About

Romain Teyssier is a Professor of Astrophysical Sciences and Applied and Computational Mathematics at Princeton University. His expertise includes cosmology, galaxy formation, and star formation. He is the main author of the RAMSES code, a massively parallel Adaptive Mesh Refinement code designed for self-gravitating, magnetized, radiative flows. His primary research involves performing simulations of cosmic structures using supercomputers to understand the origins of astrophysical objects such as stars and galaxies. Additionally, he models the evolution of the entire Universe within the context of Euclid and LSST. Teyssier has held positions at the University of Zurich and CEA Saclay, and has a background that includes a PhD in Astrophysics from Paris Saclay University and education at Ecole Polytechnique.

Research topics

• Physics
• Astrophysics
• Astronomy
• Theoretical physics

Selected publications

• The SPHINX public data release. II. Using low-ionisation absorption lines and dust attenuation to predict Lyman continuum escape

  Springer Link (Chiba Institute of Technology) · 2026-03-17

  articleSenior author

  Low-ionisation state (LIS) metal absorption lines, such as siiilineb, are widely used to trace the properties and dynamics of the interstellar medium (ISM) in galaxies. These lines provide crucial insights into galaxy evolution, including feedback mechanisms, metal enrichment, and the escape fraction of ionising photons (fesc) during the epoch of reionisation. We expand our understanding of LIS absorption lines as diagnostic tools for ISM properties and fesc. Using the high-resolution sphinx 20 $ cosmological radiation-hydrodynamics simulation, we generated a comprehensive synthetic dataset of LIS absorption lines and tested their predictive power for $ in star-forming galaxies. Synthetic ISM absorption lines, focusing on siiiline and siiilineb, were computed with the radiative transfer code rascas , incorporating resonant scattering of photons, fluorescent emission, and interactions with dust grains. The simulated data enhance the public sphinx 20 $ dataset with high-resolution LIS lines for the full 1380 galaxies and ten viewing angles per galaxy. We analysed correlations between line properties (width, depth, and Doppler shift), dust attenuation, and $ extending previous single-galaxy studies to a statistically significant mock galaxy sample. We also tested our predictions on observed data using the LzLCS and CLASSY surveys. We found a strong correlation between the dust-corrected residual flux of siiilineb, tilde R ≡ ⋅ 10^ R_ flux ^ 1526 -0.4A_ 1500 , and fesc. More precisely, we found fesc ≈ 1.041 ^ R 1.887 - 0.002, with an average absolute error of 0.02. When we applied observational conditions, the error increased, but the escape fraction was still well recovered. In particular, the measurement of residual fluxes required a very high spectral resolution, and the dust attenuation is not directly observable. We show by applying common tools for fitting the spectral energy distribution to our mock data that the inferred dust attenuation is often far from the correct value, with a tendency to underestimate the attenuation when the effect of dust is strongest. Our results demonstrate that the residual flux of siiilineb is a powerful predictor of the escape fraction of ionising photons when it is corrected for dust. The spectra, line measurements, and escape fraction values used in this work are made publicly available.

  PublisherDOI

• Numerical cosmology

  SciPost Physics Lecture Notes · 2026-01-30

  articleOpen access1st authorCorresponding

  In these lecture notes, we describe the current state-of-the-art for numerical simulations of large-scale structure and galaxy formation. Numerical simulations play a central role in the preparation and the exploitation of large-scale galaxy surveys, in which galaxies are the fundamental observational objects. We first describe basic methods for collisionless N-body dynamics that enable us to model dark matter accurately by solving the Vlasov-Poisson equations. We then discuss simple methods to populate dark matter halos with galaxies, such as Halo and Sub-halo Abundance Matching techniques and baryonification techniques for capturing baryonic effects on the matter distribution. We finally describe how to model the gas component by solving the Euler-Poisson equations, focusing on the foundational assumptions behind these equations, namely local thermodynamical equilibrium, and the nature of the truncation errors of the numerical scheme, namely numerical diffusion. We show a few examples of simulations of a Milky-Way-like halo without cooling, with cooling and with star formation. We finally describe different subgrid prescriptions recently developed to model star formation, supernovae feedback and active galactic nuclei and how they impact cosmological simulations.

