Modeling CO Line Luminosity Using FIRE Simulations: I. Methodology and Validation of Molecular Gas and CO Abundance

arXiv:2610.12159 · astro-ph.GA, astro-ph.CO · Submitted 2026-10-08 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Astrophysics Radio. Generated commentary on the latest astrophysics papers.

Vera: Today's paper: "Modeling CO Line Luminosity Using FIRE Simulations".

Jocelyn: The combination of these excerpts allows for a highly detailed and accurate synthesis of the methodology, validation steps, and key findings.

Vera: First, who's behind it and why it matters.

Title and authors: Vera: The paper summarizes this pipeline as a way to predict CO line luminosities, molecular hydrogen masses, and CO abundances for galaxies within the FIRE-two simulations <ref:2610.12159#pg1,CO}$ line luminosities, molecular hydrogen masses, and>. It frames it as a necessary first step toward generating those response functions mentioned earlier.

Jocelyn: They highlight that this method allows us to compute the H two and CO abundances and line luminosities of each simulated gas cell as a function of their simulated properties, which moves beyond just using empirical relations <ref:2610.12159#pg1>.

Subrahmanyan: That ability to compute those properties based on the simulation parameters is what gives this work its power, because it lets us test the physics in a controlled way before we even look at real data.

Vera: The validation part of the paper shows they are testing this pipeline against several observational domains, including molecular gas masses and CO luminosities.

Jocelyn: They show good agreement with observations for molecular gas masses at z=zero when comparing to xCOLD GASS for star formation rates above zero point three M yr-one.

Subrahmanyan: But they also point out that at high redshifts, the observed gas masses seem to exceed the simulation predictions for fixed stellar mass, though they admit that comparison is limited by selection effects and assumed conversion factors.

The paper's summary: Vera: One of the main improvements suggested by this work is focusing on capturing how metallicity affects the dust shielding of CO from photodissociation in a way that’s consistent across different galaxy types and redshifts.

Jocelyn: They show that a factor of three increase in metallicity causes an order-of-magnitude change in the total L'CO(one-zero), which is crucial for accurately modeling CO emission in both low-mass and high-redshift galaxies <ref:2610.12159#pg3>.

Subrahmanyan: That dependence on metallicity is a very important theoretical insight because it shows that we can't just treat the dust properties as fixed constants across all cosmic time and environments.

Vera: They also show that the onset of CO luminosity occurs when individual clouds reach an optical depth of around one, which tells us about efficient shielding from the surrounding radiation field.

Jocelyn: So, for someone who only listens to this show, it means we can better model how much gas is actually emitting light in different galaxies depending on how dusty they are.

The paper's improvements: Vera: To wrap up the FIRE simulations paper, the authors conclude that this pipeline provides a rigorous way to predict these molecular properties and is a crucial first step for creating those CO response functions.

Jocelyn: They stress that this method is designed to connect large-volume dark matter maps with the simulated physics through those response functions, which we see illustrated in Figure one of the paper <ref:2610.12159#pg1>.

Subrahmanyan: From a theoretical standpoint, the implication is that we are getting closer to a framework where we can connect cosmological structure—the web of dark matter—to observable galaxy properties in a way that is constrained by detailed physics within those simulations.

Vera: So, this work sets up the necessary tools for making those large-volume maps more predictive, and they are moving toward using this for mock surveys.

Jocelyn: It’s a big step in how we interpret the integrated line emission from many unresolved sources, since LIM surveys can map spatial fluctuations in that line emission.

Subrahmanyan: It’s about connecting the microscopic physics of a single gas cell to the macroscopic structure of the entire cosmic web, which is always a challenging task.

Vera: That’s all we have for this paper on modeling CO line luminosity using FIRE simulations. We’ll be looking at what these response functions actually enable in our next session.

