Conditional Fluctuations of Mean-Field Systems with Stable Common Noise
Listen
Radio episode about this paper
Transcript
Introduction to the show: ident: Mathematics. The day in mathematics: new theorems, new proofs, and what they open.
Ada: We're looking at the paper: "Conditional Fluctuations of Mean-Field Systems with Stable Common Noise".
Emmy: Here's the gist of the paper. Conditional fluctuations of smooth mean-field systems on a torus with endogenous symmetric stable common jumps are studied to understand how sampling and common-signal errors combine and how their scales and path limits depend on the coupling with the limiting signal N,five or more words copied word for word from that page.
Ada: The conditional laws given the full Poisson master converge in probability to a centered Gaussian kernel in a negative Sobolev path space N,five or more words copied word for word from that page.
Emmy: Conditional Central Limit Theorem The paper establishes a conditional functional central limit theorem under empirical Poisson thinning N,five or more words copied word for word from that page.
Ada: Specifically, the limit is centered Gaussian, starts from the µ0-Brownian bridge, and satisfies a specific equation involving a continuous Gaussian H−s-valued martingale N,five or more words copied word for word from that page.
Paper summary: Ada: So wrapping up this paper "Conditional Fluctuations of Mean-Field Systems with Stable Common Noise," it's about the conditional fluctuations of mean-field systems on a torus with these common jumps. The authors explore how sampling and common signal errors combine and what their scales are, depending on the coupling with the limiting signal N sub five or more words copied word for word from that page.
Emmy: In simple terms, they're showing that even when you have these complicated interactions, if you meet certain conditions about the noise index alpha less than one and rho greater than alpha over two, things stabilize into a centered Gaussian shape in a negative Sobolev path space N−alpha < one <ref:2610.01391#pg1,in a negative Sobolev path space>.
Kolya: The main implication is that we can use this to understand how errors scale in these complex systems and what those tails look like, which is vital for simulation because it tells us how to handle the errors.
Ada: And they also found specific points, like criticality where things hit a Gaussian limit with mean chi and a covariance of eta N sub five or more words copied word for word from that page. This gives us concrete targets to aim for when designing these systems.
Emmy: The title itself hints at the deep structure they're uncovering—how the rank and clock couplings lead to the same microscopic laws, which points toward a consistent underlying mathematical framework across different ways of looking at the system.
Conclusion: Ada: So we're wrapping up this talk on conditional fluctuations of mean-field systems with stable common noise. We've seen how these complex interactions behave when you have those common jumps mixed in there, and now we need to talk about what that really means for the bigger picture.
Emmy: Exactly. The title itself, "Conditional Fluctuations of Mean-Field Systems with Stable Common Noise," it sounds super technical, but at its core, it's about how errors combine when you have a shared signal influencing a whole system. It’s not just one error type; it’s the combination of sampling errors and common signal errors that we're tracking here.
Kolya: From my side, what I want to make sure everyone gets is that this isn't just abstract math in a paper. It’s about understanding the actual shape of these fluctuations, especially when things get tricky with those stable processes involved. The results show how the scaling of these errors changes depending on how much you couple that common signal to the system itself.
Ada: Right. So what does that scaling actually tell us practically? We're looking at how these limits behave for different coupling strengths, like with a signal N sub five or more words copied word for word from that page. It’s about finding the right conditions where things settle down into a predictable Gaussian shape.
Emmy: And that predictability is key. The paper shows us exactly when those fluctuations become centered Gaussian kernels in a specific path space, which means we know what kind of "noise" to expect if you're running these simulations or modeling these real-world systems. It gives us the structure of the limiting behavior.
Kolya: The interesting part is seeing how that structure changes across three different sampling regimes based on a parameter rho. It shows that depending on the setting, you get different functional limits for things like the root-N field N sub five or more words copied word for word from that page. That's a lot of information about when and where these systems are most stable or most volatile.
Ada: So, in short, this paper lays out a roadmap for understanding how sampling errors and common signal errors work together in these mean-field settings. It sets the stage for knowing exactly what kind of statistical behavior we should be looking for under different conditions.
