Disentangling the galactic and intergalactic components in 313 observed Lyman-alpha line profiles between redshift 0 and 5

Using the zELDA code to analyze 313 Lyman-alpha spectra, this study disentangles galactic and intergalactic effects to reveal that intrinsic galactic line profiles remain remarkably stable from redshift 0 to 6, while the intergalactic medium's transmission efficiency drops significantly at higher redshifts, eventually dominating Lyman-alpha observability beyond $z \approx 5$.

The Gist

Here is an explanation of the paper, translated into everyday language with some creative analogies.

The Big Picture: A Cosmic Game of "Whisper in the Wind"

Imagine you are standing in a vast, dark forest (the Universe). A friend in a distant clearing (a Galaxy) is shouting a specific word: "Lyman-Alpha" (a specific color of light).

You want to hear exactly what they shouted to understand their voice, their mood, and how they are moving. But there are three obstacles between you and your friend:

1. The Room (ISM): The room your friend is in is full of fog and furniture. Their voice bounces off the walls and gets muffled before it even leaves the room.
2. The Garden (CGM): Outside the room is a garden with tall grass and bushes. The sound has to weave through these plants, changing its shape again.
3. The Forest (IGM): Finally, the sound travels through the vast forest between you. The wind blows, trees rustle, and the sound gets scattered, absorbed, or shifted in pitch before it reaches your ear.

The Problem: When you finally hear the sound, it's a mess. You don't know if the friend shouted quietly, if the room was foggy, or if the wind just stole half the words. For decades, astronomers struggled to figure out which part of the "noise" came from the galaxy itself and which part came from the empty space (the Intergalactic Medium, or IGM) in between.

The Solution: This paper introduces a new tool called zELDA. Think of zELDA as a super-smart AI detective that can listen to the messy, distorted sound and mathematically "un-mix" it. It separates the friend's original voice (the galaxy) from the wind and trees (the IGM).

---

What Did They Do?

The team took 313 real recordings of these cosmic shouts from galaxies ranging from our local neighborhood (low redshift) to the very edge of the observable universe (high redshift). They fed these recordings into their AI (zELDA), which had been trained on millions of computer simulations.

The AI's job was to answer two questions:

1. What did the galaxy actually sound like? (Reconstructing the "intrinsic" profile).
2. How much of the sound did the forest steal? (Measuring the "escape fraction").

The Key Discoveries

1. The "Wind" Gets Stronger the Further You Go

When looking at nearby galaxies (close to us), the "forest" (IGM) is very clear. The wind is calm. The AI found that for these close galaxies, almost all the light escapes the forest. It's like shouting in a quiet park; your friend hears you clearly.

However, as they looked at galaxies from the early universe (billions of years ago), the forest became a dense, foggy jungle. The AI found that the "wind" (IGM) started swallowing the high-pitched parts of the shout (the "blue" side of the light). By the time we look at galaxies from 12 billion years ago, the forest is so thick that it blocks more than half the light.

2. The Galaxy's Voice Didn't Change Much

Here is the most surprising part. Once the AI stripped away the "forest noise" (the IGM effects), they looked at the original voice of the galaxies.

They found that galaxies have sounded remarkably similar for the last 13 billion years.

Whether the galaxy was young and close or ancient and far away, the way the light bounced around inside the galaxy (the "Room" and "Garden") didn't change much. The reason the light looked different to us wasn't because the galaxies changed; it was because the space between them changed. The "forest" got denser and more neutral over time, swallowing more light as we look further back.

3. The "Tipping Point"

The paper identifies a specific moment in cosmic history, around 12 billion years ago (Redshift 5).

• Before this time: The main reason we couldn't hear the galaxies clearly was the dust and gas inside the galaxies themselves.
• After this time: The main reason we can't hear them is the IGM (the space between galaxies).

It's like a party where, at first, the music is hard to hear because the room is full of people (dust). But as the party moves to a huge, windy field, the wind (IGM) becomes the main problem, drowning out the music even if the room is empty.

Why Does This Matter?

