Pairwise Independent Brownian Motions

• Essentially pairwise independent Brownian motions are defined as families where each indexed process is a standard Brownian motion and almost every two processes are independent.
• The framework employs Fubini extensions to guarantee joint measurability in continuum-agent models, ensuring idiosyncratic noise with global averaging properties.
• Asymptotic constructions, such as bootstrap random walks, show that strongly dependent pre-limit processes converge to independent Gaussian processes under diffusive scaling.

Essentially pairwise independent Brownian motions are families of Brownian motions indexed by a set $I$ such that, for almost every index $u$, the process indexed by $u$ is a standard Brownian motion, and for almost every pair $(u,v)$, the two indexed processes are independent. In the recent literature, the term appears in two closely related but distinct senses. In one sense, it is a measure-theoretic notion for continua of Brownian motions on a Fubini extension space, where full joint measurability is compatible with “almost everywhere” pairwise independence (Amini et al., 15 Sep 2025). In another sense, it describes asymptotic phenomena in which strongly dependent pre-limit processes converge, after scaling, to Brownian motions with independent coordinates; the dependence persists at finite scale but disappears in the limit (Collevecchio et al., 2015, Burdzy, 2014). These usages are connected by a common theme: independence at the Brownian level can emerge either as an exact structural property on an extended probability space or as a scaling-limit property of nonlinearly coupled stochastic systems.

1. Formal definition and probabilistic setting

The precise measure-theoretic notion is formulated on two probability spaces: an index space $(I,\mathcal I,\lambda)$ and a sample space $(\Omega,\mathcal F,\mathbb Q)$. A random variable $\Theta:I\times\Omega\to\mathbb R$ is called essentially pairwise independent if, for $\lambda$-almost every $u\in I$, the random variables $\Theta(u,\cdot)$ and $u$0 are independent for $u$1-almost every $u$2 (Amini et al., 15 Sep 2025). The qualifier “essentially” means that independence may fail on index sets of $u$3-measure zero.

For Brownian motions, the corresponding notion is an indexed family $u$4 such that, for every $u$5, $u$6 is a standard Brownian motion on $u$7, the map $u$8 is jointly measurable for each $u$9, and the family is essentially pairwise independent (Amini et al., 15 Sep 2025). This formulation is designed for continuum-agent models, where one wants idiosyncratic Brownian noise for almost every agent while retaining joint measurability.

A distinct but compatible interpretation appears in scaling-limit results. In "Bootstrap Random Walks" (Collevecchio et al., 2015), a collection of random walks is generated from a single i.i.d. increment sequence by repeated nonlinear “bootstrapping”; at the discrete level the coordinates are deterministic functionals of the same underlying randomness, but under diffusive scaling the limit is a multidimensional Brownian motion with independent components. In that setting, “essentially pairwise independent Brownian motions” means that the Brownian objects are independent in the limit even though the pre-limit coordinates are far from independent (Collevecchio et al., 2015).

This suggests a useful conceptual distinction. One may speak of exact e.p.i. Brownian families when independence is formulated directly on an extended probability space, and of asymptotically independent Brownian coordinates when independent Brownian motions arise as limits of dependent systems. The literature supports both interpretations, but keeps them technically separate.

2. Fubini extension spaces and continuum Brownian families

The need for Fubini extensions arises because the classical product space

$u$0

does not support a jointly measurable continuum of independent random variables (Amini et al., 15 Sep 2025). Sun’s Fubini extension enlarges this product to

$u$1

with the property that Fubini’s theorem continues to hold for every integrable $u$2-measurable function $u$3: $u$4 This framework permits jointly measurable continuum families with prescribed marginal laws (Amini et al., 15 Sep 2025).

Within this setting, a process $u$5 on $u$6 is a standard Brownian motion if $u$7 $u$8-a.s., its increments are Gaussian with variance $u$9, its increments over disjoint intervals are independent, and for $(u,v)$0-almost every $(u,v)$1 the path $(u,v)$2 is continuous (Amini et al., 15 Sep 2025). The associated coordinate family is given by

$(u,v)$3

The main structural equivalence is explicit. If $(u,v)$4 is an e.p.i. collection of Brownian motions, then

$(u,v)$5

defines a standard Brownian motion on the Fubini extension space (Amini et al., 15 Sep 2025). Conversely, if $(u,v)$6 is a standard Brownian motion on the Fubini extension whose coordinates are pairwise independent and identically distributed, then the coordinate processes $(u,v)$7 form an e.p.i. collection of Brownian motions on the marginal sample space (Amini et al., 15 Sep 2025). The identically distributed condition is essential: the paper notes that one can construct families whose aggregate process is Gaussian under $(u,v)$8 while individual coordinates are not Brownian under $(u,v)$9 (Amini et al., 15 Sep 2025).