Emmy: And that leads right into how we actually test those limits—the translation extensions and the covariance formulas they develop are crucial because they give us a way to check if our approximations hold up when we move from theory to actual computation.
Alexandre Autran
École Normale Supérieure de Rennes
math.PR
Submitted: 2026-10-01
Updated: 2026-11-08
Comments: 39 pages, 1 table
License: http://creativecommons.org/licenses/by/4.0/
Importance score: 76/100
The gist: Conditional fluctuations of smooth mean-field systems on a torus with endogenous symmetric stable common jumps are studied to understand how sampling and common-signal errors combine and how their…
Key concepts
- Conditional Central Limit Theorem
- This theorem describes the statistical behavior of empirical fluctuations under specific conditions. It shows that when sampling is done via empirical Poisson thinning, these fluctuations converge to a centered Gaussian distribution. The limit starts from a specific process called the $\mu_0$-Brownian bridge and satisfies an equation involving a continuous Gaussian martingale in a path space defined by the Skorokhod $J^1$ topology.
- Rank Coupling
- Rank coupling is used to separate and analyze the common-signal error from other empirical fluctuations. The work develops criteria, like an inverse-tail criterion and rank trichotomy, to understand this contribution. Under certain assumptions, it leads to limits involving a stable process $Z$ with an index $\kappa = \alpha / (1 - \rho)$ in the first sampling regime.
- Sharpness and Critical Regimes
- The analysis categorizes the system's behavior into three distinct sampling regimes based on the parameter $\rho$. These regimes dictate different functional limits for the root-$N$ field. At a critical point, where $\rho = 0$, the limit becomes Gaussian with a specific mean $\chi$ and covariance structure for the martingale part.
- Translation Extensions
- This extension shows that stability holds even when considering an isometry of every $H_r(R)$, which is a completion of the Schwartz space under a specific norm. By combining joint convergence with $S$ and L1 approximation tests, the conditional covariance structure between two fields is precisely determined by the covariance of their martingale parts.
Terminology
Summary
Conditional fluctuations of smooth mean-field systems on a torus with endogenous symmetric stable common jumps are studied to understand how sampling and common-signal errors combine and how their scales and path limits depend on the coupling with the limiting signal N,five or more words copied word for word from that page. p. 0 Conditional Fluctuations of Mean-Field Systems
with Stable Common Noise
Alexandre Autran1
Abstract
We study conditional fluctuations of smooth mean-field systems on a torus with endogenous symmetric stable
common jumps of index 0 < α < 1. A positive-strip estimate makes empirical Poisson acceptance errors negligible at
the sampling scale. A positive-strip estimate makes empirical Poisson acceptance errors negligible at
the sampling scale. The conditional laws given the full Poisson master converge in probability to a centered Gaussian
kernel in a negative Sobolev path space.Where the paper says it
Where the paper says it
Conditional Central Limit Theorem
The paper establishes a conditional functional central limit theorem under empirical Poisson thinning N,five or more words copied word for word from that page. Specifically, the limit is centered Gaussian, starts from the µ0-Brownian bridge, and satisfies a specific equation involving a continuous Gaussian H−s-valued martingale N,five or more words copied word for word from that page. p. 4 Conditionally on G0
T
, the limit is centered Gaussian, starts from the
µ0-Brownian bridge, and satisfies, for ϕ ∈ C∞(Td
),
ηt(ϕ) = η0(ϕ) + Mt(ϕ) + Z t
0
ηr(Bµr ϕ) dr (7)
+
Z
[0,t]×Rd×[0,f¯]
1{u≤µr−(f)}ηr−
ϕ(· + z) − ϕ
N0
( dr, dz, du). Here M is a continuous, conditionally centered Gaussian H−s
-valued martingale, conditionally independent of η0, with
⟨M(ϕ), M(ψ)⟩
t =
Z t
0
µr(∇ϕ
⊤a∇ψ) dr. Then
η
N st(G
0
T )
====⇒
N→∞
η in D([0, T]; H−s
). (6)
The path space carries the Skorokhod J1 topology.Where the paper says it
