovae data, performed with CLASSNET and the COBAYA inference package. We have eliminated the perturbation module as a bottleneck of the EBS, with a speedup that is even more remarkable in extended cosmologies, where the usual approach would have been more expensive while the network's performance remains the same. We obtain a speedup factor of order 150 for the emulated perturbation module of CLASS. For the whole code, this translates into an overall speedup factor of order 3 when computing CMB harmonic spectra (now dominated by the highly parallelizable and further optimizable line-of-sight integration), and of order 50 when computing matter power spectra (less than 0.1 seconds even in extended cosmologies).

astro-ph.CO↗

arXiv2022

A Minimal Model for Massive Neutrinos in Newtonian N-body Simulations

We present a novel method for including the impact of massive neutrinos in cold dark matter N-body simulations. Our approach is compatible with widely employed Newtonian N-body codes and relies on only three simple modifications. First, we use commonly employed backscaling initial conditions, based on the cold dark matter plus baryon power spectrum instead of the total matter power spectrum. Second, the accurate Hubble rate is employed in both the backscaling and the evolution of particles in the N-body code. Finally, we shift the final particle positions in a post-processing step to account for the integrated effect of neutrinos on the particles in the simulation. However, we show that the first two modifications already capture most of the relevant neutrino physics for a large range of observationally interesting redshifts and scales. The output of the simulations are the cold dark matter and baryon distributions and can be analysed using standard methods. All modifications are simple to implement and do not generate any computational overhead. By implementing our methods in the N-body codes GADGET and gevolution, we show that any state-of-the-art Newtonian N-body code can be utilised out of the box. Our method is also compatible with higher order Lagrangian perturbation theory initial conditions and accurate for combined neutrino masses of up to at least 0.3 eV. Being formulated in relativistic gauge theory, in addition to including the impact of massive neutrinos, our method further includes relativistic corrections relevant on the large scales for free.

astro-ph.CO↗

arXiv2020

Fast simulations of cosmic large-scale structure with massive neutrinos

Accurate cosmological simulations that include the effect of non-linear matter clustering as well as of massive neutrinos are essential for measuring the neutrino mass scale from upcoming galaxy surveys. Typically, Newtonian simulations are employed and the neutrino distribution is sampled with a large number of particles in order to beat down the shot noise. Here we perform very efficient simulations with light massive neutrinos which require virtually no extra cost over matter-only simulations, and furthermore do not require tracer particles for the neutrinos. Instead, we use a weak-field dictionary based on the recently developed Newtonion motion approach, where Newtonian simulations for matter are paired with a linear relativistic Boltzmann code to allow for an absorption of the neutrino evolution into a time-dependent coordinate transformation. For this, only minimal modifications in existing N-body codes are required, which we have explicitly implemented in $\textit{gevolution}$ and $\textit{gadget-2}$. Our fast method determines the non-linear matter power spectrum to permille-level precision when compared against state-of-the-art simulations that have been performed for $0.1\,{\rm eV} \leq \sum m_ν\leq 0.3\,$eV.

astro-ph.CO↗

arXiv2019

Lensing anomalies from the epoch of reionisation

Reionisation blurring is a non-linear correction to the cosmic microwave background that acts similar to weak gravitational lensing and that can be computed from linear perturbations through a blurring potential. Its impact on the cosmic microwave background is roughly two order of magnitude smaller than that of lensing, in isolation. But the blurring potential is strongly correlated with the lensing potential thereby generating a potentially observable cross-correlation. We compute for the first time the inclusive impact of reionisation blurring on the temperature angular power spectrum and discuss how much it could induce lensing anomalies

astro-ph.CO↗

arXiv2019

The large-scale general-relativistic correction for Newtonian mocks

We clarify the subtle issue of finding the correct mapping of Newtonian simulations to light-cone observables at very large distance scales. A faithful general-relativistic interpretation specifies a gauge, i.e. a chart that relates the simulation data to points of the space-time manifold. It has already been pointed out that the implicit gauge choice of Newtonian simulations is indeed different from the Poisson gauge that is commonly adopted for relativistic calculations, the difference being most significant at large scales. It is therefore inconsistent, for example, to predict weak-lensing observables from simulations unless this gauge issue is properly accounted for. Using perturbation theory as well as fully relativistic N-body simulations we quantify the systematic error introduced this way, and we discuss several solutions that would render the calculations relativistically self-consistent.

astro-ph.CO↗

arXiv2019

CosmicNet I: Physics-driven implementation of neural networks within Boltzmann-Einstein solvers

