Controlled Rough Paths

• Controlled rough paths are paths whose increments admit a first-order expansion against a reference rough path with a higher-order remainder, ensuring precise local control.
• They underpin rough integration by converting it into a sewing problem, which allows for fixed-point formulations and the construction of Banach spaces for differential equations.
• Extensions include higher-order, combinatorial, and Hopf-algebraic formulations, with applications in averaging, stochastic control, and Sobolev/càdlàg settings.

Searching arXiv for recent and foundational papers on controlled rough paths to ground the article in the supplied literature. Controlled rough paths are paths whose increments admit a first-order expansion against a reference rough path, with a remainder of strictly higher regularity. In the classical formulation, a path $y$ is controlled by a rough path $X$ if

$y_t-y_s=y'_sX_{s,t}+R^y_{s,t},$

where $y'$ is the Gubinelli derivative and $R^y$ is a higher-order remainder (Duc, 2020). This perspective turns rough integration into a sewing problem and provides Banach spaces in which rough differential equations (RDEs) can be solved by fixed-point arguments (Inahama, 2022, Duc, 2020). Subsequent work has extended the notion across geometric, branched, planar-branched, Sobolev, càdlàg, and mean-field settings, and has clarified that controlled rough paths are not tied to one combinatorial model but can be formulated in a general Hopf-algebraic language (Zhu et al., 27 Sep 2025).

1. Classical definition and analytic role

In the standard level-2 regime, with roughness $2\le p<3$ or Hölder exponent $\alpha\in(1/3,1/2]$, a path $y:[0,T]\to\mathbb R^m$ is controlled by $X$ if there exists a path $y':[0,T]\to \mathcal L(\mathbb R^n,\mathbb R^m)$ such that

$X$0

Here $X$1 is the Gubinelli derivative and $X$2 is a higher-order remainder (Duc, 2020). In the Hölder formulation used in the averaging literature, a controlled path over $X$3 is a triple $X$4 with

$X$5

satisfying

$X$6

This is the formulation used in (Inahama, 2022).

The corresponding controlled-path spaces are Banach spaces. In (Inahama, 2022) the space is denoted

$X$7

with seminorm

$X$8

and full norm

$X$9

In (Duc, 2020), the space of controlled pairs over $y_t-y_s=y'_sX_{s,t}+R^y_{s,t},$0 is denoted $y_t-y_s=y'_sX_{s,t}+R^y_{s,t},$1, with norm built from $y_t-y_s=y'_sX_{s,t}+R^y_{s,t},$2 and $y_t-y_s=y'_sX_{s,t}+R^y_{s,t},$3. These Banach structures are the functional-analytic core of the theory: rough integration preserves controlledness, composition by smooth maps preserves controlledness, and the RDE solution map is realized as a fixed point (Duc, 2020, Inahama, 2022).

A recurrent identification in RDEs is

$y_t-y_s=y'_sX_{s,t}+R^y_{s,t},$4

for equations of the form

$y_t-y_s=y'_sX_{s,t}+R^y_{s,t},$5

The controlled expansion of the solution then takes the explicit form

$y_t-y_s=y'_sX_{s,t}+R^y_{s,t},$6

which shows that the solution is controlled by the driver with Gubinelli derivative $y_t-y_s=y'_sX_{s,t}+R^y_{s,t},$7 (Duc, 2020).

2. Rough integration, composition, and fixed-point calculus

Controlled rough paths are designed so that rough integrals can be reconstructed from local expansions. In the level-2 Hölder setting of (Inahama, 2022), if $y_t-y_s=y'_sX_{s,t}+R^y_{s,t},$8, one defines

$y_t-y_s=y'_sX_{s,t}+R^y_{s,t},$9

Using Chen’s relation and the controlled expansion, the defect satisfies

$y'$0

Since $y'$1, the sewing lemma applies, and the rough integral is

$y'$2

Moreover,

$y'$3

with estimate

$y'$4

where

$y'$5

This is the main sewing-type estimate in that framework (Inahama, 2022).

In the $y'$6-variation/Hölder formulation of (Duc, 2020), Gubinelli’s rough integral is

$y'$7

with estimate

$y'$8

A $y'$9-variation version is also given there (Duc, 2020).

Composition is equally central. In (Inahama, 2022), if $R^y$0 and $R^y$1, then $R^y$2 with

$R^y$3

and

$R^y$4

This is the classical controlled-path chain rule. In (Duc, 2020), if $R^y$5 is controlled with $R^y$6, then $R^y$7 is controlled with

$R^y$8

These two operations—composition and integration—are the ingredients needed for the fixed-point method. In (Inahama, 2022), RDEs are written as

$R^y$9

and are solved by contraction on controlled-path spaces. Global existence and uniqueness hold under the stated boundedness and Lipschitz assumptions, first for bounded globally Lipschitz drift and then for bounded locally Lipschitz drift (Inahama, 2022). In (Duc, 2020), linear diffusion coefficients are treated directly by a controlled-path fixed point, while nonlinear diffusion with unbounded drift is handled via a Doss–Sussmann transform after first analyzing the driftless rough equation (Duc, 2020).