Where the paper says it
Rank Coupling and Microscopic Transfer
The rank coupling identifies the common-signal error and its contribution to empirical fluctuations N,five or more words copied word for word from that page. The work develops an inverse-tail criterion and a rank trichotomy to analyze the behavior of the limit N,five or more words copied word for word from that page. p. 0 In dimension one, quantile coupling preserves this limit for microscopic marks
with relative tail remainder O(x
−ρ
), ρ > α/2. A quantified second-order tail expansion gives the endogenous rank
response: it shifts the conditional Gaussian mean at criticality and, when nonzero, dominates sampling below criticality. Symmetry controls small-jump errors
without assuming that U has a density or a finite second moment. Under the quantified expansion of Lemma 9 with
nonzero second-order coefficient, Theorem 10 gives a stable rank tangent of index κ := α/(1 − ρ) for ρ < 1 − α/2.Where the paper says it
Where the paper says it
Clock-Coupled Counterpoint
The clock-coupled counterpoint compares two empirical centers for a finite-state population translated by a stable process at a random clock N,five or more words copied word for word from that page. Lemma 16 shows that the joint population and clock fluctuations converge to a continuous, Gaussian limit independent of S N,five or more words copied word for word from that page. p. 1 Section 4 treats a finite-state model with L
N = SKN /N , where KN counts type transitions and S is
an independent symmetric α-stable Lévy process. A joint population–clock CLT gives the finite-dimensional signal
correction. The martingale part is
McN
t
(ϕ) := 1
√
N
X
N
i=1
Z t
0
∇ϕ(Y
i,N
u
)
⊤σ dWi
u
,
D
McN (ϕ), McN (ψ)
E
t
=
Z t
0
ν
N
u
(∇ϕ
⊤a∇ψ) du. For r > d/2 + 1, Sobolev embedding gives
|ζ(gj,t)| ≤ C∥ζ∥H−r , ∥∇ · (c
N
j,tνt)∥H−r ≤ C,
and the covariance trace per unit time is bounded by
4π
2 X
k∈Zd
(1 + |k|
2
)
−r
k
⊤ak < ∞.Where the paper says it
Where the paper says it
Sharpness and Critical Regimes
The analysis identifies three sampling regimes based on the parameter ρ, leading to different functional limits for the root-N field N,five or more words copied word for word from that page. For a 0 N,five or more words copied word for word from that page. At criticality, the limit is Gaussian with mean χ and the covariance of η N,five or more words copied word for word from that page. p. 0 In dimension one, quantile coupling preserves this limit for microscopic marks
with relative tail remainder O(x
−ρ
), ρ > α/2. A quantified second-order tail expansion gives the endogenous rank
response: it shifts the conditional Gaussian mean at criticality and, when nonzero, dominates sampling below criticality.Where the paper says it
Translation Extensions
Translation extends to an isometry of every Hr(R), where Hr(R) is the completion of the Schwartz space S(R) under a specific norm N,five or more words copied word for word from that page. The joint convergence with S, followed by L1 approximation of bounded σ(S)-measurable tests, gives stability N,five or more words copied word for word from that page. p. 3 For a
Polish space E with bounded compatible metric dE, define
dBL(P, Q) := sup
∥F ∥∞+LipdE
(F )≤1
|P(F) − Q(F)|, P, Q ∈ P(E). If conditional kernels KN , K given G0
T
satisfy dBL(KN , K) → 0 in probability, then, for bounded G0
T
-measurable Z
and bounded Lipschitz F,
E [ZKN (F)] −→ E [ZK(F)] ,
which gives stable convergence [1]. A positive-strip estimate makes empirical Poisson acceptance errors negligible at
the sampling scale. The conditional laws given the full Poisson master converge in probability to a centered Gaussian
kernel in a negative Sobolev path space.Where the paper says it
Where the paper says it
Conclusion on Convergence
The overall result is that the conditional bounded–Lipschitz distance between the microscopic field and this approximation is bounded by a constant times E h 1 ∧ εN N,five or more words copied word for word from that page. This proves (57) and its joint version N,five or more words copied word for word from that page. p. 1 Since
α/2 < 1 − α/2 for α < 1, the sampling threshold lies in the stable rank regime. Theorem 14 proves
µ
N,mic − µ = N
