A cosmic filament catalog for COSMOS-Web Data Release 1 across 0.5 < z < 3.5

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

Vera: I'm Vera.

Jocelyn: And I'm Jocelyn, here with Subrahmanyan.

Vera: We're looking at the paper: "A cosmic filament catalog for COSMOS-Web Data Release 1 across 0.5 < z < 3.5".

Jocelyn: Here's the gist of the paper. The gist The researchers constructed a filament catalog spanning 0.5 < z < 3.5 using deep-field data and advanced methods to study galaxy evolution and large-scale structure at cosmic noon; this work provides a robust foundation for investigating galaxy–environment coevolution at cosmic noon and beyond The new procedure developed is based on the DisPerSE method, which fully incorporates the probability distribution functions (PDFs) of galaxy photometric redshifts by generating multiple realizations for each tomographic 2D slice The researchers then stacked these realizations to construct “filament probability maps” (wfil maps), from which filaments were subsequently extracted The filament extraction parameters were optimized using mock catalogs generated from TNG simulations by minimizing the discrepancies between 3D and projected 2D filaments Data and Catalogs The study utilizes the deep-field catalog COSMOS-Web Data Release 1 across 0.5 < z < 3.5 Observational data includes the COSMOS2025 catalog, which contains 784,016 sources with photometric redshifts (zphot) measured with a precision of ∼0.01 for mF444W < 24 The analysis spans 67 tomographic 2D slices, each with a thickness of 150 cMpc, covering the redshift range from z = 0.5 to z = 3.5 To verify completeness and purity, a mock catalog was constructed from the IllustrisTNG simulations using TNG300-1, which has a volume of 205,000 ckpc/h Filament Extraction and Calibration The process involves several key steps to mitigate photometric-redshift uncertainties and projection biases The method includes: 1.

Vera: Constructing a “filament probability map” (wfil map) by stacking filaments extracted from multiple PDF(z) realizations per tomographic slice 2.

Jocelyn: Calibrating optimal extraction parameters through mock experiments with TNG300-1 simulations 3.

Vera: Compiling a filament catalog that fully exploits the PDF(z) while maximizing consistency with the 3D structures The optimal persistence threshold nσ,pred was identified by minimizing the similarity metric Ξ between F3D,ref and F2D,mock extracted from wfil maps Validation and Results The extracted filaments were validated against astrophysical properties to show a strong agreement with simulations Key validations include:: Galaxy stellar mass function inside and outside filaments, which shows an upturn in the normalized GSMF at the high-mass end for galaxies inside filaments The difference ∆Φ ≡ Φin − Φout between the inside- and outside-filament GSMFs crosses from negative to positive at M⋆ ∼ 109.5 M⊙, confirming that high-mass galaxies preferentially reside in filaments Connectivity (Codis et al.

Title and authors: Vera: So we're looking at this paper called "A cosmic filament catalog for COSMOS-Web Data Release one across zero point five < z < three point five <ref:2610.31168#pg1,A cosmic filament catalog for COSMOS-Web Data Release 1 across 0>." It’s a lot of data from the deep-field COSMOS-Web release, and they’ve put together a way to map out the cosmic web structure across that redshift range.

Jocelyn: It seems like they’re tackling that big question about how galaxies actually sit within these filaments, especially around cosmic noon. They use this catalog to build something more robust than just looking at 2D slices of the sky <ref:2610.31168#pg1>.

Subrahmanyan: What I find interesting is that they are trying to bridge the gap between what simulations show and what we can actually observe with photometric redshifts. That uncertainty is usually a big hurdle when you try to pin down three dee structure <ref:2610.31168#pg1>.

Vera: Exactly, they’re using a method called DisPerSE, which involves generating multiple realizations based on the probability distributions of those photometric redshifts. They stack those together to create something called "filament probability maps," or wfil maps.

Jocelyn: That sounds like they're trying to smooth out the noise in the data by looking at many possibilities at once instead of just taking one snapshot. So, how do they actually turn those probability maps into a catalog of filaments?

Vera: They extract the filaments from those wfil maps, and they’ve optimized the extraction parameters using mock catalogs generated from TNG simulations. They minimized the difference between three dee and projected 2D filaments when doing that calibration <ref:2610.31168#pg1>.