3. Higher-order and combinatorial formulations

For $2\le p<3$0, the classical level-2 picture is no longer sufficient once $2\le p<3$1. In (Boedihardjo et al., 2020), the driving signal is a $2\le p<3$2-Hölder geometric rough path with

$2\le p<3$3

A controlled rough path over $2\le p<3$4 is then a hierarchy

$2\le p<3$5

where for $2\le p<3$6,

$2\le p<3$7

and for $2\le p<3$8,

$2\le p<3$9

Controlledness means

$\alpha\in(1/3,1/2]$0

The resulting Banach space is denoted $\alpha\in(1/3,1/2]$1 (Boedihardjo et al., 2020).

The central result of (Boedihardjo et al., 2020) is a higher-order composition theorem. If $\alpha\in(1/3,1/2]$2 is $\alpha\in(1/3,1/2]$3-Lipschitz in the Stein sense with $\alpha\in(1/3,1/2]$4, then $\alpha\in(1/3,1/2]$5 is again controlled, with transformed coefficients

$\alpha\in(1/3,1/2]$6

The proof relies on the coproduct structure on the truncated tensor algebra and the characterization of geometric rough paths by

$\alpha\in(1/3,1/2]$7

for $\alpha\in(1/3,1/2]$8 (Boedihardjo et al., 2020). This establishes closure under composition below the threshold $\alpha\in(1/3,1/2]$9, identified there as the main missing ingredient in extending controlled rough path theory to arbitrary $y:[0,T]\to\mathbb R^m$0.

A distinct combinatorial development appears in (Cass et al., 2021), which treats weakly geometric rough paths of bounded $y:[0,T]\to\mathbb R^m$1-variation without smooth approximation. There a controlled path $y:[0,T]\to\mathbb R^m$2 is an element of

$y:[0,T]\to\mathbb R^m$3

encoded as a graded element in $y:[0,T]\to\mathbb R^m$4. In coordinates,

$y:[0,T]\to\mathbb R^m$5

The paper’s main controlled-path contribution is an explicit lift of a controlled path to a genuine weakly geometric rough path: $y:[0,T]\to\mathbb R^m$6 or in coordinates,

$y:[0,T]\to\mathbb R^m$7

This yields the genuine lift

$y:[0,T]\to\mathbb R^m$8

That construction supports associativity of the rough integral, the adjunction between pushforwards and pullbacks, and change-of-variables formulas for RDEs on manifolds (Cass et al., 2021).

A further geometric development, but now at the level of the family of all controlled-path spaces, is given in (Varzaneh et al., 2022). For branched rough paths of arbitrary order, the fibers $y:[0,T]\to\mathbb R^m$9 assemble into a continuous field of Banach spaces over the base rough path space. The total space

$X$0

carries an intrinsic tube topology, and in the geometric case

$X$1

is Polish (Varzaneh et al., 2022). This shows that controlled rough paths are naturally organized as Banach fibers varying with the driver, rather than as a single fixed Banach space.

4. Hopf-algebraic unification

A major unifying development is the formulation of controlled rough paths on a general class of combinatorial Hopf algebras (Zhu et al., 27 Sep 2025). The paper works with an $X$2-truncated connected graded Hopf algebra

$X$3

with finite-dimensional homogeneous components, truncated characters, convolution

$X$4

and Chen relation

$X$5

This framework simultaneously encompasses the shuffle Hopf algebra, the Butcher–Connes–Kreimer Hopf algebra, and the Munthe-Kaas–Wright Hopf algebra (Zhu et al., 27 Sep 2025).

In this language, an $X$6-controlled rough path is a path

$X$7

such that for homogeneous $X$8,

$X$9

satisfies

$y':[0,T]\to \mathcal L(\mathbb R^n,\mathbb R^m)$0

The degree-zero component recovers the underlying path: $y':[0,T]\to \mathcal L(\mathbb R^n,\mathbb R^m)$1 This is the exact Hopf-algebraic controlled expansion (Zhu et al., 27 Sep 2025).

The associated Banach norm is defined through the two-parameter remainder: $y':[0,T]\to \mathcal L(\mathbb R^n,\mathbb R^m)$2 and

$y':[0,T]\to \mathcal L(\mathbb R^n,\mathbb R^m)$3

This makes the space $y':[0,T]\to \mathcal L(\mathbb R^n,\mathbb R^m)$4 a Banach space (Zhu et al., 27 Sep 2025).