−1/2
ηe
N + N
−ρ/αχ + oP(N
−ρ/α), 0 < ρ ≤ α/2,
uniformly in the negative Sobolev norm, where ηe
N := √
N(µe
N − µ) and χ is the linear response to the rank forcing
accepted by the reference threshold. A positive-strip estimate makes empirical Poisson acceptance errors negligible at
the sampling scale. The conditional laws given the full Poisson master converge in probability to a centered Gaussian
kernel in a negative Sobolev path space. A positive-strip estimate makes empirical Poisson acceptance errors negligible at
the sampling scale. The conditional laws given the full Poisson master converge in probability to a centered Gaussian
kernel in a negative Sobolev path space.Where the paper says it
Where the paper says it
Where the paper says it
References
[1] D. J. Aldous and G. K. Eagleson, On mixing and stability of limit theorems N,five or more words copied word for word from that page>. p. 0 Set
A
N
t
:= Z t
0
|µ
N
r
(f) − µr(f)| dr, V N
t
:= X
r≤t
|∆L
N
r − ∆Lr|. Predictable compensation and synchronous stability give, for stopping times τ ≤ T and x, v > 0,
P(V
N
τ > x, AN
τ ≤ v) ≤ Cvx−α
, AN
T = OP(N
−1/2
),
V
N
T = OP(N
−1/(2α)
) = oP(N
−1/2
), 0 < α < 1.Where the paper says it
[2] S. Asmussen and J. p. 1 Rank representations and stable invariance principles are developed in Ferguson and Klass [6], LePage
et al. [10, 11], Rosinski ´ [17], and Gaussian small-jump approximations in Asmussen and Rosinski ´ [2]. Rank representations and stable invariance principles are developed in Ferguson and Klass [6], LePage
et al. [10, 11], Rosinski ´ [17], and Gaussian small-jump approximations in Asmussen and Rosinski ´ [2].Where the paper says it
Where the paper says it
[3] A. Budhiraja, E. p. 1 Corollary 13 covers
source exclusion. Conditional Gaussian fluctuations with a common factor are studied in Budhiraja et al. The main results are a conditional
functional central limit theorem under empirical Poisson thinning (Theorem 2), an endogenous rank-response expansion
(Theorem 14), and a model with identical microscopic laws but different additive path fluctuations under the two
couplings (Proposition 23). Sampling and common-signal errors.Where the paper says it
Where the paper says it
[4] F. Delarue, D. p. 1 Conditional Gaussian fluctuations with a common factor are studied in Budhiraja et al. [3], Delarue
et al. We
study two couplings: matching Poisson ranks and changing the clock of a stable path. The main results are a conditional
functional central limit theorem under empirical Poisson thinning (Theorem 2), an endogenous rank-response expansion
(Theorem 14), and a model with identical microscopic laws but different additive path fluctuations under the two
couplings (Proposition 23).Where the paper says it
Where the paper says it
[5] S. N. Ethier and T. G. Kurtz, Markov Processes: Characterization and Convergence N,five or more words copied word for word from that page>. p. 1 We use the Hilbert-space sampling method
of Fernandez and Méléard [7], Kurtz and Xiong [9], then control acceptance errors by a positive-strip estimate and
microscopic errors by quantile estimates. Both steps preserve convergence in probability of the conditional laws given
the full master.Where the paper says it
[6] T. S. Ferguson and M. p. 1 Both steps preserve convergence in probability of the conditional laws given
the full master. Rank representations and stable invariance principles are developed in Ferguson and Klass [6], LePage
et al.Where the paper says it
[7] B. Fernandez and S. p. 1 Nikolaev [15] treats Brownian common noise with less regular interactions. We use the Hilbert-space sampling method
of Fernandez and Méléard [7], Kurtz and Xiong [9], then control acceptance errors by a positive-strip estimate and
microscopic errors by quantile estimates. Above each cutoff, there are finitely many candidate translations, with Lipschitz diffusions between them. Between candidates the reference law solves the usual Lipschitz McKean–Vlasov equation.Where the paper says it
Where the paper says it
[8] J. Jacod and A. N. Shiryaev, Limit Theorems for Stochastic Processes, volume 288 of Grundlehren der mathematischen Wissenschaften N,five or more words copied word for word from that page>.