  PublisherOA PDFDOI

• Cosmology with one galaxy: An analytic formula relating $Ω_{\rm m}$ with galaxy properties

  ArXiv.org · 2026-02-07

  articleOpen access

  Standard cosmological analyses typically treat galaxy formation and cosmological parameter inference as decoupled problems, relying on population-level statistics such as clustering, lensing, or halo abundances. However, classical studies of baryon fractions in massive galaxy clusters have long suggested that gravitationally bound systems may retain cosmological information through their baryonic content. Building on this insight, we present the first analytic and physically interpretable cosmological tracer that links the matter density parameter, $Ω_m$, directly to intrinsic galaxy-scale observables, demonstrating that cosmological information can be extracted from individual galaxies. Using symbolic regression applied to state-of-the-art hydrodynamical simulations from the CAMELS project, we identify a compact functional form that robustly recovers $Ω_m$ across multiple simulation suites (IllustrisTNG, ASTRID, SIMBA, and Swift-EAGLE), requiring only modest recalibration of a small number of coefficients. The resulting expression admits a transparent physical interpretation in terms of baryonic retention and enrichment efficiency regulated by gravitational potential depth, providing a clear explanation for why $Ω_m$ is locally encoded in galaxy properties. Our work establishes a direct, interpretable bridge between small-scale galaxy physics and large-scale cosmology, opening a complementary pathway to cosmological inference that bypasses traditional clustering-based statistics and enables new synergies between galaxy formation theory and precision cosmology.

  PublisherOA PDFDOI

• Redshift Evolution of the Ratio of Supermassive Black Hole Mass to Stellar Mass

  arXiv (Cornell University) · 2026-05-06

  preprintOpen accessSenior author

  We run and analyze a suite of high-redshift zoom-in cosmological simulations with varying supernova feedback and supermassive black hole (SMBH) accretion prescriptions to study the joint evolution of stellar and SMBH mass in high-redshift galaxies down to $z=10$. The simulations reproduce the observed high-$z$ $M_{\mathrm{BH}}/M_{\star}$ relation if super-Eddington accretion is allowed prior to the final self-regulated phase. To extend the evolution to lower redshift, we model subsequent black hole and host growth using analytic halo assembly histories combined with a redshift-dependent effective Eddington duty cycle, $f_{\rm duty}=0.0004(1+z)^3$, calibrated to observations at $z\le6$, with conservative uncertainties at higher redshift. Within this framework, $M_{\mathrm{BH}}/M_{\star}$ exhibits a broad peak at $z\sim7$--10, reaching a few percent up to $\sim30\%$, followed by a steady, approximately power-law decline toward $z=0$. The model predicts $M_{\mathrm{BH}}/M_{\star}\sim(0.002,0.003,0.006,0.016,0.071,0.156)$ at $z=(0,1,2,3,5,10)$, consistent with available observations. This evolution is driven by rapid SMBH growth at high redshift, with effective mass e-folding times shorter than those of stellar mass, while at later times galaxy growth dominates, leading to the decline in $M_{\mathrm{BH}}/M_{\star}$. These results demonstrate that the emergence of a high-redshift peak and subsequent decline is robust despite uncertainties in the duty-cycle normalization.

  PublisherDOI

• The AGORA High-resolution Galaxy Simulations Comparison Project. X. Formation and Evolution of Galaxies at the High-redshift Frontier

  The Astrophysical Journal · 2026-03-27

  articleOpen accessSenior author

  Abstract Recent observations from the James Webb Space Telescope have revealed unexpectedly luminous galaxies, exhibiting stellar masses and luminosities significantly higher than predicted by theoretical models at Cosmic Dawn. In this study, we present a suite of cosmological zoomed-in simulations targeting high-redshift ( z ≥ 10) galaxies with dark matter halo masses in the range 10 10 –10 11 M ⊙ at z = 10, using state-of-the-art galaxy formation simulation codes ( Enzo , Ramses , Changa , Gadget-3 , Gadget-4 , and Gizmo ). This study aims to evaluate the convergence of the participating codes and their reproducibility of high-redshift galaxies with the galaxy formation model calibrated at relatively low redshift, without additional physics for high-redshift environments. The subgrid physics follows the AGORA CosmoRun framework, with adjustments to resolution and initial conditions to emulate similar physical environments in the early Universe. The participating codes show consistent results for key galaxy properties (e.g., stellar mass), but also reveal notable differences (e.g., metallicity), indicating that galaxy properties at high redshifts are highly sensitive to the feedback implementation of the simulation. Massive halos ( M halo ≥5 × 10 10 M ⊙ at z = 10) succeed in reproducing observed stellar masses, metallicities, and UV luminosities at 10 ≤ z ≤ 12 without requiring additional subgrid physics, but tend to underpredict those properties at higher redshift. We also find that varying the dust-to-metal ratio modestly affects UV luminosity of simulated galaxies, whereas the absence of dust significantly enhances it. In future work, higher-resolution simulations will be conducted to better understand the formation and evolution of galaxies at Cosmic Dawn.