Think of the universe as a giant puzzle. For a long time, we were trying to solve the picture of the early universe, but we were looking at a blurry photo. We didn't know if the blur was because the subject was moving or because the camera lens was dirty.

This paper says: "We cleaned the lens."

By using zELDA to remove the "dirty lens" (IGM effects), astronomers can now:

• See the true shape and speed of ancient galaxies.
• Understand how the "fog" of the early universe cleared up (a process called Reionization).
• Predict which galaxies we should be able to see with new telescopes like the James Webb Space Telescope.

The Bottom Line

The authors built a digital "noise-canceling headphone" for the universe. They used it to listen to 313 ancient galaxies and realized that while the space between them has gotten much "noisier" and more opaque over time, the galaxies themselves have been singing the same song for billions of years. The change we see isn't in the singers; it's in the hall they are singing in.

Technical Summary

Here is a detailed technical summary of the paper "Disentangling the Galactic and Intergalactic Components in 313 Observed Lyman-Alpha Line Profiles Between Redshift 0 and 5".

1. Problem Statement

The Lyman-alpha (Ly$\alpha$) spectral line is a primary tool for detecting distant galaxies and probing the interstellar (ISM), circumgalactic (CGM), and intergalactic (IGM) media. However, the observed Ly$\alpha$ line profile is the result of complex radiative transfer processes across these three distinct environments.

• The Degeneracy: It is difficult to distinguish whether specific features in an observed spectrum (e.g., the suppression of the blue peak) are caused by galactic outflows (ISM/CGM) or by absorption from neutral hydrogen in the IGM.
• The Challenge: While the evolution of the IGM's mean optical depth is well-studied, the specific contribution of the IGM to the shape of individual Ly$\alpha$ line profiles and the resulting Ly$\alpha$ IGM escape fraction ($f_{\text{esc}}$) has not been directly measured on a source-by-source basis across cosmic time. Previous studies often relied on stacked profiles or assumed the IGM was negligible at low redshifts.

2. Methodology

The authors utilize the open-source Python package zELDA (redshift Estimator for Line profiles of Distant Lyman-Alpha emitters) to disentangle galactic and intergalactic effects.

• Data Source: The study analyzes 313 observed Ly$\alpha$ spectra from the Lyman Alpha Spectral Database (LASD), spanning redshifts $0 < z < 6$.
  • HST/COS: 111 sources ($z < 0.55$), high spectral resolution ($R \sim 16,000$).
  • MUSE: 202 sources ($z > 3.0$), lower resolution ($R \sim 3,000$).
• The zELDA Pipeline:
  • Training Data: zELDA employs Artificial Neural Networks (ANNs) trained on mock spectra generated via Monte Carlo radiative transfer (using the LyaRT code) through "thin-shell" models.
  • IGM Modeling: The training incorporates IGM transmission curves derived from the TNG100 cosmological simulation (Byrohl & Gronke 2020).
  • Three ANN Models:
    1. IGM+z: Trained with IGM attenuation curves specific to the source's redshift (encodes expected redshift evolution).
    2. IGM-z: Trained with IGM curves randomized across redshifts (redshift-independent control).
    3. NoIGM: Trained without IGM attenuation (fits a shell model directly to observed data).
• Reconstruction Process: The ANNs take the observed spectrum as input and output:
  1. The systemic redshift.
  2. The intrinsic thin-shell parameters (expansion velocity $V_{\text{exp}}$, column density $N_H$, dust optical depth, etc.).
  3. The IGM escape fraction ($f_{\text{esc}}^{4\mathring{A}}$), defined as the ratio of flux emerging from the ISM/CGM to the flux observed after IGM transmission within a $\pm 2\mathring{A}$ window.
  4. A reconstructed intrinsic spectrum (the profile before IGM attenuation).