[9] T. G. Kurtz and J. p. 1 Nikolaev [15] treats Brownian common noise with less regular interactions. We use the Hilbert-space sampling method
of Fernandez and Méléard [7], Kurtz and Xiong [9], then control acceptance errors by a positive-strip estimate and
microscopic errors by quantile estimates.Where the paper says it
[10] R. LePage, K. Podgórski, and M. Ryznar, Strong and conditional invariance principles for samples attracted to stable laws N,five or more words copied word for word from that page>. p. 1 Both steps preserve convergence in probability of the conditional laws given
the full master. Rank representations and stable invariance principles are developed in Ferguson and Klass [6], LePage
et al.Where the paper says it
[11] R. LePage, M. Woodroofe, and J. Zinn, Convergence to a stable distribution via order statistics N,five or more words copied word for word from that page>. p. 1 Our rank tangent
describes the residual of a fixed quantile coupling jointly with its leading signal. The clock analysis combines martingale
CLTs and time-change theory [5, 16, 18] to identify additive second-order errors and the obstruction to their path
convergence.Where the paper says it
[12] E. Löcherbach and D. p. 1 Related work. Stable common-noise limits and conditional propagation of chaos are established in Löcherbach and
Loukianova [12], Löcherbach et al. Conditional Fluctuations of Mean-Field Systems
with Stable Common Noise
Alexandre Autran1
Abstract
We study conditional fluctuations of smooth mean-field systems on a torus with endogenous symmetric stable
common jumps of index 0 < α < 1. A positive-strip estimate makes empirical Poisson acceptance errors negligible at
the sampling scale.Where the paper says it
Where the paper says it
[13] E. Löcherbach, D. p. 1 Related work. Stable common-noise limits and conditional propagation of chaos are established in Löcherbach and
Loukianova [12], Löcherbach et al. Related work. Stable common-noise limits and conditional propagation of chaos are established in Löcherbach and
Loukianova [12], Löcherbach et al. Related work. Stable common-noise limits and conditional propagation of chaos are established in Löcherbach and
Loukianova [12], Löcherbach et al. Related work. Stable common-noise limits and conditional propagation of chaos are established in Löcherbach and
Loukianova [12], Löcherbach et al. Stable common-noise limits and conditional propagation of chaos are established in Löcherbach and
Loukianova [12], Löcherbach et al. The strong result gives a quantitative coupling [13, Theorem 2.Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
[14] E. Löcherbach, D. Loukianova, and E. Marini, Weak conditional propagation of chaos for systems of interacting particles with nearly stable jumps N,five or more words copied word for word from that page>. p. 1 Related work. Stable common-noise limits and conditional propagation of chaos are established in Löcherbach and
Loukianova [12], Löcherbach et al.Where the paper says it
[15] P. Nikolaev, Fluctuation behaviour for interacting particle systems with common noise N,five or more words copied word for word from that page>. p. 1 Our selection has infinite stable activity and allows degenerate private diffusion. Nikolaev [15] treats Brownian common noise with less regular interactions.Where the paper says it
[16] R. Rebolledo, Central limit theorems for local martingales N,five or more words copied word for word from that page>. p. 0 We
study two couplings: matching Poisson ranks and changing the clock of a stable path. The main results are a conditional
functional central limit theorem under empirical Poisson thinning (Theorem 2), an endogenous rank-response expansion
(Theorem 14), and a model with identical microscopic laws but different additive path fluctuations under the two
couplings (Proposition 23).Where the paper says it
[17] J. Rosinski, On series representations of infinitely divisible random vectors N,five or more words copied word for word from that page>. p. 1 Rank representations and stable invariance principles are developed in Ferguson and Klass [6], LePage
