Pith. sign in

REVIEW 3 major objections 4 minor 93 references

Vector Resonant Relaxation and Statistical Closure Theory. II. One-loop Closure

T0 review · 3 major / 4 minor · reviewed 2026-08-15 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read The one-loop MSR closure predicts isotropic VRR two- and three-point correlations, cutting the two-point error by 80 percent and matching N-body skewness within 4 to 20 percent.

desk verdict A concrete one-loop MSR closure for VRR that improves on bare DIA and produces nonzero skewness where bare order vanishes; the main caveat is the un-tested seed independence of the fixed point. read the letter →

arxiv 2509.22164 v2 pith:2JPP2GRU submitted 2025-09-26 cond-mat.stat-mech astro-ph.GA

Sofia Flores , Jean-Baptiste Fouvry This is my paper · ORCID
classification cond-mat.stat-mechastro-ph.GA
keywords vectorresonantrelaxationstatisticalclosuretheoryMSRformalismone-looprenormalisationDirectInteractionApproximationtwo-pointcorrelationthree-pointskewnessfixed-pointiteration
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper tries to establish that the MSR statistical closure scheme, taken one order beyond the bare Direct Interaction Approximation, accurately predicts correlation functions in a fully non-linear long-range system: vector resonant relaxation, the reorientation of stellar orbital planes around a supermassive black hole. The authors implement the one-loop MSR equations with a fixed-point iteration and compare the predicted two-point two-time correlation, renormalised three-point vertex, and three-point skewness against direct N-body simulations. Their central finding is that the one-loop prediction improves the two-point correlation by about 80 percent over the bare DIA, and produces non-zero skewness within 4 to 20 percent of simulations for triangles where the bare order predicts exactly zero. A sympathetic reader would care because this system has no linear term and no external driving, so it puts closure theory to the test in the strongly non-linear regime where standard perturbation theory fails.

What carries the argument

The load-bearing object is the renormalised three-point interaction vertex $\Gamma_{123}$, defined by $\Gamma_{123}=G^{-1}_{11'}G^{-1}_{22'}G^{-1}_{33'}G_{1'2'3'}$; it is the target of the closure and a proxy for the skewness. The closure is supplied by the one-loop relation $\Gamma_{123}=\gamma_{123}+(\delta\Sigma_{12}/\delta G_{45})G_{44'}G_{55'}\Gamma_{34'5'}$ together with the Dyson equation $G^{-1}=g^{-1}-\Sigma$, with self-energy $\Sigma_{12}=\tfrac12\gamma_{134}G_{33'}G_{44'}\Gamma_{23'4'}$. The numerical scheme iterates these coupled equations for the pair $[G,\Gamma]$ until convergence, exploiting isotropy, time stationarity, the fluctuation-dissipation theorem, and angular-momentum contraction rules to keep the cost manageable; the auxiliary tensor $\Lambda_{123}=\Gamma_{123'}G_{3'3}$ reduces the dominant one-loop evaluation from $O(N_{\rm STEPS}^8L_{\rm MAX}^6)$ to $O(N_{\rm STEPS}^5L_{\rm MAX}^6)$ operations.

What would settle it

Run the same one-loop fixed-point iteration starting from several distinct initial propagators—the bare propagator, the Gaussian propagator with coherence time $T_c$, and a perturbed Gaussian—and compare the converged $[G,\Gamma]$ and predicted $C_\ell$; if the final correlation functions differ by more than numerical tolerance, the claimed one-loop prediction is not unique. Alternatively, measure the renormalised vertex directly in N-body simulations by inverting the measured two- and three-point propagators and compare its structure to the predicted $\Gamma$.

Watch

Extended reading notes

Core claim

The paper's central claim is that the self-consistent one-loop MSR closure, in which the renormalised three-point vertex $\Gamma$ and the dressed two-point propagator $G$ are solved together, gives quantitatively accurate predictions for isotropic vector resonant relaxation. Concretely, for harmonic $\ell=4$ the maximum absolute error of the two-point correlation drops from about 0.15 at bare DIA order to about 0.03 at one-loop order, and the one-loop prediction recovers the late-time exponential decay and the sign change of the correlation. For triangles $(\ell_1,\ell_2,\ell_3)$ whose bare vertex, and hence bare skewness, vanishes, the one-loop renormalisation generates a non-zero skewness that matches N-body measurements within roughly 4 to 20 percent, with the $(1,3,3)$ triangle underestimated by about 20 percent. The same scheme shows that expanding the vertex in powers of the bare coupling diverges at late times, while the self-consistent closure in terms of $\Gamma$ converges.

Load-bearing premise

The fixed-point iteration is assumed to converge to a unique physical solution, but the paper only demonstrates convergence from one Gaussian-seeded initial condition; if multiple fixed points exist, the one-loop prediction depends on the starting guess.