  PublisherDOI

• The SPHINX public data release

  Astronomy and Astrophysics · 2026-03-16

  articleOpen accessSenior author

  Context. Low-ionisation state (LIS) metal absorption lines, such as Si II λ 1526, are widely used to trace the properties and dynamics of the interstellar medium (ISM) in galaxies. These lines provide crucial insights into galaxy evolution, including feedback mechanisms, metal enrichment, and the escape fraction of ionising photons ( f esc ) during the epoch of reionisation. Aims. We expand our understanding of LIS absorption lines as diagnostic tools for ISM properties and f esc . Using the high-resolution SPHINX 20 cosmological radiation-hydrodynamics simulation, we generated a comprehensive synthetic dataset of LIS absorption lines and tested their predictive power for f esc in star-forming galaxies. Methods. Synthetic ISM absorption lines, focusing on Si II λ 1260 and Si II λ 1526, were computed with the radiative transfer code RASCAS , incorporating resonant scattering of photons, fluorescent emission, and interactions with dust grains. The simulated data enhance the public SPHINX 20 dataset with high-resolution LIS lines for the full 1380 galaxies and ten viewing angles per galaxy. We analysed correlations between line properties (width, depth, and Doppler shift), dust attenuation, and f esc , extending previous single-galaxy studies to a statistically significant mock galaxy sample. We also tested our predictions on observed data using the LzLCS and CLASSY surveys. Results. We found a strong correlation between the dust-corrected residual flux of Si II λ 1526, R ∼ ≡ R flux 1526 · 10 −0.4 A 1500 , and f esc . More precisely, we found f esc ≈ 1.041 R ∼ 1.887 −0.002, with an average absolute error of 0.02. When we applied observational conditions, the error increased, but the escape fraction was still well recovered. In particular, the measurement of residual fluxes required a very high spectral resolution, and the dust attenuation is not directly observable. We show by applying common tools for fitting the spectral energy distribution to our mock data that the inferred dust attenuation is often far from the correct value, with a tendency to underestimate the attenuation when the effect of dust is strongest. Conclusions. Our results demonstrate that the residual flux of Si II λ 1526 is a powerful predictor of the escape fraction of ionising photons when it is corrected for dust. The spectra, line measurements, and escape fraction values used in this work are made publicly available.

  PublisherDOI

• The SPHINX public data release. II. Using low-ionisation absorption lines and dust attenuation to predict Lyman continuum escape

  arXiv (Cornell University) · 2026-03-17

  preprintOpen accessSenior author

  Low-ionisation state (LIS) absorption lines, such as SiII 1526, are widely used to trace the properties of the interstellar medium (ISM) in galaxies. These lines provide crucial insights into galaxy evolution, including feedback mechanisms, metal enrichment, and the escape fraction of ionising photons ($f_{\rm{esc}}$). We expand our understanding of LIS absorption lines as diagnostic tools for ISM properties and $f_{\rm{esc}}$. Using the SPHINX20 cosmological radiation-hydrodynamics simulation, we generated a comprehensive synthetic dataset of LIS absorption lines and tested their predictive power for $f_{\rm{esc}}$ in star-forming galaxies. Synthetic SiII 1260 and SiII 1526 lines were computed with the radiative transfer code RASCAS, incorporating resonant scattering of photons, fluorescent emission, and interactions with dust grains. The simulated data enhance the public SPHINX20 dataset with high-resolution LIS lines for the full 1380 galaxies and ten viewing angles per galaxy. We analysed correlations between line properties, dust attenuation, and $f_{\rm{esc}}$. We also tested our predictions on observed data using the LzLCS and CLASSY surveys. We found a strong correlation between the dust-corrected residual flux of SiII 1526, $\tilde{R} \equiv \rm{R_{flux}^{1526}} \cdot 10^{-0.4A_{1500}}$, and $f_{\rm{esc}}$. We found $f_{\rm{esc}} \approx 1.041\tilde{R}^{1.887} - 0.002$, with small error bars. When we applied observational conditions, the error increased, but the escape fraction was still well recovered. We show by applying common tools for fitting the spectral energy distribution to our mock data that the inferred dust attenuation is often far from the correct value, with an underestimation of the attenuation when the effect of dust is strongest. Our results demonstrate that the residual flux of SiII 1526 is a powerful predictor of the escape fraction of ionising photons.