Conclusion: Vera: So we’ve been looking at this paper, "Modeling CO Line Luminosity Using FIRE Simulations: I. Methodology and Validation of Molecular Gas and CO Abundance." Essentially, they built a whole pipeline to predict CO line luminosities, molecular hydrogen masses, and CO abundances inside the FIRE-two simulations.

Jocelyn: Right? And it’s not just a simple lookup table; they used SKIRT for the radiation field calculation and then fed that into CLOUDY to get all those line luminosities. It’s a pretty detailed process for figuring out what gas is actually doing.

Subrahmanyan: From my end, what I find really compelling is how they handle the metallicity scaling, because they show that just a factor of three increase in metallicity can change the total L'CO(one-zero) by an order of magnitude. That means we need to be super careful about those assumptions when looking at different galaxies.

Vera: Exactly, Subrahmanyan. And the validation against observations is pretty strong, especially when they look at molecular gas masses at z=zero. They agreed well with xCOLD GASS for star formation rates above zero point three M yr-one.

Jocelyn: I also see that sharp rise in CO luminosity predicted for Milky Way clouds when the column density gets around one, which seems to match what we see in the data. It’s a nice consistency there.

Subrahmanyan: But they also admit their geometry is a big simplification; they treat each gas cell as a plane-parallel slab and don't transport line photons between cells, which is something we definitely need to keep in mind when we try to scale this up.

Vera: So, the main point here is that capturing that metallicity-dependent dust shielding of CO from photodissociation is essential for modeling CO emission across different redshifts.

Jocelyn: And they frame this whole thing as a crucial first step toward building those response functions we talked about in other work, which will help us paint those large-volume dark matter maps.

Subrahmanyan: That connection between the simulation results and the eventual LIM surveys is where the real power of this pipeline lies; it moves us away from just seeing what happens in a single cloud to mapping out how everything interacts on a larger scale.

Vera: So, to wrap up, this paper lays down the groundwork for predicting molecular gas properties within FIRE-two simulations by using SKIRT and CLOUDY, and it validates those predictions against observations like L'CO(one-zero) rises in Milky Way clouds.

Jocelyn: And they show that understanding how metallicity affects dust shielding is key for modeling CO emission in both low-mass galaxies and high-redshift systems.

Subrahmanyan: It’s a solid methodological foundation, but we still have to worry about those geometric simplifications when we try to push these models into the next phase of cosmological simulations.

Vera: Indeed, this is just the first part of building that full response function, and it sets us up well for what comes next.

Jocelyn: Next time we talk about how these calculated luminosities actually get convolved with those beam profiles to make mock surveys.

Doga Tolgay ˘, Norman W. Murray, Dongwoo Chung, J. Richard Bond, Nathan J. Carlson, Robert Feldmann, Andrew Wetzel, Guochao Sun

Canadian Institute for Theoretical Astrophysics, University of Toronto Department of Physics, Cornell University Department of Astronomy, Department of Astrophysics, University of Zurich Department of Physics & Astronomy, University of California Davis Department of Physics & Astronomy, CIERA and Department of Physics and Astronomy Northwestern University

astro-ph.GA, astro-ph.CO

Submitted: 2026-10-08

Updated: 2026-10-08

Comments: 31 pages, 14 figures, 2 tables

Code: https://github.com/mikegrudic/meshoid

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Importance score: 89/100

The gist: The combination of these excerpts allows for a highly detailed and accurate synthesis of the methodology, validation steps, and key findings.

Key concepts

SKIRT
This is a Monte Carlo dust radiative transfer code used to calculate the local interstellar radiation field reaching every gas cell in the simulation. It determines how much ultraviolet light is available to affect molecular clouds, which is essential input for subsequent physical modeling of CO emission.
CLOUDY
A photoionization code used to model how radiation from SKIRT affects molecular gas. Instead of running it on every cell, the authors create a lookup table based on five key parameters (like metallicity and density) to quickly find the resulting $ ext{CO}$ luminosities and masses for any given environment.
L'CO(1-0)
This is a specific measure of how bright the CO emission line at 1-0 km/s is, which directly relates to the amount of molecular gas present. The pipeline successfully predicted this luminosity, showing it rises sharply when clouds reach certain optical depths ($ ext{A}_V$), matching real astronomical observations.
Dust Shielding
This refers to how dust within a galaxy protects CO molecules from being destroyed by intense ultraviolet radiation. The research shows that the metallicity-dependent shielding of CO is crucial for accurately modeling its emission, especially in galaxies with low mass or those observed at high redshifts.