Editorial extensions

If this is right

  • The one-loop MSR closure reduces the maximum absolute error of the two-point correlation for $\ell=4$ from about 0.15 to about 0.03, an 80 percent improvement over the bare DIA.
  • For odd triangles whose bare skewness vanishes, the one-loop scheme predicts a non-zero skewness that agrees with N-body measurements within 4 to 20 percent, so the closure captures weak non-Gaussianities generated purely by renormalisation.
  • The self-consistent one-loop vertex expansion converges at late times, while expanding the same diagram in powers of the bare vertex diverges; convergence requires dressing the vertex in terms of itself.
  • The renormalised vertex stays close to the bare vertex, with dressing amplitudes of order $10^{-4}$ relative to the bare maximum, indicating that the closure corrections are small perturbative improvements rather than large resummations.
  • The same fixed-point scheme formally extends to two-loop order, but the estimated cost, about $10^{10}$ hours on 128 cores for the chosen parameters, makes two-loop predictions impractical without new numerical methods.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Because the governing equation $\partial_t\varphi=\tfrac12\gamma\varphi\varphi$ appears in many settings, including plasmas, decaying turbulence, and cosmological structure formation, the validated one-loop vertex closure suggests the same self-consistent treatment could improve correlation predictions wherever no linear timescale exists.
  • A direct test of internal consistency would be to measure the dressed vertex $\Gamma$ in simulations via its definition in terms of the two- and three-point propagators; the paper notes this is difficult, but such a measurement would distinguish genuine one-loop dressing from an effective fit.
  • The documented sensitivity to the initial propagator seed implies a uniqueness test: exploring additional seeds, including a delayed Gaussian or a vertex initialised away from the bare value, could expose fixed-point branching not visible in the ITER=10 runs reported here.
  • The cost scaling suggests that solving the MSR equations in Fourier time, or replacing the vertex expansion with a regulator-based flow, are natural next steps; if either works, two-loop predictions could become feasible and would test whether the residual 20 percent underestimate in the $(1,3,3)$ triangle shrinks.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 4 minor

Summary. This paper develops a numerical implementation of the one-loop Martin-Siggia-Rose (MSR) closure for the isotropic Vector Resonant Relaxation (VRR) model, a purely quadratically nonlinear long-range interacting system with no linear term. The authors solve the coupled Dyson equation (Eq. 22) and the self-consistent vertex equation (Eq. 24) by fixed-point iteration, exploiting isotropy, time stationarity, the fluctuation-dissipation theorem, and angular-momentum contraction rules to manage the numerical cost. They compare the resulting two-point correlation function C_ell (Fig. 2) and three-point skewness S^L (Figs. 4-5) with direct N-body simulations, reporting 4-20% agreement. The one-loop closure is shown to improve upon the bare DIA prediction, in particular by producing nonzero skewness for odd triangles where the bare vertex exactly vanishes. The authors also show that a naive one-loop expansion in powers of the bare vertex diverges (Fig. 9), whereas the self-consistent one-loop closure is stable. Numerical convergence with respect to ITER, LMAX, DT, and TMAX is documented in Appendix F, and the code is publicly available.

Significance. If the central claim holds, this is a substantial methodological advance: it provides a parameter-free, self-consistent one-loop MSR calculation in a fully nonlinear, non-perturbative setting, with falsifiable predictions for both two- and three-point statistics. The manuscript is careful in several respects: it documents convergence with respect to all numerical truncation parameters (Appendix F), reproduces the bare and naive benchmarks from the companion paper, uses N-body measurements with explicit bootstrap error bars, and makes the code publicly available. The reported improvement over bare DIA is concrete and quantified, and the demonstration that a non-self-consistent naive expansion diverges is an important sanity check. The main concerns are not with the derivation but with the uniqueness of the fixed point and with the strength of the validation claims: the quantitative results rest on a single fixed-point seed, and the abstract overstates the validation of the renormalised vertex, which is not directly measured. These issues are addressable and do not, on the present evidence, invalidate the derivation.