et al. [10, 11], Rosinski ´ [17], and Gaussian small-jump approximations in Asmussen and Rosinski ´ [2].Where the paper says it
[18] W. Whitt, Stochastic-Process Limits: An Introduction to Stochastic-Process Limits and Their Application to Queues N,five or more words copied word for word from that page>. p. 0 Introduction. Heavy-tailed collective jumps can produce stable common noise in mean-field limits.Where the paper says it
Pruning redundant reasoning tokens
The paper establishes a conditional functional central limit theorem under empirical Poisson thinning N,five or more words copied word for word from that page>. Specifically, the limit is centered Gaussian, starts from the µ0-Brownian bridge, and satisfies a specific equation involving a continuous Gaussian H−s-valued martingale N,five or more words copied word for word from that page>. p. 4 Conditionally on G0
T
, the limit is centered Gaussian, starts from the
µ0-Brownian bridge, and satisfies, for ϕ ∈ C∞(Td
),
ηt(ϕ) = η0(ϕ) + Mt(ϕ) + Z t
0
ηr(Bµr ϕ) dr (7)
+
Z
[0,t]×Rd×[0,f¯]
1{u≤µr−(f)}ηr−
ϕ(· + z) − ϕ
N0
( dr, dz, du). Here M is a continuous, conditionally centered Gaussian H−s
-valued martingale, conditionally independent of η0, with
⟨M(ϕ), M(ψ)⟩
t =
Z t
0
µr(∇ϕ
⊤a∇ψ) dr. Then
η
N st(G
0
T )
====⇒
N→∞
η in D([0, T]; H−s
). (6)
The path space carries the Skorokhod J1 topology.Where the paper says it
Where the paper says it
How the training signal is built
The work develops an inverse-tail criterion and a rank trichotomy to analyze the behavior of the limit N,five or more words copied word for word from that page>. Under Assumption 8, which controls the inverse-tail control, Theorem 10 gives limits involving a stable process Z with index κ:= α/(1 − ρ) in the first regime N,five or more words copied word for word from that page>. p. 1 Symmetry controls small-jump errors
without assuming that U has a density or a finite second moment. Under the quantified expansion of Lemma 9 with
nonzero second-order coefficient, Theorem 10 gives a stable rank tangent of index κ := α/(1 − ρ) for ρ < 1 − α/2.Where the paper says it
Where the gains come from
The analysis identifies three sampling regimes based on the parameter ρ, leading to different functional limits for the root-N field N,five or more words copied word for word from that page>. For a 0 N,five or more words copied word for word from that page>.
Translation extends to an isometry of every Hr(R), where Hr(R) is the completion of the Schwartz space S(R) under a specific norm N,five or more words copied word for word from that page>. The joint convergence with S, followed by L1 approximation of bounded σ(S)-measurable tests, gives stability N,five or more words copied word for word from that page>. p. 3 For a
Polish space E with bounded compatible metric dE, define
dBL(P, Q) := sup
∥F ∥∞+LipdE
(F )≤1
|P(F) − Q(F)|, P, Q ∈ P(E). If conditional kernels KN , K given G0
T
satisfy dBL(KN , K) → 0 in probability, then, for bounded G0
T
-measurable Z
and bounded Lipschitz F,
E [ZKN (F)] −→ E [ZK(F)] ,
which gives stable convergence [1].Where the paper says it
The overall result is that the conditional bounded–Lipschitz distance between the microscopic field and this approximation is bounded by a constant times E h 1 ∧ εN N,five or more words copied word for word from that page>. This proves (57) and its joint version N,five or more words copied word for word from that page>. p. 1 Since
α/2 < 1 − α/2 for α < 1, the sampling threshold lies in the stable rank regime. Theorem 14 proves
µ
N,mic − µ = N
−1/2
ηe
N + N
−ρ/αχ + oP(N
−ρ/α), 0 < ρ ≤ α/2,
uniformly in the negative Sobolev norm, where ηe
N := √
N(µe
N − µ) and χ is the linear response to the rank forcing
accepted by the reference threshold.Where the paper says it
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
-
Enhanced understanding of conditional fluctuations in mean-field systems allows for modeling empirical sampling errors more accurately using positive-strip estimates to ensure negligible errors at sampling scale.