Subrahmanyan: That part about minimizing discrepancies with TNG simulations is crucial because it gives them a benchmark for what a real filament should look like in three dimensions. If the method works well on those mocks, then we can trust what it finds in the real data.

Jocelyn: And they found some specific parameters that work best, like this persistence threshold nsigma pred, which is determined by minimizing a similarity metric called xi between their three dee and 2D mock extractions <ref:2610.31168#pg1>. That sounds like a very careful tuning process.

Title and authors: Vera: It means they aren't just dumping whatever the algorithm spits out; they are actively calibrating it to trace the underlying structure as well as possible across the entire redshift range from z equals zero point five up to three point five.

Subrahmanyan: That range is right around cosmic noon, which is when galaxy formation and large-scale structure growth were really at their peak, so getting a good map here is important for understanding how the cosmic web fuels star formation.

Jocelyn: So they’ve got this final product called the COSMOS-Web filament catalog, F2D obs, and these sixty-seven calibrated filament-probability maps for each slice. What does that actually mean for someone who just listens to this show?

Vera: It means you can look at any spot on the sky at any redshift in that range and get a probability, P, between zero and one of whether or not there’s a real filament there. This is encoded as ln(one plus wfil), where x is just the pixel value <ref:2610.31168#pg1>.

Subrahmanyan: That mapping itself is interesting because they modeled it as a piecewise linear plus saturation function. The near-zero intercept of that mapping encodes how the large-scale structure evolves with cosmic time.

Jocelyn: It sounds like a way to see the hierarchical growth of these structures directly from the observed probability distribution, which is powerful information for understanding structure formation over time.

Vera: They’ve validated this catalog against astrophysical properties, and they found some pretty strong agreement with simulations. For instance, looking at the galaxy stellar mass function inside and outside these filaments shows an upturn at the high-mass end for galaxies inside them.

Subrahmanyan: That's significant because it confirms that high-mass galaxies preferentially reside in filaments when you look at this redshift range, as shown by the difference delta phi between inside and outside filament galaxy mass functions crossing from negative to positive around stellar mass one hundred nine point five solar masses.

Jocelyn: And they also checked connectivity for galaxy groups, finding consistency between the COSMOS-Web data and TNG mock samples at the group-mass scale, which gives confidence in their extraction method.

Title and authors: Vera: So, to summarize this paper "A cosmic filament catalog for COSMOS-Web Data Release one across zero point five < z < three point five," they’ve developed a method using DisPerSE and mock calibration to produce a filament catalog that's robust against photometric redshift uncertainties, showing that high-mass galaxies prefer filaments and connectivity holds up when compared to TNG simulations <ref:2610.31168#pg1,A cosmic filament catalog for COSMOS-Web Data Release 1 across 0>.

Subrahmanyan: It sets a solid foundation for investigating galaxy–environment coevolution at cosmic noon and beyond, because now we have a better way to link the large-scale structure we see with the actual galaxy properties in those environments.

Jocelyn: It’s going to help future studies probe how these filaments influence star formation and AGN activity, which is exactly what the paper points toward when discussing the scarcity of robust filament identifications before cosmic noon.

Vera: We can now use these calibrated maps to map environmental trends directly, like seeing how galaxy mass functions change depending on whether you are inside or outside a filament at different redshifts.

Subrahmanyan: That capability to map environmental trends is what opens up the next step, connecting the observable structure of the cosmic web more firmly to cosmological models that predict how this structure forms.

Jocelyn: It’s a solid piece of data for studying how these large-scale structures guide galaxy evolution across cosmic time.

Vera: So, that's what we have with this work on "A cosmic filament catalog for COSMOS-Web Data Release one across zero point five < z < three point five <ref:2610.31168#pg1,A cosmic filament catalog for COSMOS-Web Data Release 1 across 0>." We’ve got a catalog and a probability map to explore how galaxies are organized in the cosmic web during this important epoch.

Subrahmanyan: It’s a big step forward for connecting galaxy properties to the large-scale structure we observe, and it gives us tools to test cosmological models in a way that's more constrained by real data.

Jocelyn: We’ll keep an eye on how this catalog helps settle those questions about fueling star formation and AGN activity within these cosmic environments.