major comments (3)
  1. [Section IV A and Section V] The one-loop predictions are obtained from a fixed-point iteration whose uniqueness is explicitly not guaranteed: Section IV A states that "a unique, well-behaved fixed-point solution for [G,Gamma] in the MSR scheme is not guaranteed," and the scheme is initialized with G(0)=G_G, a Gaussian propagator built from the coherence time T_c. The paper reports that starting from the bare propagator g gives "slow and marginally stable convergence," but it does not test whether the final [G,Gamma] is independent of the seed. Since every quantitative claim in Section V (Figs. 2, 4, and 5) is generated from this single seed, the excellent agreement with N-body data could in principle reflect a branch selected by the Gaussian initial condition rather than the intrinsic content of the one-loop closure. Please add a seed-independence study: for example, continue iterations from the bare-propagator seed to large ITER, and also run a perturbed Gaussian seed, then compare the resulting C_ell and S^L within the quoted few-percent accuracy. If such a test is computationally infeasible, the conclusions and abstract must be softened to state that the predictions are conditional on the chosen fixed-point branch.
  2. [Abstract and Section V B] The abstract lists among the validated predictions "(ii) the renormalised three-point interaction vertex," but Section V B explicitly states that Gamma is challenging to measure directly in N-body simulations and that doing so "will be the topic of a future work." The only quantitative N-body comparisons involving Gamma are made indirectly through the skewness S^L in Section V C, which is a convolution of G and Gamma (Eq. E4), not a direct measurement of the vertex itself. Thus the renormalised vertex is predicted and examined structurally (Fig. 3), but it is not quantitatively validated against N-body data. Please rephrase the abstract and any summary statements to say that the formalism predicts Gamma and that the skewness comparisons provide an indirect consistency test, or add a direct measurement of Gamma.
  3. [Appendix F and Section IV C] The convergence tests in Appendix F (Figs. 10-13) monitor the two-point correlation function C_ell only; no equivalent convergence diagnostic is reported for the renormalised vertex Gamma^L or for the residual of the vertex equation (Eq. 26a). Since Gamma^L is the central new object and the one-loop diagram is the computational bottleneck, it would be valuable to report the relative change in Gamma^L between successive iterations (or the norm of the residual of Eq. 26a) alongside the C_ell convergence. This would strengthen the claim that the fixed point is actually resolved for the vertex, not only for the two-point function.
minor comments (4)
  1. [Section II] There is a typo in the phrase "methods from stastical closure theory"; it should read "statistical closure theory."
  2. [Section V A] The sentence "the one-loop prediction correctly recovers the oscillation of the correlation function that predicts a negative correlation for t/T_c by 0.6" is awkward; consider rephrasing as "correctly recovers the oscillation, predicting a negative correlation for t/T_c greater than about 0.6."
  3. [Figure 2] The bottom panel of Fig. 2 shows only the one-loop absolute error; including the bare-order error in the same panel would make the claimed 80% improvement directly visible to the reader.
  4. [Equation (48)] The norm introduced in Eq. (48) is not named; please state explicitly that it is the L2 norm over the square domain [0,T_c]^2.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the one-loop MSR predictions are computed from the VRR Hamiltonian and the MSR equations with no fitted parameters; the Gaussian-seed caveat is a robustness limitation, not a circular reduction.

full rationale

The one-loop MSR predictions in this paper are self-contained: the only inputs are the VRR Hamiltonian (coupling J_l, initial amplitude C_0 = N/4pi, and coherence time T_c from Eq. D5) and the MSR equations truncated at one loop (Eqs. 22 and 24). No parameter is fitted to the N-body correlation data; the N-body measurements serve purely as external benchmarks in Figs. 2, 4, and 5. The bare-order and naive-one-loop results of FF25 are reproduced as numerical sanity checks (Appendix F, Fig. 14), so those self-citations are benchmarks rather than load-bearing derivations. The paper explicitly flags the main caveat in Section IV A: 'At this stage, a unique, well-behaved fixed-point solution for [G,Gamma] in the MSR scheme is not guaranteed,' and the Gaussian seed G_G built from T_c is not tested for seed independence. This is a genuine robustness and possible non-uniqueness limitation, but it is not circularity: the final one-loop propagator deviates substantially from the Gaussian seed (e.g., the negative l=4 correlation at t/T_c > 0.6 in Fig. 2), so the output is not equal to its input by construction. The FDT relation C_l = C_0 R_l(|t|) is imposed from equilibrium statistical mechanics, not fitted. No enumerated circularity pattern — self-definition, fitted-input-called-prediction, load-bearing self-citation, uniqueness imported from authors, ansatz smuggled in via citation, or renaming a known result — is exhibited with a concrete reduction of the paper's own equations.

Assumptions & free parameters 4 free parameters · 7 assumptions · 0 invented entities

The MSR closure rests on standard functional-integral identities; the physical model adds equilibrium (FDT), isotropy, single-population ℓ=2 assumptions. The numerical solution relies on truncation parameters LMAX, DT, TMAX, and ITER whose convergence is tested but not proven asymptotically. No new particles or forces are introduced.