-
Improved simulation fidelity for heavy-tailed collective jumps by incorporating endogenous rank response expansions, which shift the conditional Gaussian mean at criticality and dominate sampling below criticality.
-
Development of a stable-to-Brownian transition analysis for rank error provides a more nuanced view of the rank error behavior compared to standard Brownian noise models.
-
Construction of coupling methods, such as matching Poisson ranks and changing the clock, allows comparison between different empirical fluctuation scales and path limits in mean-field systems.
-
Ability to distinguish between rank coupling and clock coupling effects on microscopic laws, leading to a more detailed analysis of additive path fluctuations under different master-measurable forcings.
-
Implementation of a functional central limit theorem under Poisson thinning enables the conditional law given the full Poisson master to converge in probability to a centered Gaussian kernel in negative Sobolev path space.
-
Application of quantile coupling methods, such as rank and clock couplings yielding the same microscopic laws, provides a method for comparing different limiting behaviors while maintaining identical underlying mark structures.
-
Capacity to compute an endogenous rank response expansion that determines whether the conditional Gaussian mean shifts at criticality or dominates sampling below criticality based on the second-order tail expansion.
-
Identification of a stable-to-Brownian transition for rank error provides a criterion for determining when the rank error behaves like Brownian noise versus stable noise based on its square integrability.
-
Implementation of a clock-coupled counterpoint allows for comparison between empirical centers translated by a stable process and subsequent construction of rank and grid couplings with identical microscopic laws.
-
Capacity to establish joint convergence results for the initial field and martingale, yielding a centered Gaussian martingale independent of the initial bridge with covariance determined by the integral of the quadratic variation.
-
Development of a pathwise unique adapted solution for the master equation under specific assumptions provides rigorous tools for analyzing conditional law flows in finite-particle settings.
-
Ability to prove tightness and convergence of fields like L N,r - L r in probability using positive disagreement strips and strip estimates, showing that
the paths are Cauchy in probability for the uniform metric.
-
Capability to analyze three distinct sampling regimes based on the tail index alpha (0 < alpha < 2), providing explicit functional limits for the root-N field under different conditions.
-
Tool to detect non-tightness of fields at a fixed time if the subcritical expansion shows that
the root-N field is not tight at that time,
which is crucial for model selection and sharpness proofs.
Abstract
We study conditional fluctuations of smooth mean-field systems on a torus with symmetric stable common jumps of index 0<α<1. A positive-strip estimate makes empirical Poisson acceptance errors negligible at the sampling scale. Given the full Poisson master, the conditional laws converge in probability to a centered Gaussian kernel in a negative Sobolev path space. In dimension one, quantile coupling preserves this limit for microscopic marks with relative tail remainder O(x-ρ), where ρ> α 2. A second-order tail expansion identifies the response when this condition fails: it shifts the conditional Gaussian mean at criticality and, when nonzero, dominates sampling below criticality. We also identify the transition from a stable to a Brownian rank error. In a translation model, rank and clock couplings give the same microscopic laws but different path fluctuations. Rank coupling gives a functional Gaussian limit. Clock coupling admits no diverging additive path normalization at the limiting center.
Sources
Related papers
- Rank-One HCIZ Integrals and Multiplicative Convolutions
- Markov Chain CLTs: Resolving Open Problems
- Marked stable L'evy processes in Brownian local time
- A Sharp Refinement of Samuels' Inequality via Cap Moves
- A zero-one law for swap-invariant events in uniform spanning forests
- Bernoulli flow for Erd s-R'enyi graphs