Einstein-Boltzmann Solvers (EBSs) are run on a massive scale by the cosmology community when fitting cosmological models to data. We present a new concept for speeding up such codes with neural networks. The originality of our approach stems from not substituting the whole EBS by a machine learning algorithm, but only its most problematic and least parallelizable step: the integration of perturbation equations over time. This approach offers two significant advantages: the task depends only on a subset of cosmological parameters, and it is blind to the characteristics of the experiment for which the output must be computed (for instance, redshift bins). These allow us to construct a fast and highly re-usable network. In this proof-of-concept paper, we focus on the prediction of CMB source functions, and design our networks according to physical considerations and analytical approximations. This allows us to reduce the depth and training time of the networks compared to a brute-force approach. Indeed, the calculation of the source functions using the networks is fast enough so that it is not a bottleneck in the EBS anymore. Finally, we show that their accuracy is more than sufficient for accurate MCMC parameter inference from Planck data. This paves the way for a new project, CosmicNet, aimed at gradually extending the use and the range of validity of neural networks within EBSs, and saving massive computational time in the context of cosmological parameter extraction.

astro-ph.CO↗

arXiv2019

A Relativistic Interpretation of Bias in Newtonian Simulations

Observables of cosmic structures are usually not the underlying matter field but biased tracers of matter, such as galaxies or halos. We show how the bias found in Newtonian N-body simulations can be interpreted in terms of the weak-field limit of General Relativity (GR). For this we employ standard Newtonian simulations of cold dark matter and incorporate GR/radiation via a weak-field dictionary that we have recently developed. We find that even when a simple local biasing scheme is employed in the Newtonian simulation, the relativistic bias becomes inherently scale-dependent due to the presence of radiation and GR corrections. This scale-dependence could be in principle observed on large scales in upcoming surveys. As a working example, we apply our methodology to Newtonian simulations for the spherical collapse and recover permille-level agreement between the approaches for extracting the relativistic bias on all considered scales.

astro-ph.CO↗

arXiv2019

A new approach to cosmological structure formation with massive neutrinos

We show how Newtonian cosmological simulations can be employed to investigate the non-linear evolution of two particle species in a relativistic context. We discuss the application for massive neutrinos and other multi-species systems such as Cold Dark Matter (CDM) plus baryons or Warm Dark Matter (WDM). We propose a method that allows us to perform simulations including massive neutrinos and general relativistic effects at almost the same computational cost as ordinary CDM only N-body simulations, employing tailor-made initial conditions and a dictionary for the interpretation of the simulation output.

astro-ph.CO↗

arXiv2018

Relativistic bias in neutrino cosmologies

Halos and galaxies are tracers of the underlying dark matter structures. While their bias is well understood in the case of a simple Universe composed dominantly of dark matter, the relation becomes more complex in the presence of massive neutrinos. Indeed massive neutrinos introduce rich dynamics in the process of structure formation leading to scale-dependent bias. We study this process from the perspective of general relativity employing a simple spherical collapse model. We find a characteristic signature at the neutrino free-streaming scale in addition to a large-scale feature from general relativity. The scale-dependent halo bias opposes the suppression in the matter distribution due to neutrino free-streaming and leads to corrections of a few percent in the halo power spectrum. It is not only sensitive to the sum of the neutrino-masses, but respond to the individual masses. Accurate models for the neutrino bias are a crucial ingredient for the future data analysis and play an important role in constraining the neutrino masses.

astro-ph.CO↗

arXiv2018

Suitable Initial Conditions for Newtonian Simulations with Massive Neutrinos

Initial conditions for cosmological N-body simulations are usually calculated by rescaling the present day linear power spectrum obtained from an Einstein-Boltzmann solver to the initial time employing the scale-independent matter growth function. For the baseline Lambda-CDM model, this has been shown to be consistent with General Relativity (GR) even in the presence of relativistic species such as photons. We show that this approach is not feasible in cosmologies with massive neutrinos and present an alternative method employing the Newtonian motion gauge framework.

astro-ph.CO↗

arXiv2018

CMB anisotropies from patchy reionisation and diffuse Sunyaev-Zel'dovich effects

Anisotropies in the Cosmic Microwave Background (CMB) can be induced during the later stages of cosmic evolution, and in particular during and after the Epoch of Reionisation. Inhomogeneities in the ionised fraction, but also in the baryon density, in the velocity fields and in the gravitational potentials are expected to generate correlated CMB perturbations. We present a complete relativistic treatment of all these effects, up to second order in perturbation theory, that we solve using the numerical Boltzmann code SONG. The physical origin and relevance of all second order terms are carefully discussed. In addition to collisional and gravitational contributions, we identify the diffuse analogue of the blurring and kinetic Sunyaev-Zel'dovich (SZ) effects. Our approach naturally includes the correlations between the imprint from patchy reionisation and the diffuse SZ effects thereby allowing us to derive reliable estimates of the induced temperature and polarisation CMB angular power spectra. In particular, we show that the B-modes generated at intermediate length-scales (l~100) have the same amplitude as the B-modes coming from primordial gravitational waves with a tensor-to-scalar ratio r=10^{-4}.

astro-ph.CO↗