Vera: That’s our take on "A cosmic filament catalog for COSMOS-Web Data Release one across zero point five < z < three point five <ref:2610.31168#pg1,A cosmic filament catalog for COSMOS-Web Data Release 1 across 0>." It gives us the tools to see the connection between filaments and galaxy properties across a wide range of redshift.

The paper's summary: Vera: So we just finished talking about how they built this catalog, and now we need to talk about what they actually found across that whole range from z equals zero point five up to three point five.

Jocelyn: Yeah, the main thing is they took all this raw data from COSMOS-Web and created something called a filament catalog, F2D obs, which is their two-dimensional version of these cosmic structures.

Vera: It’s not just a list of points; it’s these sixty-seven calibrated filament probability maps for each redshift slice. The core idea is that instead of just saying "this galaxy is in a filament," they are giving you the chance, P, between zero and one, that the location actually traces a true three-dimensional filament.

Jocelyn: That's the crucial part because it handles all that mess with photometric redshift uncertainties. They used this DisPerSE method to generate multiple realizations based on those probability distributions of the redshifts.

Vera: And then they optimized the extraction parameters using mock catalogs from TNG simulations, which is how they made sure their 2D extraction matched what a real three dee structure should look like in the simulation <ref:2610.31168#pg1>. They even found an optimal persistence threshold, nsigma pred, by minimizing a similarity metric between the extracted filaments and those perfect TNG references.

Jocelyn: That optimization step is where you really see the engineering side of it—how they tune the algorithm so it actually traces that underlying cosmic web without getting too noisy.

Vera: And what this means for us is that we have a tool now to map environmental trends directly across cosmic time, not just in snapshots. They validated this by checking things like the galaxy stellar mass function inside and outside these filaments.

Jocelyn: That validation was pretty telling, right? They saw an upturn in the normalized GSMF at high-mass ends for galaxies inside filaments, and they found that the difference between those inside and outside mass functions crosses over around stellar mass one hundred nine point five solar masses.

Vera: So it confirms what we suspected—that massive galaxies preferentially reside in these filaments as we look back to cosmic noon. It’s a direct observation linking galaxy properties to the large-scale structure.

Jocelyn: And they also checked connectivity, looking at how groups of galaxies are linked, and they found consistency between their COSMOS-Web data and those TNG mock samples at the group mass scale. That gives confidence in the whole extraction process.

Vera: The final product is these probability maps modeled as a piecewise linear plus saturation function, where that near-zero intercept actually encodes how the large-scale structure grows with cosmic time. It’s a way to see hierarchical growth directly from the observed probability distribution.

Jocelyn: So, we moved from just seeing structures to quantifying their presence and evolution across a wide stretch of cosmic history. This gives us a better handle on how these filaments guide galaxy formation across different epochs.

Vera: It's definitely a foundation for looking at galaxy-environment coevolution much further out than we could before this work. But what does this catalog tell us about the actual physics happening *inside* those filaments?

The paper's improvements: Tom: So we talked about what they found in the main paper, and now we’re going to look at how they think they can make this tool even better than it is right now.

Vera: The authors suggest a few key improvements to take this catalog from just a good map to something more useful for real science. First, there's this idea for an adaptive extraction method based on those probability maps we built.

Jocelyn: What they mean is that instead of using one fixed threshold, the system should be able to directly apply those calibrated filament probability maps to any tomographic slice you want.

Vera: So, if you pick a specific redshift slice, the AI can give you a per-pixel probability P between zero and one for whether that location actually traces a true three-dimensional filament. It’s described in Equation D1 and Table D1 of their work.

Jocelyn: That sounds like it makes the whole catalog much more flexible for researchers using it. And then there's the part about robust threshold calibration, which they tie into those mock experiments again.

Vera: They want the AI to use those mock experiments to determine the optimal extraction parameters by minimizing that similarity metric X between their extracted filaments and the TNG300-one reference skeleton. This ensures that nsigma pred is really doing its job in tracing the three dee structure in a way that’s best for simulations.

Jocelyn: That means if you were to use this tool on a new dataset, it wouldn't just give you an output; it would actively try to find the settings that make the most sense cosmologically. It’s like having an assistant tuning itself.