free parameters (4)
  • LMAX = 7
    Maximum harmonic number for truncation of the ℓ sums; convergence checked in Appendix F, within 2 percent for ℓ=2.
  • DT = Tc/80
    Timestep for discretization of time integrals; convergence better than 1 percent for Tc/DT between 48 and 80.
  • TMAX = Tc
    Maximum temporal depth stored for the response function; test with TMAX=2Tc shows no boundary effects.
  • ITER = 10
    Number of fixed-point iterations; convergence within 1 percent from iteration 5 to 10.
assumptions (7)
  • standard math Exactness of the MSR Dyson and vertex equations, Eqs. (17) and (18), as derived in Martin, Siggia, and Rose (1973).
    The central closure scheme relies on the functional-integral identities of the MSR formalism; the derivation is cited, not re-derived.
  • domain assumption Fluctuation-dissipation theorem holds exactly for isotropic VRR: C_ℓ(t) = C_0 R_ℓ(|t|).
    Invoked in Section IV B 3 to reconstruct the full propagator from the response function; the system is assumed to be at thermodynamic equilibrium.
  • standard math Isotropy and time stationarity are preserved under the one-loop renormalization.
    This follows from the contraction rules of the Elsasser coefficients (Appendix B), using the 6j-symbol identity of Eq. (B8).
  • domain assumption The VRR interaction is dominated by the ℓ=2 quadrupole term, so J_ℓ vanishes for ℓ ≠ 2.
    Section II states this restriction is physically motivated by the dominance of the large-scale quadrupolar component.
  • domain assumption Single-population system: all stars have the same semi-major axis and eccentricity.
    Section II introduces this simplification, noting that multi-population cases can be included in the formalism but are not treated.
  • domain assumption Initial conditions are quasi-Gaussian for large N; the initial skewness scales as N^{-1/2} and is negligible for odd triangles at N=10^4.
    Section VI A shows the N-dependence and uses N=10^4 to ensure convergence; the MSR prediction assumes exactly Gaussian initial statistics for the retained terms.
  • ad hoc to paper Numerical truncations: R_ℓ(t)=0 for t>TMAX and ℓ>LMAX, and Γ is set to zero for time separations exceeding TMAX.
    Needed for finite computations; convergence tests in Appendix F show effects within a few percent but do not establish the continuum limit rigorously.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Vector Resonant Relaxation and Statistical Closure Theory. II. One-loop Closure." pith.science (2026). https://pith.science/paper/2JPP2GRU

@misc{pith2026250922164,
  author       = {Pith},
  title        = {Pith review of: Vector Resonant Relaxation and Statistical Closure Theory. II. One-loop Closure},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/2JPP2GRU}},
  note         = {Machine review of arXiv:2509.22164}
}
read the original abstract

We use stellar dynamics as a testbed for statistical closure theory. We focus on the process of "Vector Resonant Relaxation," a long-range, non-linear, and correlated relaxation mechanism that drives the reorientation of stellar orbital planes around a supermassive black hole. This process provides a natural setting to evaluate the predictive power of generic statistical closure schemes for dynamical correlation functions, in the fully non-linear and non-perturbative regime. We develop a numerical scheme that explicitly implements the seminal "Martin-Siggia-Rose" formalism at one-loop order via an iterative fixed-point approach, thereby improving upon the bare order from the "Direct Interaction Approximation." Using this framework, we quantitatively validate the ability of the formalism to predict (i) the two-point two-time correlation function; (ii) the renormalised three-point interaction vertex; (iii) the three-point three-time correlation function. These predictions are compared to direct measurements from numerical simulations. We conclude by discussing the limitations of this approach and presenting possible future venues.

Figures

Figures reproduced from arXiv: 2509.22164 by the authors.

Figure 1
Figure 1. FIG. 1: Illustration of the process of VRR. Each point on [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. FIG. 2: Top panel: Normalised isotropic two-point correla [PITH_FULL_IMAGE:figures/full_fig_p009_2.png] view at source ↗
Figure 3
Figure 3. FIG. 3: Top (resp. Bottom) panels: Dependence of [PITH_FULL_IMAGE:figures/full_fig_p010_3.png] view at source ↗
Figures from the paper (11 more)
Figure 4
Figure 4. Figure 4: FIG. 4: Dependence of the three-point isotropic skewness, [PITH_FULL_IMAGE:figures/full_fig_p011_4.png]
Figure 5
Figure 5. Figure 5: FIG. 5: Slice of the isotropic skewness from Fig. [PITH_FULL_IMAGE:figures/full_fig_p012_5.png]
Figure 6
Figure 6. Figure 6: FIG. 6: Profile of the three-point isotropic skewness [PITH_FULL_IMAGE:figures/full_fig_p013_6.png]
Figure 7
Figure 7. Figure 7: FIG. 7: Profile of the three-point isotropic skewness [PITH_FULL_IMAGE:figures/full_fig_p014_7.png]
Figure 8
Figure 8. Figure 8: FIG. 8 [PITH_FULL_IMAGE:figures/full_fig_p014_8.png]
Figure 9
Figure 9. Figure 9: FIG. 9: Same as Fig [PITH_FULL_IMAGE:figures/full_fig_p015_9.png]
Figure 12
Figure 12. Figure 12: FIG. 12: Convergence with respect to the timestep [PITH_FULL_IMAGE:figures/full_fig_p023_12.png]
Figure 10
Figure 10. Figure 10: FIG. 10: Convergence with respect to the number of itera [PITH_FULL_IMAGE:figures/full_fig_p023_10.png]
Figure 11
Figure 11. Figure 11: FIG. 11: Convergence with respect to the maximum har [PITH_FULL_IMAGE:figures/full_fig_p023_11.png]
Figure 14
Figure 14. Figure 14: FIG. 14: Same as Fig [PITH_FULL_IMAGE:figures/full_fig_p024_14.png]
Figure 15
Figure 15. Figure 15: FIG. 15: Relative error in the total angular momentum, [PITH_FULL_IMAGE:figures/full_fig_p024_15.png]

