pith. sign in

arxiv: 2308.10926 · v2 · submitted 2023-08-21 · 🌌 astro-ph.CO

Symfind: Addressing the Fragility of Subhalo Finders and Revealing the Durability of Subhalos

Pith reviewed 2026-05-24 06:55 UTC · model grok-4.3

classification 🌌 astro-ph.CO
keywords subhalo finderparticle trackingdark matter simulationssubhalo survivalsatellite galaxiescosmological simulationshalo findingsubhalo mass function
0
0 comments X

The pith

A new particle-tracking subhalo finder tracks objects to orders-of-magnitude lower masses than common tools detect.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper introduces Symfind to locate and follow subhalos in dark-matter simulations more reliably at low particle counts. It establishes that this approach finds 15 to 40 percent more subhalos inside the virial radius and 35 to 120 percent more inside one-quarter of that radius at any fixed peak mass. A sympathetic reader would care because subhalo counts directly shape predictions for how many satellite galaxies should exist around larger systems. The work further shows that raising simulation resolution uncovers still more subhalos and that existing methods settle on incorrect answers for the mass function, radial distribution, and disruption masses. Symfind is presented as a way to follow resolved subhalos all the way to the typical point of galaxy disruption.

Core claim

Symfind is a particle-tracking subhalo finder that can track subhalos to orders-of-magnitude lower masses than commonly used halo-finding tools. In the Symphony dark-matter-only simulation suite this produces approximately 15-40 percent more subhalos within R_vir and 35-120 percent more within R_vir/4 at fixed peak subhalo mass. Mass loss itself becomes resolvable at modest particle counts while maximum circular velocity requires much higher resolution. The method traces resolved subhalos until the point of typical galaxy disruption without invoking orphan modeling, and the paper shows that commonly used tools converge to false solutions for the mass function, radial distribution, and subhal

What carries the argument

Symfind, a particle-tracking-based subhalo finder that follows particles to identify bound structures at low particle counts.

If this is right

  • At fixed peak mass the number of subhalos rises with simulation resolution.
  • Mass loss can be resolved at particle counts around 4000 while v_max requires counts around 30000.
  • Subhalos can be followed to the typical galaxy disruption point without separate orphan modeling.
  • Commonly used finders produce resolution-dependent errors in the subhalo mass function and radial distribution.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Adopting this approach would raise the expected number of satellite galaxies around Milky Way-mass hosts.
  • The concrete steps given for testing other finders could become a standard check for reliability at low masses.
  • Disruption seen in current simulations may partly reflect finder limits rather than purely physical processes.

Load-bearing premise

The particle-tracking implementation correctly identifies physically bound subhalos at low particle counts without introducing spurious detections or missing real disruption events.

What would settle it

A side-by-side run of Symfind and a standard finder on a simulation at substantially higher resolution that shows whether the extra low-mass subhalos persist or disappear.

Figures

Figures reproduced from arXiv: 2308.10926 by Elise Darragh-Ford, Ethan O. Nadler, Philip Mansfield, Risa H. Wechsler, Yunchong Wang.