A further consequence is the Exact Law of Large Numbers. If $(I,\mathcal I,\lambda)$0 is integrable and e.p.i. on the Fubini extension, then for every $(I,\mathcal I,\lambda)$1,

$(I,\mathcal I,\lambda)$2

Applied to an e.p.i. Brownian family, this yields

$(I,\mathcal I,\lambda)$3

for every $(I,\mathcal I,\lambda)$4 (Amini et al., 15 Sep 2025). In continuum-agent models, this pathwise averaging property plays the role usually associated with laws of large numbers for independent sequences.

3. Preservation under measure change and graphon SDEs

A notable feature of the Fubini-extension framework is that it supports a Girsanov theorem preserving pairwise independence at the level of coordinates. Starting from an e.p.i. Brownian family $(I,\mathcal I,\lambda)$5 on $(I,\mathcal I,\lambda)$6, and a coordinate-wise adapted process $(I,\mathcal I,\lambda)$7, one defines the exponential martingale

$(I,\mathcal I,\lambda)$8

and the new measure

$(I,\mathcal I,\lambda)$9

The drifted global process

$(\Omega,\mathcal F,\mathbb Q)$0

is then a standard Brownian motion on $(\Omega,\mathcal F,\mathbb Q)$1 (Amini et al., 15 Sep 2025).

The coordinate-wise implication is subtler. Under $(\Omega,\mathcal F,\mathbb Q)$2, the family $(\Omega,\mathcal F,\mathbb Q)$3 remains essentially pairwise independent, but the individual coordinate processes need not remain Brownian on $(\Omega,\mathcal F,\mathbb Q)$4; for instance,

$(\Omega,\mathcal F,\mathbb Q)$5

unless the drift has zero mean (Amini et al., 15 Sep 2025). Thus the global Brownian structure is preserved on the extended space, while coordinate marginals may acquire drift. The Exact Law of Large Numbers is crucial in the independence argument after measure change (Amini et al., 15 Sep 2025).

The same paper uses this noise structure to reformulate graphon stochastic differential equations. For a symmetric Borel measurable graphon

$(\Omega,\mathcal F,\mathbb Q)$6

the continuum system

$(\Omega,\mathcal F,\mathbb Q)$7

can be rewritten as a single McKean–Vlasov type equation on the Fubini extension: $(\Omega,\mathcal F,\mathbb Q)$8 where

$(\Omega,\mathcal F,\mathbb Q)$9

Under the stated Lipschitz and linear-growth assumptions on $\Theta:I\times\Omega\to\mathbb R$0 and $\Theta:I\times\Omega\to\mathbb R$1, the equation admits a unique solution in the corresponding square-integrable path space (Amini et al., 15 Sep 2025). In this application, essentially pairwise independent Brownian motions provide the idiosyncratic noise for a continuum of agents while retaining a jointly measurable, globally analyzable structure.

4. Asymptotic independence from a single source of randomness

A different route to pairwise independent Brownian motions begins from a single i.i.d. increment sequence. In "Bootstrap Random Walks" (Collevecchio et al., 2015), one starts from the simple symmetric random walk

$\Theta:I\times\Omega\to\mathbb R$2

and defines new increments by partial products,

$\Theta:I\times\Omega\to\mathbb R$3

Although $\Theta:I\times\Omega\to\mathbb R$4 is again i.i.d. symmetric $\Theta:I\times\Omega\to\mathbb R$5, the dependence between $\Theta:I\times\Omega\to\mathbb R$6 and $\Theta:I\times\Omega\to\mathbb R$7 is exact and invertible: $\Theta:I\times\Omega\to\mathbb R$8 and the filtrations coincide: $\Theta:I\times\Omega\to\mathbb R$9 Thus $\lambda$0 and $\lambda$1 share the same underlying randomness and are not independent at finite time (Collevecchio et al., 2015).

Nevertheless, under the usual diffusive scaling,

$\lambda$2

the process converges weakly to a two-dimensional Brownian motion with independent components (Collevecchio et al., 2015). More generally, after iterating the bootstrap operation and constructing

$\lambda$3

the normalized process converges in $\lambda$4 to a $\lambda$5-dimensional Brownian motion with independent components (Collevecchio et al., 2015).

The independence mechanism is encoded in the covariance structure. For fixed bootstrap levels $\lambda$6 and $\lambda$7, the increment correlations satisfy: $\lambda$8

$\lambda$9

and, for equal times, there is a threshold $u\in I$0 such that

$u\in I$1

Consequently,

$u\in I$2

so the covariance between distinct coordinates stabilizes at constant order while each variance grows linearly in $u\in I$3 (Collevecchio et al., 2015). After normalization by $u\in I$4, the cross-covariances vanish: $u\in I$5 Since the limiting field is Gaussian with diagonal covariance matrix $u\in I$6, the limiting coordinates are independent Brownian motions (Collevecchio et al., 2015).

The construction extends beyond $u\in I$7. If $u\in I$8 takes values in a finite alphabet $u\in I$9, where $\Theta(u,\cdot)$0 is prime and $\Theta(u,\cdot)$1 is an Abelian cyclic group of order $\Theta(u,\cdot)$2, one defines the bootstrap operator

$\Theta(u,\cdot)$3

Under the mean-zero condition

$\Theta(u,\cdot)$4

each bootstrap increment process has the same i.i.d. law as the original increments, and the corresponding vector of walks again converges to a multidimensional Brownian motion with independent components (Collevecchio et al., 2015).