Review history (2 revisions) →

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

93 extracted references · 76 canonical work pages

  1. [1]

    (26a), the most challenging part is to compute its one-loop di- agram

    Updating the dressed vertex In order to compute the right-hand side of Eq. (26a), the most challenging part is to compute its one-loop di- agram. Indeed, it must be evaluated for all possible val- ues of the external coordinates, orlegs, i.e. all values of 1=(ϵ,t,ℓ,m), while summing over all the internal ones. Owing to the symmetries from Section IV B, co...

  2. [2]

    Updating the inverse propagator We can now turn our interest to updating the in- verse two-point propagator, as given by Eq. (26b). This requires the computation of the loop diagram associ- ated with the self-energy (Eq. 20). Leveraging once again isotropy (Section IV B 1) and time-stationarity (Section IV B 2), computing directly the self-energy in- volv...

  3. [3]

    bare, (a)

    Inverting the propagator We recall that the scheme of Eq. (26), together with the isotropy property from Section IV B 1, provides us with the isotropic renormalised vertex Γ L, and thein- verseisotropic two-point propagator, [G L]−1. However, it still needs to be inverted to obtainGL. As mentioned in Section IV B 3, the isotropic two-point propagator,G L,...

  4. [4]

    Definition In Eq

    Bare vertex a. Definition In Eq. (8), we introduced the coupling coefficientγ 123, which encapsulates all the information about the VRR interaction. Following eq. (A14) in FF25, it reads γ123 =γ abcδD(ta,tb)δ D(ta,tc),(A1) with γabc =E abc Jℓc−Jℓb ,(A2) where 1=(t,a) anda=(ℓ,m). In Eq. (A2), the Elsasser coefficientsE abc are defined in Eq. (B1). The coup...

  5. [5]

    Isotropy From Eq

    Bare propagator a. Isotropy From Eq. (16) and the identityg −1 12g23 =δ 13, we find that the bare propagator can be recast as g12 =δ a2 a1gL ℓ1(ξ1,ξ 2),(A9) whereξ=(ϵ,t). In particular, its non-zero components are gL ℓ1[(+,t 1),(−,t 2)] = Θ(t1−t2),(A10a) gL ℓ1[(−,t 1),(+,t 2)] = Θ(t2−t1),(A10b) where Θ is the usual Heaviside function, with the con- ventio...

  6. [6]

    (A2) are in- dependent of the considered orbits

    Definition The Elsasser coefficients appearing in Eq. (A2) are in- dependent of the considered orbits. They stem directly from the geometry of the problem, namely from the unit sphere. Following the same convention as in [51], they are defined as Eabc =E L abcEM abc,(B1) whereE L abc is independent of the indices (m a,mb,mc). In order to emphasise the sym...

  7. [7]

    B5), sat- isfy a few important contraction rules

    Contraction rules The excluded Elsasser coefficients,E ∆M abc (Eq. B5), sat- isfy a few important contraction rules. These play a cru- cial role in ensuring the isotropic symmetries of (G,Γ), as introduced in Section IV B 1. Following [76], for the diagram of the self-energy in Eq. (26b), we use the relation X m2,m3 E∆M a1a2a3E∆M a1′a2a3 =δ a1′ a1 δ∆ ℓ1ℓ2...

  8. [8]

    Isotropy We are interested in VRR in the isotropic limit

    Symmetries a. Isotropy We are interested in VRR in the isotropic limit. Within the iterative scheme of Eq. (26), the isotropy conditions from Eqs. (A5) and (A9) are consistently transmitted to the dressed vertex Γ and the two-point propagatorG, as shown in Eqs. (28) and (27). Moreover, from Eq. (15), this property is also satisfied by the self-energy Σ. T...

Show all 93 references
  1. [9]

    Updating the dressed vertex The numerical cost of evaluating all the possible one- loop diagrams from Eq

    Iteration a. Updating the dressed vertex The numerical cost of evaluating all the possible one- loop diagrams from Eq. (26a) is significantly reduced by symmetries. Owing to isotropy (Section IV B 1 and C 1 a), the sums overmare immediately carried out. Owing to time stationar...