Figure 1
Figure 1. Figure 1: Cartoon illustrating the major steps in our subhalo-finding method, Symfind. Panel I: First, we annotate an input merger tree (for this paper, input catalogs are generated by Rockstar), identifying and correcting various errors (Appendix A.1). Here, the red X’s indicate portions of the input merger tree that would be removed or corrected. Panel II: Next, using this annotated tree, we find the first halo br… view at source ↗
Figure 3
Figure 3. Figure 3: Projected dark matter density of the subhalo shown in [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗
Figure 2
Figure 2. Figure 2: The evolution of a representative subhalo over time, as measured by both Rockstar (red) and Symfind (blue). In the top panel, the Rockstar curve is dashed during the period when it overlaps with the Symfnd curve. Snapshots during which Rockstar has identified an incorrect subhalo center are shown in orange (see Appendix C), and the virial mass/virial radius of the host halo is shown in black. When followed… view at source ↗
Figure 4
Figure 4. Figure 4: The amount of mass subhalos lose before subhalo finders lose track of them, measured across the entire Symphony suite. Left: The probability that subhalos with 104.5 < 𝑛peak < 105 disrupt before reaching a given mass loss ratio, 𝜇 ≡ 𝑚/𝑚peak, when subhalos are followed by Rockstar (red) and by Symfind (blue). Survivor bias is accounted for using the Kaplan–Meier estimator, and 1-𝜎 confidence intervals (shad… view at source ↗
Figure 5
Figure 5. Figure 5: The convergence limits of subhalos in Symfind. Left: Comparison between several idealized convergence limits and the mass-loss￾history of a fairly well-resolved 𝑛peak ≈ 104 subhalo. Lines compare the force-softening limit (Eq. 6; orange), the discreteness limit (Eq. 7; blue), and the two-body relaxation limit (Eq. 9; red). Each limit predicts that the subhalo is only converged when the black curve is above… view at source ↗
Figure 6
Figure 6. Figure 6: The median relationship between 𝑣max/𝑣max,infall and 𝑚/𝑚infall for subhalos, stacked across the entire Symphony suite, as measured by Symfind. The red curve shows the predictions of high￾resolution idealized simulations (Green & van den Bosch 2019), performed over a range of orbital parameters and matched to the infall concentration distribution of our subhalo sample. The blue curves show the relation when… view at source ↗
Figure 7
Figure 7. Figure 7: A toy model for the evolution of a toy mass, 𝑚˜ , as a function of toy time, 𝑡˜, which illustrates the impact of survivor bias. A population of 103 mock subhalos was generated with random exponential timescales and random mass scales at which they are lost by the subhalo finder. 30 random subhalos are shown in black. The true median of the population is shown in red. In orange and blue, we show two flawed … view at source ↗
Figure 8
Figure 8. Figure 8: Subhalo mass loss rates compared against idealized numerical limits. Each panel shows a set of Symfind subhalos from the SymphonyMilkyWayHR suite (red) and their mass-matched subhalos from the fiducial-resolution SymphonyMilkyWay (blue), grouped by 𝑛peak of the fiducial-resolution subhalos. The Kaplan-Meier estimator has been used to correct for survivor bias and to estimate 1 𝜎 uncertainties (shaded bands… view at source ↗
Figure 9
Figure 9. Figure 9: The 𝑧 = 0 subhalo 𝑚peak functions of Milky Way-mass halos, normalized by hosts masses, 𝑀vir. Left: Comparison between the subhalo mass functions measured by Rockstar (dashed) and Symfind (solid) for the fiducial-resolution SymphonyMilkyWay suite. Mass functions within different radii are shown as different colors, and the bottom panel shows the ratio between Symfind mass functions and Rockstar mass functio… view at source ↗
Figure 10
Figure 10. Figure 10: Radial distribution of subhalos in the SymphonyMilky￾Way suite. We compare the radial distribution of dark matter parti￾cles (dotted) against Symfind (solid) and Rockstar (dashed) . The bottom panel shows the ratios of these curves to a fit to the dashed blue curve, CDFRCT (Eq. 15). Rockstar subhalos show no appre￾ciable dependence on resolution. This is caused by the effect shown in the right panel of [… view at source ↗
Figure 11
Figure 11. Figure 11: Comparison between subhalo finder disruption limits and other limiting factors for subhalo/satellite galaxy analysis. The two colored curves show 𝜇90, the 𝑚/𝑚peak ratio at which Symfind can still resolve 90% of the subhalo population (reproduced from [PITH_FULL_IMAGE:figures/full_fig_p020_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: The fraction of disrupted Rockstar subhalos that experienced an error where they were associated with an incorrect density peak. All the Symphony suites have been combined to improve number statistics. The left panel shows the error rate at 𝑧 = 0 as a function of radius among surviving subhalos, and the right panel shows the error rate among disrupted subhalos as a function of the final radius of the subh… view at source ↗
Figure 13
Figure 13. Figure 13: The fraction of Rockstar branches which started out as subhalos. Symphony’s snapshot cadence is fine enough that these branches are likely numerical artifacts: either subhalos that are flitting in and out of existence, or subhalos that Rockstar had previously lost track of and was unable to match with its true pro￾genitors. The red curve shows all the branches that have ever been subhalos of one of Sympho… view at source ↗
Figure 14
Figure 14. Figure 14: A comparison of Rockstar and Symfind error rates as a function of our algorithm’s two free parameters, 𝑘 and 𝑁core. The left panel shows low-resolution subhalos and the right panel shows moderate-resolution subhalos. The 𝑦-axis shows 𝑓err, the fraction of halos misidentified or missed entirely by either catalog. The best performance is achieved when the error rate in the core catalogs (dashed lines) is sm… view at source ↗
Figure 15
Figure 15. Figure 15: The relationship between subhalo masses and 𝑚bound, the bound mass calculated from iteratively unbinding the tracked particles. The black curve compares Rockstar mass to 𝑚bound, the two red curves compare a limited number of iterative unbindings on tracked particles, and the blue curve compares against full itera￾tive unbinding where the potential is calculated assuming spherical symmetry. pool of particl… view at source ↗
Figure 16
Figure 16. Figure 16: The evolution of eight randomly selected subhalos from the first host halo, Halo023, in SymphonyMilkyWay, using the same representation as [PITH_FULL_IMAGE:figures/full_fig_p041_16.png] view at source ↗
Figure 17
Figure 17. Figure 17: Survival curves for several 𝑛peak bins using a variety of particle masses (see [PITH_FULL_IMAGE:figures/full_fig_p042_17.png] view at source ↗
Figure 18
Figure 18. Figure 18: The 𝑧 = 0 SHMF for both fiducial-resolution and high-resolution simulations. The left panel shows Symfind and the right panel shows Rockstar. The format of these figures is the same as [PITH_FULL_IMAGE:figures/full_fig_p043_18.png] view at source ↗
Figure 19
Figure 19. Figure 19: Comparison between the radial distribution of subhalos as measured with Subfind and particle-tracking. This Figure is the same as [PITH_FULL_IMAGE:figures/full_fig_p044_19.png] view at source ↗
read the original abstract

