An information-theoretic framework for quantifying coherence in the cosmic density field
Listen
Radio episode about this paper
Transcript
Introduction to the show: ident: Astrophysics Radio. Generated commentary on the latest astrophysics papers.
Vera: We're looking at the paper: "An information-theoretic framework for quantifying coherence in the cosmic density field".
Jocelyn: Here's the gist of the paper. An information-theoretic bound on cosmic coherence in finite-volume simulations quantifies how much physical information about the cosmic density field is shared between different epochs and establishes a fundamental limit on large-scale coherence imposed by finite simulation volumes Quantifying Shared Information The paper introduces an information-theoretic perspective to ask how much physical information about the cosmic density field is shared between the same comoving volume at two different epochs By comparing mutual information measured from cosmological Nbody simulations to that obtained from appropriately constructed random catalogs, the authors isolate the physically meaningful component of shared mutual information and track how it depends on spatial scale This method treats the evolving cosmic density field as an information channel where gravity transmits memory of the early state across scales and times Scale Dependence of Coherence The analysis shows that the shared physical mutual information between density fields at different redshifts is strongly suppressed on small scales, rises rapidly with increasing smoothing length, and reaches a clear maximum at a characteristic scale This characteristic scale occurs at approximately one-eighth of the simulation box size, d ≃ L/8, across boxes of different volumes Beyond this scale, the shared information declines sharply, indicating that the coherence of large-scale structure across cosmic time is fundamentally limited by the finite size of the simulated volume Robustness and Physical Interpretation The analysis confirms that the location of the peak in shared physical mutual information peaks at a voxel size approximately one-eighth of the box size, d ≃ L/8, for different simulation volumes This scaling demonstrates that the location of the peak is not associated with a universal physical scale of structure formation but is instead set by the finite size of the simulation volume The robustness tests indicate that the observed scale dependence is not an artifact of voxelization or binning but reflects genuine finite-volume constraints on the coherence of large-scale structure Implications for Large-Scale Structure The central result establishes a general and modelindependent limitation: any attempt to infer the largest coherent structures, the scale of homogeneity, or large-scale environmental effects using finite simulation boxes is intrinsically constrained by the available information content The suppression of shared physical mutual information beyond d ∼ L/8 reflects a fundamental finite-volume bound on the coherence that can be retained across cosmic time in N-body simulations This result suggests that large-scale coherence in cosmic structure and inferences about homogeneity are intrinsically limited by the absence of long-wavelength modes in N-body simulations Robustness to Evolution and Discretization The shared physical mutual information is robust across widely separated redshifts, with the peak persistence demonstrating that this scale is set by the finite volume rather than by nonlinear gravitational evolution Furthermore, the analysis shows that at sufficiently large voxel sizes, particularly in smaller simulation volumes (L = 128 Mpc), the shared physical mutual information Iphys(X; Y) becomes negative This negative value emerges when the physical signal is suppressed below this irreducible null floor The overall shape of Iphys(X; Y) including the rapid rise, the peak near d ≃ L/8, and the subsequent decline remains unchanged across binning choices The work provides a unified and physically transparent way to assess the impact of finite volume on cosmological inference by framing finite-volume effects in terms of shared information across cosmic time The authors conclude that the suppression of shared physical mutual information reported here reflects a loss of largescale coherence in the cosmic density field, rather than a reduction in clustering amplitude This approach opens new avenues for applying information theory to nonlinear gravitational systems The existence of extremely large observed structures requires cosmological N-body simulations spanning volumes of at least several gigaparsecs on a side so that finite-volume effects and missing long-wavelength modes do not artificially limit large-scale coherence References
1: S.
Vera: Bharadwaj, S.
Jocelyn: P.
Vera: Bhavsar, and J.
Paper summary: Vera: So Jocelyn you’re looking at this paper "An information-theoretic framework for quantifying coherence in the cosmic density field." The big idea here is using mutual information between density fields at different times to measure how much physical information about structure survives gravitational evolution across cosmic time.
Jocelyn: Right, it’s about quantifying that physical memory of the cosmic density field. What they’re claiming is that this approach isolates the excess statistical dependence generated by gravity, which is what conventional clustering statistics often miss because they don't capture all orders of correlation.