  2. [10]

    Following Eqs

    Two-point function In this Appendix, we compute the ampli- tude of the initial two-point correlation function, Cab(ta =0,t b =0)=C (0) ab , witha=(ℓ,m) and assuming ℓa,ℓb>0. Following Eqs. (6) and (7), the stochastic field,φ a(t), is given instantaneously by φa(t) = NX i=1 Ya[...

  3. [11]

    (19) of [51], in the single-population ℓ=2 interaction model of VRR, the coherence time, i.e

    Coherence time According to eq. (19) of [51], in the single-population ℓ=2 interaction model of VRR, the coherence time, i.e. the characteristic timescale for the decay of correlations, is given by Tc = 4 √ 5√ 3 s a3⋆ GM• M• M⋆ √ N,(D5) wherea ⋆ is the semi-major axis of a sta...

  4. [12]

    (D2), assumingℓ a,ℓb,ℓc>0, we can write the initial three-point correlation function as C(0) abc = NX i,j,k=1 Ya[bLi(0)]Yb[bLj(0)]Yc[bLk(0)] .(D7) The average appearing in Eq

    Three-point function Similarly to Eq. (D2), assumingℓ a,ℓb,ℓc>0, we can write the initial three-point correlation function as C(0) abc = NX i,j,k=1 Ya[bLi(0)]Yb[bLj(0)]Yc[bLk(0)] .(D7) The average appearing in Eq. (D7) is zero unlessi=j=k. Thus, we get C(0) abc =N Z dbL 4π Ya(...

  5. [13]

    We refer to appendix D therein for the specific numerical setup

    Numerical simulations Throughout this paper, we investigate the exact same N-body system as in FF25. We refer to appendix D therein for the specific numerical setup. Compared to FF25, we improved the integration scheme by using a fourth-order Munthe-Kaas scheme [78]. We refer ...

    2025

  6. [14]

    We refer to appendix D of FF25 for more detail on the measure- ment of the isotropic two-point correlation from Eq

    Ensemble average Once a set of simulations (realisations) is available, we must average over them appropriately to obtain the N-body measurements of the correlations. We refer to appendix D of FF25 for more detail on the measure- ment of the isotropic two-point correlation fro...

  7. [15]

    Figure 2

    Figure parameters In this section, we specify the values of the parameters used when performing theN-body measurements of the two- and three-point correlation functions. Figure 2. We usedN=1 000 andNSEED=10 4. Figure 4 and 5. We usedN=10 4 andNSEED=10 4 for triangles (a), (b) ...

  8. [16]

    Frisch,Turbulence: The Legacy of A

    U. Frisch,Turbulence: The Legacy of A. N. Kolmogorov (Cambridge Univ. Press, 1995)

    1995

  9. [17]

    Berera, M

    A. Berera, M. Salewski, and W. D. McComb, Phys. Rev. E87, 013007 (2013)

    2013

  10. [18]

    D. R. Nicholson,Introduction to Plasma Theory (Krieger, 1992)

    1992

  11. [19]

    J. A. Krommes, JPP81, 205810601 (2015)

    2015

  12. [20]

    Lynden-Bell, MNRAS136, 101 (1967)

    D. Lynden-Bell, MNRAS136, 101 (1967)

    1967

  13. [21]

    Binney and S

    J. Binney and S. Tremaine,Galactic Dynamics: Second Edition(Princeton Univ. Press, 2008)

    2008

  14. [22]

    Bernardeau, S

    F. Bernardeau, S. Colombi, E. Gazta˜ naga, and R. Scoc- cimarro, Phys. Rep.367, 1 (2002)

    2002

  15. [23]

    Bernardeau, N

    F. Bernardeau, N. Van de Rijt, and F. Vernizzi, Phys. Rev. D85, 063509 (2012)

    2012

  16. [24]

    Kocsis and S

    B. Kocsis and S. Tremaine, MNRAS448, 3265 (2015)

    2015

  17. [25]

    J. A. Krommes, Phys. Rep.360, 1 (2002)

    2002

  18. [26]

    P. H. Diamond, S.-I. Itoh, and K. Itoh,Modern Plasma Physics(Cambridge Univ. Press, 2010)

    2010

  19. [27]

    Nazarenko,Wave Turbulence(Springer, Berlin, Hei- delberg, 2011)

    S. Nazarenko,Wave Turbulence(Springer, Berlin, Hei- delberg, 2011)

    2011

  20. [28]

    A. A. Schekochihin,Lectures on Kinetic The- ory and Magnetohydrodynamics of Plasmas, https://www-thphys.physics.ox.ac.uk/people/ AlexanderSchekochihin/KT/2015/KTLectureNotes.pdf (2025)

    2025

  21. [29]