  PublisherDOI

• Redshift Evolution of the Ratio of Supermassive Black Hole Mass to Stellar Mass

  ArXiv.org · 2026-05-06

  articleOpen accessSenior author

  We run and analyze a suite of high-redshift zoom-in cosmological simulations with varying supernova feedback and supermassive black hole (SMBH) accretion prescriptions to study the joint evolution of stellar and SMBH mass in high-redshift galaxies down to $z=10$. The simulations reproduce the observed high-$z$ $M_{\mathrm{BH}}/M_{\star}$ relation if super-Eddington accretion is allowed prior to the final self-regulated phase. To extend the evolution to lower redshift, we model subsequent black hole and host growth using analytic halo assembly histories combined with a redshift-dependent effective Eddington duty cycle, $f_{\rm duty}=0.0004(1+z)^3$, calibrated to observations at $z\le6$, with conservative uncertainties at higher redshift. Within this framework, $M_{\mathrm{BH}}/M_{\star}$ exhibits a broad peak at $z\sim7$--10, reaching a few percent up to $\sim30\%$, followed by a steady, approximately power-law decline toward $z=0$. The model predicts $M_{\mathrm{BH}}/M_{\star}\sim(0.002,0.003,0.006,0.016,0.071,0.156)$ at $z=(0,1,2,3,5,10)$, consistent with available observations. This evolution is driven by rapid SMBH growth at high redshift, with effective mass e-folding times shorter than those of stellar mass, while at later times galaxy growth dominates, leading to the decline in $M_{\mathrm{BH}}/M_{\star}$. These results demonstrate that the emergence of a high-redshift peak and subsequent decline is robust despite uncertainties in the duty-cycle normalization.

  PublisherOA PDF

• Cosmic-Ray and Plasma Coupling for Isothermal Supersonic Turbulence in the Magnetized Interstellar Medium

  The Astrophysical Journal · 2026-04-08

  preprintOpen access

  Abstract Cosmic rays (CRs) are an integral part of the nonthermal pressure budget in the interstellar medium (ISM) and are the leading-order ionization mechanism in cold molecular clouds. We study the impacts that different microphysical CR diffusion coefficients and streaming speeds have on the evolution of isothermal, magnetized, turbulent plasmas, relevant to the cold ISM. We utilized a two-moment CR magnetohydrodynamic model, allowing us to dynamically evolve both CR energy and flux densities with contributions from Alfvénic streaming and anisotropic diffusion. We identify coupled and decoupled regimes, and define dimensionless Prandtl numbers Pm c and Pm s , which quantify whether the plasma falls within these two regimes. In the coupled regime—characteristic of slow streaming (Pm s < 1) and low diffusion (Pm c < 1)—the CR fluid imprints upon the plasma a mixed equation of state between P c ∝ ρ 4/3 (relativistic fluid) and P c ∝ ρ 2/3 (streaming), where P c is the CR pressure and ρ is the plasma density. By modifying the sound speed, the coupling reduces the turbulent Mach number, and hence the amplitude of the density fluctuations, while supporting secular heating of the CR fluid. In contrast, in the decoupled regime (Pm s > 1 or Pm c > 1) the CR fluid and the plasma have negligible interactions. We further show that CR heating is enabled by coherent structures within the compressible velocity field, with no impact on the turbulence spectrum of incompressible modes.

  PublisherDOI

• Comparison between a priori and a posteriori slope limiters for high-order finite volume schemes: the Euler equations

  SSRN Electronic Journal · 2026-01-01

  preprintOpen accessSenior author
  PublisherDOI

Frequent coauthors

• F. Bournaud

  339 shared

• Jean‐Luc Starck

  CEA Cadarache

  235 shared

• A. Amara

  193 shared

• S. Farrens

  192 shared

• Alexandre Réfrégier

  183 shared

• S. Pires

  Université Paris Cité

  172 shared

• V. Pettorino

  Astrophysique, Instrumentation et Modélisation

  169 shared

• S. Pires

  CEA Paris-Saclay

  142 shared

Labs

• Department of Astrophysical Sciences, Princeton UniversityPI

Education

• PhD in Astronomy and Astrophysics, Physics

  Université Paris Cité Faculté de Santé

  1996

• Master in Science

  ENSTA ParisTech

  1994

• Bachelor

  École Polytechnique

  1992