Subrahmanyan: That makes sense from a theoretical side, but I wonder if isolating that information actually reveals anything new about the growth of structure compared to things like power spectrum analysis. What’s the core mechanism here?
Vera: Well, they define this physical mutual information as the data mutual information minus a null mutual information calculated using independent random point distributions with the same particle number and box size. That subtraction is what isolates those genuine gravitational correlations, removing stuff caused just by the way we sample things or by shot noise.
Jocelyn: And they show that this shared physical mutual information has a strong dependence on spatial scale. It rises as you use a larger voxel size to coarse-grain the density field, peaks at a characteristic scale, and then declines again on larger scales.
Subrahmanyan: A peak in coherence across cosmic time is interesting because it suggests there’s an optimal length scale where structure retains its memory best, but what do those numbers say about the physical universe?
Vera: They found that this characteristic scale peaks at approximately one-eighth of the simulation box size, which is around six forty four Mpc for a five hundred twelve megaparsec box. This location of the peak isn't necessarily a universal physical scale of structure formation; it’s tied directly to the finite size of the simulation volume they used.
Jocelyn: So, if we use this method to study large-scale structure, what does that imply about what we can actually infer? The paper suggests that this peak location is governed by these finite sampling effects rather than some fundamental physical coherence scale.
Subrahmanyan: That’s a big caveat. It means when you try to infer things like the scale of homogeneity or very large-scale environmental effects, your measurements are intrinsically limited by the size of your simulation box because you’re missing those long-wavelength modes that would be needed to see them clearly.
Vera: Exactly. The suppression of this shared physical mutual information beyond that one eighth length scale shows a fundamental limit on how much coherence we can retain across cosmic time in N-body simulations. It suggests that if we want to talk about homogeneity, we really need simulation volumes spanning several gigaparsecs on each side so these finite-volume effects don't artificially cap our measurement of coherence.
Jocelyn: So the paper is essentially using this information theory tool to give us a way to quantify how much memory the cosmic density field has between redshift one and redshift two, while also telling us exactly where the simulation volume limits that memory.
Subrahmanyan: It’s a powerful new tool for quantifying coherence, yes. But I'm still curious about how this relates to different structure types. The paper mentions that filamentary environments retain more statistically accessible shared information between cosmic epochs than sheets and clusters.
Vera: That’s a specific finding because it tells us something about the geometry of the cosmic web. It suggests that if we want to study how much information is preserved, looking at filaments might give us a better signal than just looking at dense clusters or flat sheets.
Jocelyn: It’s interesting because they also found that the shared information between the initial and present-day density fields is positive but extremely small. That small positive value reflects how much information gets progressively redistributed as structure forms during nonlinear evolution.
Subrahmanyan: So, to wrap up, this paper shows that the coherence we measure isn't just about how much stuff is clustered now; it’s about the memory of structure across time and scale, and it puts a concrete limit on what that memory can be if your simulation box is too small.
Vera: Yeah, so for observational astronomers, this gives us a new way to frame our errors when we try to infer large-scale properties using density field statistics. It ties the statistical uncertainty directly to the physical volume of our observations or simulations.
Jocelyn: And for time-domain work, it helps us understand what’s preserved in the signal between different cosmic epochs, like comparing a snapshot at redshift one to one at redshift two, beyond just looking at standard two-point statistics.
Subrahmanyan: It opens up avenues for applying information theory to nonlinear gravitational systems in a way that’s sensitive to higher-order correlations. We need more observational data that can probe these scale dependencies beyond what current methods can provide with the available resolution.
Vera: That sounds like the next step for us—building surveys or simulations large enough to push past this finite volume constraint and see if we can find that coherence peak at a truly physical scale.
Jocelyn: It’s certainly a thought-provoking way to look at what’s happening on the largest scales of the universe.
Subrahmanyan: Indeed, it grounds our understanding of structure formation in the constraints imposed by our computational limits.
Conclusion: Vera: So, this paper, "An information-theoretic framework for quantifying coherence in the cosmic density field," is basically saying we can use mutual information to measure how much physical memory about structure survives gravitational evolution across cosmic time.