Terminology

Summary

The combination of these excerpts allows for a highly detailed and accurate synthesis of the methodology, validation steps, and key findings.

Here is the long and detailed summary:

Detailed Research Summary: Post-Processing Pipeline for Molecular Gas Properties in FIRE-2 Simulations

This research presents a sophisticated post-processing pipeline designed to predict critical molecular gas properties—specifically CO line luminosities (L'CO(1-0)), molecular hydrogen (H 2) masses, and CO abundances—for galaxies within the Feedback in Realistic Environments (FIRE-2) cosmological simulations. The ultimate goal of this initial work is to establish a foundation for developing ** CO response functions** necessary for line intensity mapping mocks.

I. Core Methodology: The Post-Processing Pipeline

The pipeline integrates three major computational components:

  1. Radiation Field Calculation (SKIRT): The process begins by employing the three-dimensional Monte Carlo dust radiative transfer code, SKIRT. This code is used to compute the local interstellar radiation field (ISRF) reaching every individual gas cell within the simulation volume. The radiation field considered is specifically in the Far-Ultraviolet (FUV) band, defined between 6 and 13.6 eV.

  2. Molecular Physics Modeling (CLOUDY): The resulting radiation field from SKIRT is then fed into the CLOUDY photoionization code. Instead of running CLOUDY on every gas cell individually—which would be computationally prohibitive—the authors adopt a crucial simplification: they pre-calculate and tabulate the emission per unit mass. They run CLOUDY on a massive grid of 109,296 models spanning five key parameters (metallicity, density, turbulent velocity, radiation field strength, and cloud scale), which are then interpolated onto the simulated gas cells. This lookup table yields line luminosities (L'CO(1-0), L'CO(2-1)), H alpha, H beta, [OII], [OIII], and [NII] luminosities, alongside the molecular masses (M H 2 and M CO).

  3. Interpolation: The luminosity-per-mass values obtained from the CLOUDY look-up table are then used to assign the specific line luminosities to each gas cell in the FIRE-2 simulation, as detailed in Section 4.4 of the full paper.

II. Simulation Setup and Assumptions (Crucial Caveats)

The pipeline operates on a foundation built upon several key simulation choices and necessary simplifications:

  • Simulation Suites: The analysis utilizes various FIRE-2 simulation suites to cover different cosmic epochs and galaxy types:

  • Core Suite: Zoom-in simulations of Milky Way (MW) and Large Magellanic Cloud (LMC) mass galaxies run to z=0.

  • Triple Latte Suite: Higher-resolution resimulations of m12 i galaxies.

  • MassiveHalo (MassiveFIRE) Suite: Zoom-in simulations run to z=1.

  • FIREbox Suite: Simulates the full cosmological volume with baryonic physics, focusing on the entire volume rather than individual zoom-in regions.

  • Gas Cell Definition: Simulated gas cells are defined as star particles when they meet strict criteria: exceeding a local density threshold, being predominantly molecular, and being locally self-gravitating and Jeans unstable.

  • Simplifying Assumptions (Caveats): The authors explicitly acknowledge significant simplifications in their approach, which must be noted for accurate interpretation:

  1. Geometry: Each gas cell is treated as a one-dimensional, constant-density, plane-parallel slab. Crucially, the model does not transport line photons between cells.

  2. Dust Properties: A fixed dust-to-metal ratio of 0.4 and a single grain mixture (Weingartner & Draine 2001) are adopted across all galaxies and redshifts.

