Going Wide and Deep with Roman: The z~6-9 UV luminosity function in a Roman Deep Field

This paper presents a trade study for the Nancy Grace Roman Space Telescope, recommending an ultra-deep survey covering at least 0.56 square degrees with all six filters to optimize the measurement of the high-redshift (z~6–9) UV luminosity function and significantly reduce uncertainties in rest-UV luminosity density compared to existing JWST programs.

The Gist

Imagine the universe as a giant, dark ocean, and the very first galaxies are like tiny, bioluminescent fish swimming in the deepest, darkest parts of that ocean. For decades, astronomers have been trying to count these fish and understand how they swim, but they've been using flashlights that are either too weak or too narrow.

This paper is a blueprint for a new, super-powered flashlight (the Nancy Grace Roman Space Telescope) and a guide on how to use it to finally get a clear picture of the early universe.

Here is the breakdown of the research in simple terms:

1. The Problem: The "Flashlight Dilemma"

Imagine you are trying to count fish in a massive, dark ocean.

• The Old Way (Hubble & JWST): You have a flashlight that is incredibly bright (very deep), but it only shines on a tiny patch of water (a small area). If you happen to shine it on a school of fish, you think there are millions of fish. If you shine it on an empty spot, you think there are none. This is called Cosmic Variance. It's like trying to guess the population of a whole country by counting people in just one city block. You might get lucky, or you might get a very wrong answer.
• The New Way (Roman): The Roman telescope has a flashlight that is just as bright as the old ones, but it shines on a patch of water 100 times wider. This allows astronomers to see the "average" number of fish, smoothing out the lucky or unlucky spots.

2. The Experiment: A "Trade-Off" Study

The authors didn't just guess; they ran a massive simulation. They built a virtual universe inside a computer containing over 7.6 million fake galaxies. Then, they simulated taking 16 different types of "photos" with the Roman telescope to see which setup worked best.

They tested different combinations of:

• Area: How wide the photo is (1 spot, 2 spots, 4 spots, or 7 spots).
• Depth: How long they stare at the spot to see fainter objects.
• Filters: The colored lenses they use to see different colors of light.

3. The Big Discoveries (The "Secret Sauce")

The study found that to get the best results, you can't just pick any combination. You need a specific recipe:

• The "Red" Filter (r062) is Essential:
  Imagine trying to spot a red car in a foggy forest. If you only have a green filter, the red car looks black and invisible. To find the very first galaxies (which are very far away and their light has stretched to red), you must have a specific filter that looks at the "red" part of the spectrum. Without it, you might mistake a nearby, dusty galaxy for a distant one. The study found that without this filter, your sample could be 100% wrong (contaminated). With it, the error drops to almost zero.

• The "Deep" Filter (F184) is the Safety Net:
  For the very oldest, faintest galaxies (the ones at the edge of the universe), you need an extra "deep" filter to confirm they are actually there and not just a trick of the light. This filter also helps distinguish real galaxies from brown dwarfs (failed stars that look like galaxies but aren't).

• Don't Spread Yourself Too Thin:
  The study showed that it's better to have a moderately large area (about 2 Roman "pointings" or spots) that is very deep, rather than a huge area that is shallow. If you spread your time too thin over a huge area, you miss the faintest fish. If you focus on a slightly smaller area but stare longer, you catch the faint ones and get enough data to avoid the "Cosmic Variance" problem.

4. The Recommendation: The "Perfect Shot"

Based on their simulations, the authors recommend a specific plan for the Roman telescope:

• Target: Two specific spots in the sky (covering about 0.56 square degrees).
• Time: About 500 hours of observation.
• Filters: Use all six available filters (including the critical red and deep ones).

Why this matters:
If we do this, we can measure the density of light in the early universe with 2 to 4 times more precision than our current best telescopes (like JWST). This helps us answer a huge question: Did these first galaxies have enough power to turn the universe from dark to light? (This process is called "Reionization").

The Bottom Line

Think of this paper as a chef testing recipes for the perfect soup. They realized that to get the best flavor (scientific data), you need the right ingredients (filters) and the right cooking time (depth). If you skip the "red" ingredient, the soup tastes wrong. If you cook it too fast over a huge pot, it's watery.

Their conclusion? Cook a slightly smaller pot, but cook it longer, and make sure you have all the spices. This will give us the clearest, most accurate view of the universe's "dawn" that we have ever had.

Technical Summary

Here is a detailed technical summary of the paper "Going Wide and Deep with Roman: The z ∼6 −9 UV luminosity function in a Roman Deep Field."

1. Problem Statement

Understanding the growth of galaxies during the Epoch of Reionization ($z \sim 6-9$) relies on precise measurements of the rest-frame ultraviolet (UV) luminosity function (LF). However, current observations face a fundamental trade-off between survey depth (to detect faint galaxies that dominate the ionizing photon budget) and survey area (to mitigate cosmic variance).

• Hubble Space Telescope (HST): Deep fields (e.g., HUDF) reach great depths but cover tiny areas ($\sim 5$ arcmin$^2$), suffering from severe cosmic variance that leads to large field-to-field variations in measured galaxy densities.
• James Webb Space Telescope (JWST): While deeper, current JWST programs (e.g., JADES, CEERS, NGDEEP) still cover relatively modest areas. Combining heterogeneous surveys introduces systematic uncertainties.
• The Gap: There is a lack of a comprehensive trade study optimizing the depth-area-filter parameter space for the upcoming Nancy Grace Roman Space Telescope to maximize high-redshift galaxy science. Specifically, it is unclear how many pointings and which filter combinations are required to accurately recover the UV LF and the integrated UV luminosity density ($\rho_{UV}$).