Jocelyn: Right, so it’s not just looking at how much stuff is clustered now, but tracking the shared information between two different epochs, like comparing a snapshot from redshift one to one from redshift two.
Subrahmanyan: The authors are using this mutual information between density fields in N-body simulations and subtracting a random baseline to isolate the part that gravity actually imprinted on the field.
Vera: They found that this shared physical mutual information shows a strong dependence on spatial scale, peaking at about one eighth of the simulation box size.
Jocelyn: And they stress that this peak isn't some universal physical scale; it’s actually set by the finite size of the box they used in their simulation.
Subrahmanyan: That means if you want to talk about homogeneity or very large-scale structure, your measurement is fundamentally limited by how big your simulation volume is.
Vera: Exactly, and they show that beyond that one eighth scale, the shared information drops off sharply, which points to a real limit on coherence across cosmic time in simulations.
Jocelyn: The authors also point out that filamentary environments retain more of this shared information between epochs than sheets or clusters do.
Subrahmanyan: And they noted the initial and present-day fields have only a very small positive mutual information, which makes sense because gravity is constantly redistributing that initial information during structure formation.
Vera: So, the main point here is using this mutual information as a new way to measure coherence and memory in the large-scale structure of the Universe.
Jocelyn: It opens up new ways to look at how structure evolves across time and scale in a way that goes beyond traditional clustering statistics.
Subrahmanyan: This method connects information theory to nonlinear gravitational systems, which is something we need more of as we try to understand the big picture.
Vera: It gives us a concrete limit on what we can infer about the largest coherent structures using these simulations, and it tells us exactly what kind of volume you need for those measurements to be meaningful.
Biswajit Pandey, Anindita Nandi
Visva-Bharati University
astro-ph.CO
Submitted: 2026-01-05
Updated: 2026-11-06
Comments: 20 pages, 10 figures, significantly revised and expanded, included new analyses on the dependence on cosmological parameters and cosmic web environments, Accepted for publication in JCAP
Code: https://github.com/rajeshmondal18/N-body
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Importance score: 84/100
The gist: An information-theoretic bound on cosmic coherence in finite-volume simulations quantifies how much physical information about the cosmic density field is shared between different epochs and…
Key concepts
- Mutual Information
- A mathematical measure quantifying the statistical dependence between two random variables. In this context, it measures how much knowledge about the density field at one time (redshift $z_1$) helps predict the density field at another time ($z_2$), capturing the 'memory' of gravitational evolution.
- Shared Physical Mutual Information
- This is the crucial quantity derived by subtracting a null value from total mutual information. The null value represents spurious correlations caused purely by finite sampling and discretization. The physical version isolates the genuine statistical dependence imprinted by gravity, showing how structure's memory evolves over time.
- Scale Dependence of Coherence
- The shared information between density fields is not uniform across all spatial scales. It rises sharply as the smoothing length increases, peaks at a characteristic scale ($d hickapprox L/8$), and then declines. This dependence reveals that coherence across cosmic epochs is fundamentally limited by the simulation's finite volume.
- Finite-Volume Limit
- The analysis demonstrates that the peak in shared physical mutual information occurs at a voxel size related to the box size ($d hickapprox L/8$), not at a universal physical scale of structure formation. This proves that inferences about large-scale homogeneity are intrinsically constrained by the absence of long-wavelength modes in finite N-body simulations.