Vera: And they also look at validating those reconstructions against real astrophysical trends, which is where they test if their optimized thresholds actually matter in a physical sense. They check whether the reconstructed filament networks, F2D mock and F2D obs, recover things like galaxy mass gradients and that connectivity-mass relation at the group mass scale.

Jocelyn: So they’re not just checking if their math works; they’re checking if the resulting structure actually looks like what we expect from cosmological models. They found that the optimized thresholds do indeed yield a 2D skeleton that reliably traces the underlying three-dimensional connection <ref:2610.31168#pg1>.

Vera: That brings us to environmental property mapping, which is where you get to analyze trends without needing a perfect three dee view. The system can map these by analyzing normalized stellar mass functions inside and outside filaments, which shows that high-mass galaxies preferentially reside in those filaments across different redshift bins.

Jocelyn: It’s like we can now use this map to quickly see how the environment affects galaxy growth as we look back further in time. And they also have a specific analysis for connectivity of galaxy groups, calculating a metric to show that F2D obs recovers that positive trend at the group mass scale.

Vera: The main limitation they mention is that this method relies on these photometric redshifts, so the accuracy of your P value depends directly on how good those initial zphot measurements are. It’s not perfect because it’s based on observational constraints, not a direct three dee measurement.

Jocelyn: So what this changes for someone listening? It means you can use this catalog to probe environmental trends across cosmic time with much higher confidence than before, provided you understand the error bars coming from those initial redshift estimates.

Vera: This work gives us a clear path forward for using these filament maps as a direct probe of how structure dictates galaxy evolution. But next week, we’re going to talk about what these filaments actually *do*—how they influence star formation and black hole growth within them.

Conclusion: Vera: So we’re wrapping up this look at "A cosmic filament catalog for COSMOS-Web Data Release one across zero point five < z < three point five." Basically, they built a robust way to map the cosmic web structure using photometric data from z equals zero point five up to three point five.

Jocelyn: Right, it’s about moving beyond just seeing structures in snapshots and giving us a calibrated probability map for any spot on the sky at any redshift.

Vera: The big implication is we now have a tool that lets us study galaxy environment coevolution much more directly, because we can quantify how likely a location is to be part of a filament.

Jocelyn: It’s powerful because it’s designed to handle the mess from photometric redshifts, giving us confidence in the results through rigorous calibration against TNG simulations.

Subrahmanyan: From a theoretical perspective, this catalog gives us a massive observational anchor for testing structure formation models across cosmic time; if we can keep mapping these filaments consistently, we can constrain how dark matter halos merge and grow.

Vera: And that’s what they’ve done: they validated the results against galaxy mass functions, confirming that high-mass galaxies definitely prefer residing in those filaments.

Jocelyn: We also saw consistency in connectivity analyses between the real data and the TNG mocks at the group mass scale, which really boosts our trust in this whole extraction method.

Subrahmanyan: It’s a strong piece of evidence for how environmental density affects galaxy evolution at cosmic noon, and it provides a framework for future theoretical work on galaxy assembly in filaments.

Vera: So to recap, this paper provides a calibrated filament catalog and probability maps that link observable large-scale structure directly to the properties of galaxies inside them across a wide redshift range.

Jocelyn: It’s a really useful observational resource for anyone studying how cosmic web structures guide galaxy growth as we look back at earlier epochs.

Subrahmanyan: This work sets a high bar for how we should approach these observational constraints when building the next generation of cosmological simulations and tests.

Vera: So that’s our take on "A cosmic filament catalog for COSMOS-Web Data Release one across zero point five < z < three point five." It’s a solid foundation for connecting what we see in the sky with how galaxies evolve within those structures.

Jocelyn: It gives us the tools to map environmental trends directly, and that's huge for seeing how these filaments influence star formation and AGN activity across different redshifts.

Vera: We’ll keep an eye on how this catalog helps settle those questions about fueling star formation and AGN activity within these cosmic environments as we move into the next round of data analysis.

Jocelyn: It’s a big step forward for studying how these large-scale structures guide galaxy evolution across cosmic time, and we expect follow-up studies to use this probability map as a reference point.

Subrahmanyan: This is exactly what we need to see more robustly in future observations so that our simulations can accurately predict the physical processes occurring within these filaments.