This line of work provides a concrete realization of asymptotic independence: many independent Brownian motions arise from a single underlying source of randomness via nonlinear recycling of increments. The pre-limit dependence is strong, but it is localized in a way that becomes negligible under diffusive scaling (Collevecchio et al., 2015).

5. Interacting systems and asymptotically independent Brownian limits

A related asymptotic-independence phenomenon appears in "Stirring two grains of sand" (Burdzy, 2014). The model consists of a Brownian particle $\Theta(u,\cdot)$5 moving on a flat torus $\Theta(u,\cdot)$6, together with two unit balls whose centers $\Theta(u,\cdot)$7 and $\Theta(u,\cdot)$8 move only when pushed by the Brownian particle, and reflect when they touch so that

$\Theta(u,\cdot)$9

At finite $u$00, the two center processes are dependent: they share the same Brownian driver and interact through reflection (Burdzy, 2014).

The analysis is carried out in the local-time clock. If $u$01 and $u$02 are the local times of $u$03 on the moving spheres surrounding the balls, and

$u$04

then the rescaled center process observed at $u$05 satisfies an invariance principle. For $u$06, with

$u$07

the paper proves that, as $u$08 and $u$09,

$u$10

that is, a pair of independent $u$11-dimensional Brownian motions (Burdzy, 2014). The diffusion coefficient depends on dimension through $u$12.

The same asymptotic decorrelation appears at equilibrium. The process $u$13 has a unique stationary distribution $u$14, and the rescaled stationary law of the centers satisfies

$u$15

the product of uniform laws on the unit torus (Burdzy, 2014). Thus the time-evolving processes and their stationary positions both become asymptotically independent.

This interacting-particle example differs from the bootstrap-random-walk construction in mechanism but not in outcome. In the discrete bootstrap model, asymptotic independence arises from covariance saturation and Gaussian limits (Collevecchio et al., 2015); in the stirred-balls model, it arises from a large-domain limit, a local-time clock, and excursion-theoretic homogenization (Burdzy, 2014). In both settings, strongly coupled finite-scale dynamics converge to independent Brownian motions.

A useful contrast is provided by "Common Decomposition of Correlated Brownian Motions and its Financial Applications" (Chen et al., 2019). There, two correlated Brownian motions $u$16 and $u$17 are decomposed as

$u$18

where $u$19 and $u$20 are independent Brownian motions and

$u$21

The paper shows equivalent conditions for the triplet $u$22 to be mutually independent (Chen et al., 2019). This is not an e.p.i. construction over a continuum of indices, but it exhibits the same structural theme: correlated Brownian objects can be represented in terms of independent Brownian building blocks plus an auxiliary clock.

6. Identification, testing, and related interpretations

For Gaussian processes, pairwise independence is equivalent to vanishing cross-covariance. In "Tests of independence for pairs of paths of non-stationary Gaussian processes" (Ernst et al., 27 Oct 2025), two standard Brownian motions $u$23 are taken jointly Gaussian with

$u$24

In this framework, the two Brownian motions are independent if and only if $u$25 (Ernst et al., 27 Oct 2025). The same equivalence holds for fractional Brownian motions with $u$26, where the cross-covariance is the same covariance kernel multiplied by $u$27 (Ernst et al., 27 Oct 2025).

The paper emphasizes that pathwise empirical correlation does not directly test independence for non-stationary Gaussian processes. For a path $u$28, define

$u$29

and for two paths $u$30,

$u$31

The associated empirical correlation is

$u$32

For independent Brownian motions, $u$33 has a non-degenerate distribution that does not converge as the observation horizon grows; this is the paper’s formulation of Yule’s “nonsense correlation” (Ernst et al., 27 Oct 2025). Instead, the informative object is the discretization error $u$34, where $u$35 is computed from samples at $u$36.

For Brownian motion, the paper proves that

$u$37

and

$u$38

where $u$39 and $u$40 are explicit functionals of the observed paths (Ernst et al., 27 Oct 2025). Under the null hypothesis $u$41, this yields a path-based asymptotic test for independence of Brownian paths. The paper also provides a fully discrete version based on $u$42 (Ernst et al., 27 Oct 2025).

These testing results belong to a different layer of the subject than the construction theorems. The Fubini-extension literature studies how to realize e.p.i. Brownian families as measurable objects (Amini et al., 15 Sep 2025). The bootstrap and stirring papers study how independent Brownian components emerge from dependent dynamics in the limit (Collevecchio et al., 2015, Burdzy, 2014). The testing paper studies how to decide, from observed Gaussian paths, whether the pairwise independence condition $u$43 holds (Ernst et al., 27 Oct 2025). Taken together, these strands show that “essentially pairwise independent Brownian motions” is not a single theorem but a cluster of related ideas: exact measure-theoretic realizability, asymptotic decorrelation under scaling, structural decompositions into independent drivers, and statistical verification from path data.