Terminology
Summary
An information-theoretic bound on cosmic coherence in finite-volume simulations quantifies how much physical information about the cosmic density field is shared between different epochs and establishes a fundamental limit on large-scale coherence imposed by finite simulation volumes
Quantifying Shared Information
The paper introduces an information-theoretic perspective to ask how much physical information about the cosmic density field is shared between the same comoving volume at two different epochs By comparing mutual information measured from cosmological Nbody simulations to that obtained from appropriately constructed random catalogs, the authors isolate the physically meaningful component of shared mutual information and track how it depends on spatial scale This method treats the evolving cosmic density field as an information channel where gravity transmits memory of the early state across scales and times
Scale Dependence of Coherence
The analysis shows that the shared physical mutual information between density fields at different redshifts is strongly suppressed on small scales, rises rapidly with increasing smoothing length, and reaches a clear maximum at a characteristic scale This characteristic scale occurs at approximately one-eighth of the simulation box size, d ≃ L/8, across boxes of different volumes Beyond this scale, the shared information declines sharply, indicating that the coherence of large-scale structure across cosmic time is fundamentally limited by the finite size of the simulated volume
Robustness and Physical Interpretation
The analysis confirms that the location of the peak in shared physical mutual information peaks at a voxel size approximately one-eighth of the box size, d ≃ L/8, for different simulation volumes This scaling demonstrates that the location of the peak is not associated with a universal physical scale of structure formation but is instead set by the finite size of the simulation volume The robustness tests indicate that the observed scale dependence is not an artifact of voxelization or binning but reflects genuine finite-volume constraints on the coherence of large-scale structure
Implications for Large-Scale Structure
The central result establishes a general and modelindependent limitation: any attempt to infer the largest coherent structures, the scale of homogeneity, or large-scale environmental effects using finite simulation boxes is intrinsically constrained by the available information content The suppression of shared physical mutual information beyond d ∼ L/8 reflects a fundamental finite-volume bound on the coherence that can be retained across cosmic time in N-body simulations This result suggests that large-scale coherence in cosmic structure and inferences about homogeneity are intrinsically limited by the absence of long-wavelength modes in N-body simulations
Robustness to Evolution and Discretization
The shared physical mutual information is robust across widely separated redshifts, with the peak persistence demonstrating that this scale is set by the finite volume rather than by nonlinear gravitational evolution Furthermore, the analysis shows that at sufficiently large voxel sizes, particularly in smaller simulation volumes (L = 128 Mpc), the shared physical mutual information Iphys(X; Y) becomes negative This negative value emerges when the physical signal is suppressed below this irreducible null floor The overall shape of Iphys(X; Y) including the rapid rise, the peak near d ≃ L/8, and the subsequent decline remains unchanged across binning choices
Conclusion
The work provides a unified and physically transparent way to assess the impact of finite volume on cosmological inference by framing finite-volume effects in terms of shared information across cosmic time The authors conclude that the suppression of shared physical mutual information reported here reflects a loss of largescale coherence in the cosmic density field, rather than a reduction in clustering amplitude This approach opens new avenues for applying information theory to nonlinear gravitational systems The existence of extremely large observed structures requires cosmological N-body simulations spanning volumes of at least several gigaparsecs on a side so that finite-volume effects and missing long-wavelength modes do not artificially limit large-scale coherence
References
[1] S. Bharadwaj, S. p. 4 P. Bhavsar, and J. p. 4 V. Sheth, ApJ 606, 25 (2004), arXiv:astro-ph/0311342 [astro-ph]. p. 4
[2] J. Yadav, S. Bharadwaj, B. p. 4 Pandey, and T. p. 0 R. Seshadri, MNRAS 364, 601 (2005), arXiv:astro-ph/0504315 [astroph]. p. 4
[3] D. W. Hogg, D. J. Eisenstein, M. p. 4 R. Blanton, N. p. 4 A. Bahcall, J. p. 4 Brinkmann, J. p. 4 E. Gunn, and D. p. 4 P. Schneider, ApJ 624, 54 (2005), arXiv:astro-ph/0411197 [astro-ph]. p. 4
[4] P. Sarkar, J. Yadav, B. Pandey, and S. p. 0 p. 0 Bharadwaj, MNRAS 399, L128 (2009), arXiv:0906. p. 43431 [astro-ph. p. 4CO].
[5] B. Pandey, G. Kulkarni, S. p. 4 Bharadwaj, and T. p. 4 Souradeep, MNRAS 411, 332 (2011), arXiv:1009. p. 42223 [astro-ph. p. 4CO].
[6] M. I. e. a. Scrimgeour, MNRAS 425, 116 (2012), arXiv:1205. p. 46812 [astro-ph. p. 4CO].
[7] S. Nadathur, MNRAS 434, 398 (2013), arXiv:1306. p. 41700 [astro-ph. p. 4CO].
[8] B. Pandey and S. p. 0 Sarkar, MNRAS 454, 2647 (2015), arXiv:1507. p. 403124 [astro-ph. p. 4CO].