A major question in $\Lambda$CDM is what this theory actually predicts for the properties of subhalo populations. Subhalos are difficult to simulate and to find within simulations, and this propagates into uncertainty in theoretical predictions for satellite galaxies. We present Symfind, a new particle-tracking-based subhalo finder, and demonstrate that it can track subhalos to orders-of-magnitude lower masses than commonly used halo-finding tools, with a focus on Rockstar and consistent-trees. These longer survival mean that at a fixed peak subhalo mass, we find $\approx 15\%{-}40\%$ more subhalos within the virial radius, $R_\textrm{vir}$, and $\approx 35\%-120\%$ more subhalos within $R_\textrm{vir}/4$ in the Symphony dark-matter-only simulation suite. More subhalos are found as resolution is increased. We perform extensive numerical testing. In agreement with idealized simulations, we show that the $v_{\rm max}$ of subhalos is only resolved at high resolutions ($n_\textrm{peak}\gtrsim3\times 10^4$), but that mass loss itself can be resolved at much more modest particle counts ($n_\textrm{peak}\gtrsim4\times 10^3$). We show that Rockstar converges to false solutions for the mass function, radial distribution, and disruption masses of subhalos. We argue that our new method can trace resolved subhalos until the point of typical galaxy disruption without invoking ``orphan'' modeling. We outline a concrete set of steps for determining whether other subhalo finders meet the same criteria. We publicly release Symfind catalogs and particle data for the Symphony simulation suite at \url{web.stanford.edu/group/gfc/gfcsims/}.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