3. Key Contributions

• First Direct Measurement: This is the first study to directly measure the redshift evolution of the Ly$\alpha$ IGM escape fraction ($f_{\text{esc}}$) using individual spectra rather than just stacked averages.
• Disentanglement of Components: Successfully separates the effects of galactic outflows (ISM/CGM) from IGM absorption, allowing for the reconstruction of the "pre-IGM" intrinsic line profile.
• Validation of zELDA: Demonstrates that zELDA can robustly recover intrinsic line shapes and redshifts ($\sim 0.3\mathring{A}$ accuracy) even in the presence of significant IGM attenuation.
• Low-Redshift Anomaly: Provides evidence that IGM attenuation may be non-negligible even at $z < 0.5$, challenging the assumption that the IGM is transparent at low redshifts for all sightlines.

4. Key Results

A. Evolution of the IGM Escape Fraction ($f_{\text{esc}}$)

• Low Redshift ($z < 0.5$): The IGM attenuation is minimal but non-zero.
  • $IGM\text{-}z$ model predicts $\langle f_{\text{esc}}^{4\mathring{A}} \rangle \approx 0.89 \pm 0.02$.
  • $IGM\text{+}z$ predicts $\approx 0.92 \pm 0.02$ (likely slightly biased high).
• High Redshift ($z > 3$): Significant attenuation occurs, primarily suppressing the blue peak of the Ly$\alpha$ line.
  • $\langle f_{\text{esc}}^{4\mathring{A}} \rangle$ decreases from $\sim 0.85$ at $z=3$ to $\sim 0.55$ at $z=5$.
  • The decline follows a modulated Fermi-Dirac distribution, consistent with independent constraints on the IGM mean optical depth.

B. Intrinsic Galactic Evolution

• After correcting for IGM effects, the stacked intrinsic galactic Ly$\alpha$ line profiles show remarkably little evolution from $z=0$ to $z=6$.
• The observed evolution in Ly$\alpha$ profiles (specifically the disappearance of the blue peak at high $z$) is driven almost entirely by IGM absorption, not by changes in the physical properties of the galactic outflows (ISM/CGM).

C. Comparison with Global Escape Fractions

• The measured IGM escape fraction ($f_{\text{esc}}$) is systematically higher than the global Ly$\alpha$ escape fraction ($f_{\text{Ly}\alpha}^{\text{esc}}$) found in literature (which includes ISM/CGM losses).
• Dominance Shift:
  • At $z < 3$, the ISM/CGM dominates the attenuation (global $f_{\text{Ly}\alpha}^{\text{esc}} < 0.1$, while IGM $f_{\text{esc}} \sim 0.8$).
  • At $z \approx 5$, the IGM becomes the primary absorber, accounting for $\sim 60\%$ of the total attenuation, while ISM/CGM accounts for $\sim 40\%$.

D. Low-Redshift IGM Features

• The models predict that even at $z < 0.5$, some sources exhibit IGM absorption features (e.g., small flux drops in the blue peak) similar to the Ly$\alpha$ forest. Six specific examples (Sources A–F in Fig. 11) show deviations from the smooth shell model that are consistent with IGM absorption.

5. Significance and Implications

• Cosmic Reionization: The results provide a direct probe of the IGM's neutral fraction and its impact on galaxy observability, confirming that the IGM becomes the dominant barrier to Ly$\alpha$ visibility at $z \gtrsim 5$.
• Galaxy Kinematics: For high-redshift studies, using the reconstructed intrinsic profile (rather than the observed one) is crucial for accurately measuring galactic outflow velocities and column densities, as the IGM significantly distorts the blue wing of the line.
• Future Applications: The methodology enables the potential for 3D tomography of the IGM. By mapping $f_{\text{esc}}$ across different sightlines, future work could correlate IGM density fluctuations with the large-scale structure of the universe.
• Methodological Robustness: The agreement between the redshift-dependent ($IGM\text{+}z$) and redshift-independent ($IGM\text{-}z$) models validates that the measured trends are driven by the data, not by training set priors.

In conclusion, this work establishes that the IGM is a dynamic and dominant filter for Ly$\alpha$ radiation at high redshifts, and that tools like zELDA are essential for recovering the true physical properties of star-forming galaxies across cosmic time.