[9] S. Sarkar and B. p. 4 p. 4 Pandey, MNRAS 463, L12 (2016), arXiv:1607. p. 406194 [astro-ph. p. 4CO].
[10] S. Sarkar and B. Pandey, MNRAS 485, 4743 (2019), arXiv:1812. p. 403661 [astro-ph. p. 4CO].
[11] P. Sarkar, B. Pandey, and S. Sarkar, MNRAS 519, 3227 (2023), arXiv:2208. p. 409472 [astro-ph. p. 4CO].
[12] V. Springel, S. p. 4 D. M. White, A. Jenkins, C. p. 4 S. Frenk, N. Yoshida, L. p. 4 Gao, J. Navarro, R. p. 4 Thacker, D Croton, J Helly, J A Peacock Guldberg and F Pearce Nature (London) 435, 629 (2005), arXiv:astro-ph/0504097 [astro-ph]. p. 4
[13] J. Kim, C. Park, G. Rossi, S. M. Lee, and J R Gott III Journal of Korean Astronomical Society 44, 217 (2011), arXiv:1112. p. 41754 [astro-ph. p. 4CO].
[14] J Schaye, R A Crain R G Bower M Furlong M Schaller T Theuns C Dalla Vecchia C S Frenk I G McCarthy J C Helly A Jenkins Y M RosasGuevara S D M White M Baes C M Booth P Camps J F Navarro Y Qu A Rahmati T Sawala P A Thomas and J Trayford MNRAS 446, 521 (2015), arXiv:1407. p. 47040 [astro-ph. p. 4GA].
[15] D Nelson V Springel A Pillepich V RodriguezGomez P Torrey S Genel M Vogelsberger R Pakmor F Marinacci R Weinberger L Kelley M Lovell B Diemer and L Hernquist Computational Astrophysics and Cosmology 6, 2 (2019), arXiv:1812. p. 405609 [astro-ph. p. 4GA].
[16] J K Yadav J S Bagla and N Khandai MNRAS 405, 2009 (2010), arXiv:1001. p. 40617 [astro-ph. p. 4CO].
[17] B Pandey MNRAS 401, 2687 (2010), arXiv:91/O7/T
[18] C Park Y-Y Choi J Kim J R Gott III S S Kim and K-S Kim ApJ Letters 759, L7 (2012), arXiv:1209.5659 [astro-ph. p. 4CO].
[19] J S Bagla and J Prasad MNRAS 370, 993 (2006), arXiv:astro-ph/0601320 [astro-ph]. p. 4
[20] C Power and A Knebe MNRAS 370, 691 (2006), arXiv:astro-ph/0512281 [astro-ph]. p. 4
[21] T Nishimichi A Shirata A Taruya K Yahata S Saito Y Suto R Takahashi N Yoshida T Matsubara N Sugiyama I Kayo Y Jing and K Yoshikawa Publications of the Astronomical Society of Japan 61, 321 (2009), arXiv:0810. p. 40813 [astro-ph]. p. 4
[22] J R Gott III M Juri´c D Schlegel F Hoyle M Vogeley M Tegmark N Bahcall and J Brinkmann ApJ 624, 463 (2005), arXiv:astro-ph/0310571 [astro-ph]. p. 4
[23] M J West R De Propris M Einasto Z L Wen and J L Han ApJ Letters 987, L24 (2025), arXiv:2506. p. 419826 [astro-ph. p. 4CO].
[24] J Einasto arXiv e-prints arXiv:2509. p. 404252 [astro-ph. p. 4CO].
[25] M Einasto Universe 11, 167 (2025), arXiv:2505. p. 422082 [astro-ph. p. 4CO].
[33] C E Shannon Bell Labs Technical Journal 27, 623 (1948). p. 4
[34] B Pandey MNRAS 463, 4239 (2016), arXiv:1607. p. 402768 [astro-ph. p. 4CO].
[35] Planck Collaboration Astronomy & Astrophysics 641, A6 (2020), arXiv:1807. p. 406209 [astro-ph. p. 4CO].