    Micha and I

    R. Micha and I. I. Tkachev, Phys. Rev. D70, 043538 (2004)

    2004

  22. [30]

    Yokoyama and M

    N. Yokoyama and M. Takaoka, Phys. Rev. E89, 012909 (2014)

    2014

  23. [31]

    Tarpin, L

    M. Tarpin, L. Canet, C. Pagani, and N. Wschebor, J. Phys. A52, 085501 (2019)

    2019

  24. [32]

    Zhou, Phys

    Y. Zhou, Phys. Rep.935, 1 (2021)

    2021

  25. [33]

    R. H. Kraichnan, Phys. Rev.109, 1407 (1958)

    1958

  26. [34]

    T. H. Dupree, Physics of Fluids15, 334 (1972)

    1972

  27. [35]

    Boutros-Ghali and T

    T. Boutros-Ghali and T. H. Dupree, Physics of Fluids 24, 1839 (1981)

    1981

  28. [36]

    Valageas, A&A421, 23 (2004)

    P. Valageas, A&A421, 23 (2004)

    2004

  29. [37]

    Bernardeau, M

    F. Bernardeau, M. Crocce, and R. Scoccimarro, Phys. Rev. D85, 123519 (2012)

    2012

  30. [38]

    A. N. Kolmogorov, Akademiia Nauk SSSR Doklady32, 16 (1941)

    1941

  31. [39]

    Baldauf, U

    T. Baldauf, U. Seljak, and L. Senatore, JCAP2011, 006 (2011)

    2011

  32. [40]

    R. H. Kraichnan, J. Fluid Mech.5, 497 (1959)

    1959

  33. [41]

    S. A. Orszag, J. Fluid. Mech.41, 363 (1970)

    1970

  34. [42]

    W. D. McComb,The physics of fluid turbulence(Claren- don Press, 1990)

    1990

  35. [43]

    W. D. McComb,Homogeneous, Isotropic Turbulence: Phenomenology, Renormalization and Statistical Clo- sures(Oxford Univ. Press, 2014)

    2014

  36. [44]

    Canet, arXiv 2509.01472 (2025)

    L. Canet, arXiv 2509.01472 (2025)

    2025

  37. [45]

    P. C. Martin, E. D. Siggia, and H. A. Rose, Phys. Rev. A8, 423 (1973)

    1973

  38. [46]

    Flores and J.-B

    S. Flores and J.-B. Fouvry, Phys. Rev. E111, 044111 (2025)

    2025

  39. [47]

    Gillessen et al., ApJ837, 30 (2017)

    S. Gillessen et al., ApJ837, 30 (2017)

    2017

  40. [48]

    Alexander, ARA&A55, 17 (2017)

    T. Alexander, ARA&A55, 17 (2017)

    2017

  41. [49]

    K. P. Rauch and S. Tremaine, New Astron.1, 149 (1996)

    1996

  42. [50]

    Kocsis and S

    B. Kocsis and S. Tremaine, MNRAS412, 187 (2011)

    2011

  43. [51]

    A. S. Hamers, B. Bar-Or, C. Petrovich, and F. Antonini, ApJ865, 2 (2018)

    2018

  44. [52]

    Roupas, B

    Z. Roupas, B. Kocsis, and S. Tremaine, ApJ842, 90 (2017)

    2017

  45. [53]

    Sz¨ olgy´ en and B

    ´A. Sz¨ olgy´ en and B. Kocsis, Phys. Rev. Lett.121, 101101 (2018)

    2018

  46. [54]

    Touma, S

    J. Touma, S. Tremaine, and M. Kazandjian, Phys. Rev. Lett.123, 021103 (2019)

    2019

  47. [55]

    Tremaine, MNRAS491, 1941 (2020)

    S. Tremaine, MNRAS491, 1941 (2020)

    2020

  48. [56]

    Tremaine, MNRAS493, 2632 (2020)

    S. Tremaine, MNRAS493, 2632 (2020)

    2020

  49. [57]

    Gruzinov, Y

    A. Gruzinov, Y. Levin, and J. Zhu, ApJ905, 11 (2020)

    2020

  50. [58]

    Magnan et al., MNRAS514, 3452 (2022)

    N. Magnan et al., MNRAS514, 3452 (2022)

    2022

  51. [59]

    M´ ath´ e,´A

    G. M´ ath´ e,´A. Sz¨ olgy´ en, and B. Kocsis, MNRAS520, 2204 (2023)

    2023

  52. [60]

    Y. B. Ginat, T. Panamarev, B. Kocsis, and H. B. Perets, MNRAS525, 4202 (2023)

    2023

  53. [61]

    Levin, ApJ975, 278 (2024)

    Y. Levin, ApJ975, 278 (2024)

    2024

  54. [62]