  3. Radiation Field Input: Only the FUV normalization of the SKIRT radiation field is passed to CLOUDY, assuming a fixed Milky Way spectral shape.

  4. Metallicity Scaling: Elemental abundances are scaled using only the total metal mass fraction of each gas cell, relative to an adopted solar metallicity (Z = 0.0134), rather than tracking the specific carbon and oxygen abundances tracked by FIRE-2. The authors stress that this assumption is highly sensitive to metallicity.

  5. Model Consistency: All galaxies originate from a single galaxy formation model.

III. Validation and Key Findings Against Observations

The pipeline's predictive power is rigorously tested against observational data across multiple domains:

  • Molecular Gas Masses (M mol):

  • At z=0, the predicted molecular gas masses agree well with local observations from xCOLD GASS for star formation rates (SFR) 0.3 M yr-1, and the predictions are converged across simulation resolutions.

  • At high redshifts (z 1), observed gas masses exceed simulation predictions at fixed stellar mass, though this comparison is limited by selection effects and assumed conversion factors.

  • ** CO Luminosities (L'CO(1-0)):**

  • For Milky Way (MW) molecular clouds, the predicted L'CO(1-0) shows a sharp rise at A v about 1, matching observed values above this column.

  • For a uniform 3Z environment, the summed luminosity is within 35% of observed totals.

  • A factor of three increase in metallicity causes an order-of-magnitude change in total L'CO(1-0).

  • The onset of CO luminosity is correctly identified as occurring when individual clouds reach A v about 1, indicating efficient shielding from the surrounding radiation field.

  • Abundances (NCO/NH 2):

  • The predicted NCO/NH 2 ratio matches observations below (NH 2 / cm-2) about 20.5.

  • The pipeline successfully recovers the sharp upturn observed in Core MW-mass galaxies at (NH 2) about 3 - 5 times 10 20 cm-2.

  • Along MW sight lines, the predictions reproduce the observed upturn within 0.3 dex in NH 2 but fall below data at higher column densities.

IV. Broader Context and Future Applications

The research emphasizes that capturing the metallicity-dependent dust shielding of CO from photodissociation is essential for accurately modeling CO emission in both low-mass and high-redshift galaxies. The pipeline itself is presented as a crucial first step toward building the necessary response functions required for line intensity mapping forecasts, which are detailed in Paper II and Paper III. Furthermore, the final panel of the paper demonstrates how this calculated L'CO(1-0) for each gas cell, incorporating radiative post-processing, is used to paint large-volume dark matter maps with response functions (g(M halo; z=0)) and convolve them with beam profiles (e.g., LIM-like beams) to generate mock surveys.

Improvements for AI systems

  1. This pipeline can predict CO line luminosities, molecular hydrogen masses, and CO abundances for galaxies in the Feedback in Realistic Environments (FIRE-2) simulations by using a post-processing approach involving SKIRT for radiation field calculation and CLOUDY for line emission prediction.

  2. The improved AI system can perform Line Intensity Mapping (LIM) by producing mock maps of CO intensity that paint the large-volume dark matter only WebSky maps with the g(Mhalo; z = 0) response function of Paper III and convolve with a LIM-like beam.

  3. The system can constrain galaxy physics by predicting molecular gas content using its pipeline, as it can compute H2 and CO abundances and line luminosities of each simulated gas cell as a function of their simulated properties, rather than relying on empirical relations.

  4. The AI can validate its predictions against observations by comparing the predicted L′CO(1−0) rises sharply at AV ∼ 1 for Milky Way (MW) molecular clouds with observed values and comparing predicted NCO/NH2 versus NH2 ratios with those from Galactic sight lines.

  5. The system can model the impact of metallicity on CO emission, as it demonstrates that a factor of three increase in metallicity changes the total L′CO(1−0) by an order of magnitude, which is crucial for accurately modeling CO emission in low-mass and high-redshift galaxies.

Sources

Related papers