1 major / 0 minor

Summary. The manuscript introduces Symfind, a new particle-tracking-based subhalo finder, and claims it tracks subhalos to orders-of-magnitude lower masses than Rockstar and consistent-trees. In the Symphony dark-matter-only suite this yields ≈15%–40% more subhalos within R_vir and ≈35%–120% more within R_vir/4 at fixed peak mass. The paper asserts that Rockstar converges to false solutions for the subhalo mass function, radial distribution, and disruption masses, supported by extensive numerical testing and agreement with idealized simulations on v_max and mass-loss resolution thresholds (n_peak ≳ 3×10^4 for v_max, ≳4×10^3 for mass loss). It argues that resolved subhalos can be traced until typical galaxy disruption without orphan modeling, outlines steps to test other finders, and publicly releases Symfind catalogs and particle data.

Significance. If the central claim holds, the result would imply substantially more durable subhalo populations than inferred from standard finders, directly affecting theoretical predictions for satellite galaxies in ΛCDM and reducing reliance on orphan modeling. The public release of catalogs and particle data for the Symphony suite is a clear strength that enables community verification. The reported agreement with idealized simulations on resolution thresholds for v_max versus mass loss is also a constructive element. Significance is contingent on demonstrating that Symfind’s tracking criteria do not introduce artifacts at low particle counts.

major comments (1)
  1. [Methods] Methods section: The precise binding-energy cut, particle-assignment rule, and tracking continuity criterion used by Symfind are not explicitly stated. This is load-bearing for the central claim because the reported 15–40% (and 35–120%) excess subhalo counts at fixed peak mass, as well as the assertion that Rockstar converges to false solutions, rest on the assumption that Symfind correctly identifies physically bound objects down to n_peak ≳ 4×10^3 without spurious detections or missed disruptions. Without these explicit rules, the numerical testing cannot be independently reproduced or used to falsify the Rockstar results.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for their thoughtful review and for identifying a key issue regarding the reproducibility of Symfind. We agree that the precise algorithmic criteria must be stated explicitly in the manuscript and will revise accordingly.

read point-by-point responses
  1. Referee: [Methods] Methods section: The precise binding-energy cut, particle-assignment rule, and tracking continuity criterion used by Symfind are not explicitly stated. This is load-bearing for the central claim because the reported 15–40% (and 35–120%) excess subhalo counts at fixed peak mass, as well as the assertion that Rockstar converges to false solutions, rest on the assumption that Symfind correctly identifies physically bound objects down to n_peak ≳ 4×10^3 without spurious detections or missed disruptions. Without these explicit rules, the numerical testing cannot be independently reproduced or used to falsify the Rockstar results.

    Authors: We agree that the specific numerical criteria employed by Symfind are essential for reproducibility and for validating the central claims. In the revised manuscript we will add a dedicated subsection to the Methods section that explicitly states (i) the binding-energy cut, including the precise threshold and procedure for removing unbound particles, (ii) the particle-assignment rule, including how particles are initially assigned and reassigned during tracking, and (iii) the tracking continuity criterion, including the conditions used to maintain subhalo identity across snapshots. These parameters are already implemented in the publicly released Symfind code and particle data, but we acknowledge they were not documented with sufficient precision in the original text. The added description will enable independent reproduction of the reported subhalo counts and the numerical tests that compare Symfind to Rockstar. revision: yes

Circularity Check

0 steps flagged

No significant circularity; Symfind claims rest on independent application to simulation data

full rationale

The paper introduces Symfind as a new particle-tracking algorithm and reports its outputs (higher subhalo counts at fixed peak mass, Rockstar converging to false solutions) by direct comparison against Rockstar/consistent-trees on the Symphony suite. No step reduces a claimed prediction to a fitted parameter from the same data, redefines a quantity in terms of itself, or relies on a load-bearing self-citation whose content is unverified outside the present work. The numerical testing is presented as external validation rather than a definitional loop. The derivation chain is therefore self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract-only review; no explicit free parameters, axioms, or invented entities are stated. The new finder is the primary addition, with implementation thresholds likely present but unspecified here.