    Panamarev and B

    T. Panamarev and B. Kocsis, MNRAS517, 6205 (2022)

    2022

  55. [63]

    Panamarev, Y

    T. Panamarev, Y. B. Ginat, and B. Kocsis, arXiv 2507.10551 (2025)

    2025

  56. [64]

    Fouvry, M

    J.-B. Fouvry, M. J. Bustamante-Rosell, and A. Zimmer- man, MNRAS526, 1471 (2023)

    2023

  57. [65]

    Y. B. Ginat and B. Kocsis, arXiv 2502.08709 (2025)

    2025 arXiv

  58. [66]

    Fouvry, B

    J.-B. Fouvry, B. Bar-Or, and P.-H. Chavanis, ApJ883, 161 (2019)

    2019

  59. [67]

    M. L. Nastac, R. J. Ewart, J. Juno, M. Barnes, and A. A. Schekochihin, arXiv 2503.17278 (2025)

    2025 arXiv

  60. [68]

    R. H. Kraichnan, Phys. Rev.113, 1181 (1959)

    1959

  61. [69]

    M. P. Bellon and E. I. Russo, Lett. Math. Phys.111, 42 (2021)

    2021

  62. [70]

    M. E. Peskin and D. V. Schroeder,An Introduction to Quantum Field Theory(Addison-Wesley, 1995)

    1995

  63. [71]

    Lancaster and S

    T. Lancaster and S. Blundell,Quantum Field Theory for the Gifted Amateur(Oxford Univ. Press, 2014). [57]https://github.com/sfloresmo/VRR_MSR

    2014

  64. [72]

    H. A. Rose, Physica D Nonlinear Phenomena14, 216 (1985)

    1985

  65. [73]

    Delamotte, inLecture Notes in Physics(Springer, 2012), vol

    B. Delamotte, inLecture Notes in Physics(Springer, 2012), vol. 852, p. 49

    2012

  66. [74]

    Dupuis, L

    N. Dupuis, L. Canet, A. Eichhorn, W. Metzner, J. M. Pawlowski, M. Tissier, and N. Wschebor, Phys. Rep. 910, 1 (2021)

    2021

  67. [75]

    Canet, J

    L. Canet, J. Fluid. Mech.950, P1 (2022)

    2022

  68. [76]

    Fontaine, M

    C. Fontaine, M. Tarpin, F. Bouchet, and L. Canet, Sci- Post Phys.15, 212 (2023)

    2023

  69. [77]

    Goldreich and S

    P. Goldreich and S. Sridhar, ApJ438, 763 (1995)

    1995

  70. [78]

    Y. B. Ginat, M. L. Nastac, R. J. Ewart, S. Konrad, M. Bartelmann, and A. A. Schekochihin, Phys. Rev. D 112, 063501 (2025)

    2025

  71. [79]

    Wang and B

    H. Wang and B. Kocsis, Phys. Rev. D108, 103004 (2023)

    2023

  72. [80]

    Fragione and A

    G. Fragione and A. Loeb, ApJL932, L17 (2022)

    2022

  73. [81]

    Paumard et al., ApJ643, 1011 (2006)

    T. Paumard et al., ApJ643, 1011 (2006). 26

    2006

  74. [82]

    Bartko et al., ApJ697, 1741 (2009)

    H. Bartko et al., ApJ697, 1741 (2009)

    2009

  75. [83]

    J. R. Lu et al., ApJ690, 1463 (2009)

    2009

  76. [84]

    Yelda et al., ApJ783, 131 (2014)

    S. Yelda et al., ApJ783, 131 (2014)

    2014

  77. [85]

    S. D. von Fellenberg et al., ApJL932, L6 (2022)

    2022

  78. [86]

    J. I. Read, T. Goerdt, B. Moore, A. P. Pontzen, J. Stadel, and G. Lake, MNRAS373, 1451 (2006)

    2006

  79. [87]

    M. I. Zelnikov and D. S. Kuskov, MNRAS455, 3597 (2016)

    2016

  80. [88]

    Banik and F

    U. Banik and F. C. van den Bosch, ApJ926, 215 (2022)

    2022

  81. [89]

    Kaur and N

    K. Kaur and N. C. Stone, MNRAS515, 407 (2022)

    2022

  82. [90]

    D. A. Varshalovich et al.,Quantum Theory of Angular Momentum(World Scientific, 1988)

    1988

  83. [91]

    Commenges and M

    D. Commenges and M. Monsion, IEEE Transactions on automatic control29, 250 (1984)

    1984

  84. [92]

    Munthe-Kaas, Appl

    H. Munthe-Kaas, Appl. Numer. Math.29, 115 (1999)

    1999

  85. [93]

    Fouvry, W

    J.-B. Fouvry, W. Dehnen, S. Tremaine, and B. Bar-Or, ApJ931, 8 (2022)

    2022

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.