[36] S Bharadwaj and P S Srikant Journal of Astrophysics and Astronomy 25, 67 (2004), arXiv:astro-ph/0402262 [astro-ph]. p. 4
[37] R Mondal S Bharadwaj S Majumdar A Bera and A Acharyya MNRAS 449, L41 (2015), arXiv:1409. p. 44420 [astro-ph. p. 4CO].
[38] S Raychaudhury Nature (London) 342, 251 (1989). p. 4
[39] R B Tully H Courtois Y Hoffman and D Pomar´ede Nature (London) 513, 71 (2014), arXiv:1409. p. 40880 [astroph. p. 4CO].
[40] J Bagchi S Sankhyayan P Sarkar S Raychaudhury J Jacob and P Dabhade ApJ 844, 25 (2017), arXiv:1707. p. 403082 [astro-ph. p. 4CO].
[41] M Einasto H Lietzen M Gramann E Saar E Tempel L J Liivam¨agi A D Montero-Dorta A Streblyanska C Maraston and J A Rubi˜no-Mart´ın Astronomy & Astrophysics 603, A5 (2017), arXiv:1703. p. 408444 [astroph. p. 4CO].
[42] H B¨ohringer G Chon J Tr¨umper R C KraanKorteweg and N Schartel Astronomy & Astrophysics 695, A59 (2025), arXiv:2501. p. 419236 [astro-ph. p. 4CO].
--- Page 1 ---
An information-theoretic bound on cosmic coherence in finite-volume simulations Biswajit Pandey and Anindita Nandi Department of Physics, Visva-Bharati University, Santiniketan, Birbhum, 731235, India. (Dated: January 6, 2026) We quantify the physical memory of the cosmic density field using mutual information between N-body snapshots at different redshifts removing a random baseline to isolate gravitational correlations. p. 0 This behaviour is robust to box size and discretization and identifies the largest coherently retained modes unaffected by missing long-wavelength power establishing a finite-volume limit on the coherence of cosmic structure with direct implications for homogeneity studies
--- Page 2 ---
finite volume Probability distributions of density: We consider two snapshots of an N-body simulation at redshifts z1 and z2 occupying the same cubic comoving volume of side length L The simulation volume is discretized into cubic voxels of side length d yielding Nvox = (L/d)3 voxels For each voxel i, we define the discrete random variables Xi ≡ ni(z1), Yi ≡ ni(z2) where ni(z) denotes the number of particles contained in voxel i at redshift z The set of Xi and Yi thus represent coarse-grained realizations of the matter density field at two different times The voxel counts are coarse-grained into nbin density bins to construct the probability distributions For a snapshot at redshift z1, the marginal probability of finding a voxel in bin a is pX (a) = Na / Nvox where Na is the number of voxels assigned to bin a and Nvox is the total number of voxels Similarly for another snapshot at redshift z2, we obtain pY (b) The joint probability, pXY (a, b), is constructed by pairing voxels at identical spatial locations across the two snapshots and counting their co-occurrences in bins (a, b) Physically, pX (a) and pY (b) encode the one-time density statistics at each epoch while pXY (a, b) captures the temporal correlation of the density field across cosmic time Mutual information between snapshots The mutual information [33] between the density fields at z1 and z2 is defined as Idata(X; Y) = nXbin a=1 nXbin b=1 pXY (a, b) log pXY (a, b) pX (a) pY (b) Idata(X; Y) measures the reduction in uncertainty in the density field at one redshift given knowledge of the field at the other reflecting their shared information By construction, Idata(X; Y) ≥ 0 with equality only if the two snapshots are statistically independent In physical terms, Idata(X; Y) quantifies the memory of the cosmic density field: it captures the amount of structure preserved during nonlinear gravitational evolution across time and scale Unlike two-point statistics, it is sensitive to all orders of correlation and therefore captures the full statistical imprint of nonlinear gravitational evolution [34]
--- Page 3 ---
Shared physical mutual information Finite sampling and voxelization can induce spurious correlations even in uncorrelated distributions To isolate the contribution from genuine gravitational correlations, we define the shared physical mutual information as Iphys(X; Y) ≡ Idata(X; Y) − Inull(d) where Idata(X; Y) is the mutual information between the voxelized density fields at redshifts z1 and z2 The null mutual information Inull(d) is defined as the mutual information computed using the same voxel size d between two independent structureless random point distributions with identical particle number and box size Inull(d) captures the spurious correlations arising purely from finite sampling and discretization in the absence of any gravitational signal The subtraction in Equation 4 removes spurious correlations arising from shot noise, discretization, and finite sampling thereby isolating the information imprinted by nonlinear gravitational evolution Physical interpretation The quantity Iphys(X; Y) has a direct and intuitive interpretation It measures how much information about the density field at redshift z2 is encoded in the density field at redshift z1 when both are viewed at a spatial resolution d In other words, it quantifies the memory retained by the cosmic density field across cosmic time on a given length scale It should be noted that mutual information is symmetric i.e.
Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.
Improvements for AI systems
- Information-theoretic bound on cosmic coherence
This approach allows AI systems to quantify how much physical information about the cosmic density field is shared between the same comoving volume at two different epochs,
enabling better assessment of memory in gravitational evolution.
- Robust inference for large-scale structure limits
AI models can now intrinsically incorporate a fundamental limit on inferred coherence, stating that any attempt to infer the largest scale of cosmic coherence or homogeneity using finite simulation boxes is intrinsically constrained by this information ceiling.
- Distinguishing physical memory from numerical artifacts
The system can differentiate genuine gravitational memory from spurious correlations by using shared physical mutual information Iphys(X; Y) ≡ Idata(X; Y) − Inull(d),
ensuring that inferences are based on correlations generated by gravitational evolution
rather than spurious correlations arising purely from finite sampling and discretization.
- Scale-dependent coherence prediction
AI can predict the characteristic scale of information retention, which is set by the simulation volume, by identifying where the shared physical mutual information peaks at a voxel size approximately one-eighth of the box size, d ≃ L/8.
- Robustness assessment for cosmological parameters
The system can perform more reliable inferences about large-scale structures and homogeneity by recognizing that the loss of coherence beyond this scale is not physical, but arises from the absence of long-wavelength modes in finite-volume simulations,
providing a modelindependent limitation on such inferences.
Sources
- The Size of the Longest Filaments in the Universe
- Testing homogeneity on large scales in the Sloan Digital Sky Survey Data Release One
- Cosmic homogeneity demonstrated with luminous red galaxies
- The scale of homogeneity of the galaxy distribution in SDSS DR6
- The size of the longest filament in the Luminous Red Galaxy distribution
- The WiggleZ Dark Energy Survey: the transition to large-scale cosmic homogeneity
- Seeing patterns in noise: Gigaparsec-scale `structures' that do not violate homogeneity
- Testing homogeneity in the Sloan Digital Sky Survey Data Release Twelve with Shannon entropy
- An information theory based search for homogeneity on the largest accessible scale
- Unravelling the Cosmic Web: An analysis of the SDSS DR14 with the Local Dimension
- The maximum extent of the filaments and sheets in the cosmic web: an analysis of the SDSS DR17
- Simulating the joint evolution of quasars, galaxies and their large-scale distribution
- The New Horizon Run Cosmological N-Body Simulations
- The EAGLE project: Simulating the evolution and assembly of galaxies and their environments
- The IllustrisTNG Simulations: Public Data Release
- Fractal Dimension as a measure of the scale of Homogeneity
- Statistically significant length scale of filaments as a robust measure of galaxy distribution
- The Challenge of the Largest Structures in the Universe to Cosmology
- Effects of the size of cosmological N-Body simulations on physical quantities -- I: Mass Function
- The Impact of Box Size on the Properties of Dark Matter Haloes in Cosmological Simulations
Related papers
- 21cmEMUv3: a hybrid diffusion-LSTM emulator of 21cmFAST summary observables
- Connecting field-level and n-point function measurements of galaxy bias
- The Impact of Cosmic Variance and Satellites on JWST Clustering Measurements at Redshift around 6
- Universality of Primordial Anisotropies in Gravitational Wave Background
- KiDS-1000 cosmic shear reanalysis using MetaCalibration
- 2Flow2SN: Dusty Supernova Cosmology with Photometric Redshifts and Contamination using Neural Simulation-Based Inference