pith-pipeline@v0.9.0 · 5891 in / 1163 out tokens · 27823 ms · 2026-05-24T06:55:37.954140+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Bulgeless Evolution And the Rise of Discs (BEARD) III. A numerical simulation view of satellites around Milky-Way analogues

    astro-ph.GA 2026-05 unverdicted novelty 5.0

    Simulation comparison finds bulgeless galaxies host more centrally concentrated, disc-aligned satellites with steeper faint-end luminosity functions than bulge-dominated controls, reflecting co-evolution and quieter m...

Reference graph

Works this paper leans on

15 extracted references · 15 canonical work pages · cited by 1 Pith paper

  1. [1]

    We perform post-processingtoremoveerrorsinthetree(Appendix A.1)

    We use an existing halo finder and merger tree code to track halos before they become subhalos. We perform post-processingtoremoveerrorsinthetree(Appendix A.1)

  2. [2]

    Foreachsubhalo,wefindalltheparticlesthataccreted onto the subhalo prior to infall, as well as the most- bound particles at infall (Appendix A.2)

  3. [3]

    Weuseanexistinghalofindertoidentifydensitypeaks within the set of tracked particles (Appendix A.3)

  4. [4]

    (Appendix A.4)

    Weusethemost-boundparticlestoselectwhichdensity peak is the true center of the subhalo. (Appendix A.4)

  5. [5]

    candidate host branches

    Wecalculatehalopropertiesbasedonthiscenterusing all the tracked particles. (Appendix A.5) At the highest level of abstraction, this strategy is similar to the codes HBT/HBT+ (Han et al. 2012, 2018) andSparta (Diemeretal.2023),althoughindetailthemethodsarequite different. We discuss this in more detail in Section 7. A.1. Merger Tree Post-Processing Merger ...

  6. [6]

    If too large a halo boundary is chosen, particles that never orbited the subhalo can be counted in subhalo property calculations

  7. [7]

    If too large a halo boundary is chosen, some particles whichneverorbitedalargerhalocanbeexcludedfrom accretion onto subhalos prior to infall

  8. [8]

    sub-subhalos,

    If too small a halo boundary is chosen, some particles whicharetrulyorbitingthesubhalowillnotbetracked. Effect 1 is not very important as long as some reasonable definition of halo radius is chosen: this choice of radius cer- tainlychangestheamountofsplashbackmassinisolatedhalos (Diemer2021),butthismassislightlyboundandonly2%of the mass outside the instan...

  9. [9]

    Theradiusofthehosthaloistakentobetheoverdensity radius corresponding to 200 times the critical density of the universe, replacing𝑅vir with 𝑅200c

  10. [10]

    A different set of resolution bins are used

  11. [11]

    No cut is made to remove subhalos with high subhalo- to-host mass ratios at infall

  12. [12]

    ThesetofresolutiontestsfromHanetal.(2016)thatwe compare against are performed at a fixed mass across resimulations rather than across different mass bins within the same simulation

  13. [13]

    Profilesaremeasuredforasinglehosthalo,ratherthan averaged across a large population of host halos

  14. [14]

    Aq-A-1 is higher mass than any of the hosts in Sym- phonyMilkyWay

  15. [15]

    Withthesedifferencesinmind,weshowthedependenceof the radial distribution of satellites on resolution in Fig

    Thelocationsofthemost-boundparticlesatinfallwere not used to confirm that late-time subhalos were truly the descendants of the initial halos they are connected to. Withthesedifferencesinmind,weshowthedependenceof the radial distribution of satellites on resolution in Fig. 19. Subfind is shown as dashed lines, while the results for particle-tracking are sh...