2. Methodology

The authors conducted a comprehensive trade study using a mock galaxy catalog derived from the Santa Cruz Semi-Analytic Model (SAM) embedded in a $2 \text{ deg}^2$ lightcone from the Small MultiDark Planck (SMDPL) N-body simulation.

• Mock Catalog: Contains over 7.6 million galaxies at $0 < z < 10$ with realistic clustering, star formation histories, and synthetic photometry.
• Survey Configurations: The study evaluated 16 distinct survey strategies, each totaling $\sim 500$ hours of exposure time.
  • Areas: 0.28, 0.56, 1.12, and 2.0 $\text{deg}^2$ (corresponding to 1, 2, 4, and $\sim$7 Roman pointings).
  • Filter Sets: A "Base" set ($z087, Y106, J129, H158$) was tested with additions of the $r062$ filter, the $F184$ filter, or both.
• Simulation Process:
  1. Exposure Time Calculation: Used Pandeia and NASA Goddard performance tables to determine depths for each configuration.
  2. Photometric Scattering: Applied non-Gaussian noise (modeled via a Voigt profile) to mock fluxes to simulate realistic detection limits and photometric scatter.
  3. Sample Selection: Used EAZY for photometric redshift ($z_{phot}$) fitting with a custom template set (including BPASS templates). Selection criteria included $S/N \ge 5$ in $H158$, specific redshift probability distribution function (PDF) constraints, and dropout requirements.
  4. Analysis: Calculated recovery fractions, contamination rates (from low-$z$ galaxies and stars), rest-UV luminosity functions (fitting Schechter functions via MCMC), and integrated UV luminosity density ($\rho_{UV}$).

3. Key Contributions

• Quantification of Cosmic Variance: Demonstrated that a single Roman pointing ($0.28 \text{ deg}^2$) reduces cosmic variance uncertainty by factors of 3–10 compared to HST-like surveys (HUDF + HFF parallels) and significantly outperforms current JWST combinations.
• Filter Optimization: Identified the critical roles of the $r062$ and $F184$ filters, which are often omitted in deep field strategies due to time constraints.
• Stellar Contamination Modeling: Developed a method to separate cool stars/brown dwarfs from high-$z$ galaxies using color-color space, proving that $r062$ and $F184$ are essential for this separation.
• Trade-off Analysis: Provided a rigorous assessment of how increasing survey area (and thus decreasing depth) impacts the recovery of the faint-end slope ($\alpha$) and $\rho_{UV}$.

4. Key Results

A. Cosmic Variance and Luminosity Function Recovery

• HST Limitations: Single HST-sized pointings show $\sim 1$ dex variation in bright-end densities and Schechter parameters ($\phi^*, M^*, \alpha$). Even combining 9 HST pointings (HFF+) fails to fully constrain the LF due to field-to-field variations.
• Roman Advantage: A single Roman pointing recovers the "true" luminosity function with high fidelity, reducing variations in Schechter parameters by factors of 8–20 compared to HST.

B. Filter Necessity

• $r062$ Filter (Critical for $z \sim 5-6$):
  • Without $r062$, sample contamination at $z \sim 6$ approaches 80–90% (mostly from low-$z$ red galaxies).
  • Inclusion of $r062$ reduces contamination to negligible levels ($<10\%$) and improves recovery from $\sim 0\%$ to nearly $100%$for$z < 6.5$.
  • It is essential for distinguishing the Lyman-$\alpha$ break from the red slopes of dusty low-$z$ galaxies.
• $F184$ Filter (Critical for $z > 9$ and Stellar Rejection):
  • Improves recovery at $z > 9$ by providing a second detection band redward of the break.
  • Stellar Contamination: Cool stars and brown dwarfs mimic high-$z$ galaxies. Separating them requires a two-step color selection involving both $r062$ and $F184$. Without $F184$, stellar contamination is significantly higher, especially at $z \gtrsim 9$ where galaxies are unresolved.

C. Depth vs. Area Trade-off

• $\rho_{UV}$ Sensitivity: The integrated UV luminosity density is highly sensitive to the faint-end slope ($\alpha$).
• The "Deep is Better" Finding: As survey area increases (e.g., to $2.0 \text{ deg}^2$), the depth per filter decreases. This leads to a steeper measured faint-end slope at$z \sim 9$due to incompleteness, causing an **overestimation of$\rho_{UV}$**.
• Optimal Strategy: The study concludes that depth is more crucial than area for measuring $\rho_{UV}$. A survey covering **two Roman pointings ($0.56 \text{ deg}^2$)** with **all six filters** ($r062, z087, Y106, J129, H158, F184$) offers the best balance. This configuration reduces uncertainties on$\rho_{UV}$ by factors of 2–4 compared to the deepest existing JWST programs.

5. Significance and Recommendations

• Scientific Impact: A Roman ultra-deep field with the recommended configuration will provide the first robust, low-variance measurement of the high-redshift UV luminosity function, directly informing models of cosmic reionization and the contribution of faint galaxies to the ionizing photon budget.
• Strategic Recommendation: The authors recommend that a Roman ultra-deep survey should:
  1. Cover at least two pointings ($0.56 \text{ deg}^2$).
  2. Include all six filters ($r062$ through $F184$).
  3. Synergy with Core Surveys: Ideally, this should be implemented by adding depth and the missing $r062$ filter to the High Latitude Time Domain Survey (HLTDS) Deep Tier. The HLTDS Deep Tier is expected to reach similar depths but currently lacks $r062$; adding $\sim 37$ hours of $r062$ imaging would create an optimal ultra-deep field.
• Future Outlook: This study establishes that Roman can overcome the limitations of HST and JWST by simultaneously achieving HUDF-like depths and wide-area coverage, fundamentally changing the precision of early-universe galaxy science.