Smooth functions are treated through a reduced coproduct. For $y':[0,T]\to \mathcal L(\mathbb R^n,\mathbb R^m)$5,

$y':[0,T]\to \mathcal L(\mathbb R^n,\mathbb R^m)$6

and for $y':[0,T]\to \mathcal L(\mathbb R^n,\mathbb R^m)$7,

$y':[0,T]\to \mathcal L(\mathbb R^n,\mathbb R^m)$8

Rough integration is encoded by degree-$y':[0,T]\to \mathcal L(\mathbb R^n,\mathbb R^m)$9 cocycles $X$00. If

$X$01

then sewing yields the rough integral, and the lifted integral is

$X$02

This allows local and global well-posedness of the lifted integral equation

$X$03

A universal limit theorem is also proved in this general setting (Zhu et al., 27 Sep 2025).

This suggests that controlled rough paths are best regarded not as a construction tied specifically to words, signatures, or rooted trees, but as an analytic structure determined by three ingredients: a graded commutative Hopf algebra, characters satisfying Chen’s relation, and degree-raising cocycles $X$04 governing rough integration (Zhu et al., 27 Sep 2025).

5. Variants: Sobolev, càdlàg, and controlled drivers

The classical theory has been extended in several non-equivalent directions.

In the Sobolev setting, (Liu et al., 2020) defines a Sobolev rough path as a path

$X$05

with

$X$06

A controlled Sobolev path above $X$07 is a pair $X$08 with

$X$09

and

$X$10

satisfying

$X$11

This stronger remainder condition is needed to ensure that rough integration preserves the same Sobolev regularity. The corresponding Itô–Lyons map is locally Lipschitz continuous with respect to the initial value, vector field, and driver, measured by the Sobolev rough path distance $X$12 together with the mixed distance $X$13 (Liu et al., 2020).

In the càdlàg setting, (Kwossek et al., 2024) extends controlled rough paths to level-2 càdlàg rough paths $X$14 with $X$15. A controlled path is again defined by

$X$16

but now in càdlàg $X$17-variation spaces: $X$18 The forward rough integral is

$X$19

with remainder estimate

$X$20

The central innovation there is that the coefficient can be a non-anticipative functional on the Banach space of controlled paths: $X$21 This gives a general rough functional differential equation framework that includes classical RDEs, controlled RDEs, and delay equations (Kwossek et al., 2024).

A third extension concerns RDEs driven not by the reference rough path $X$22 itself but by another path $X$23 controlled by $X$24. In (Li et al., 10 Mar 2026), in the level-2 regime $X$25, one considers controlled paths

$X$26

with

$X$27

The rough integral is defined by

$X$28

and an a priori estimate is proved: $X$29 This supports a universal limit theorem for equations of the form

$X$30

driven by controlled rough paths (Li et al., 10 Mar 2026).

6. Applications, extensions, and interpretation

Controlled rough paths serve as the deterministic backbone of several applied and structural developments.

In averaging for slow–fast rough systems, (Inahama, 2022) uses controlled-path spaces to adapt Khas’minskii’s time-discretizing method to rough differential equations. The slow component $X$31 is shown to be a controlled rough path over the slow rough driver $X$32, and the perturbation/stability estimate is formulated directly in controlled norms. The strong averaging theorem proves

$X$33

under the stated assumptions (Inahama, 2022).

In slow–fast large deviations under mixed fractional Brownian motion, (Yang et al., 2024) uses the continuity of the Itô–Lyons map in controlled rough path topology as the deterministic input for a variational weak-convergence argument. The slow equation is posed as a controlled rough differential equation, and the large deviation principle is proved on $X$34 with the stated good rate function (Yang et al., 2024).

In reflected rough differential equations, the supplied data for (Aida, 2016) does not include the paper text, so no paper-specific theorem statements can be extracted. The supplied material only supports the general background statement that controlled paths can be used to prove existence for reflected rough differential equations under weaker assumptions than an earlier Euler-approximation approach, but the exact assumptions and formulas are unavailable from the provided notice (Aida, 2016).

In stochastic control, (Diehl et al., 2013) formulates control problems for rough differential equations driven by geometric rough paths. The paper primarily uses Lyons-style RDEs rather than Gubinelli’s controlled paths, but it explicitly remarks that if one allows the diffusion coefficient to depend on the control, one may need controls “controlled by $X$35 in the Gubinelli sense.” A plausible implication is that controlled rough path regularity can serve as a compatibility condition when the control enters the rough channel (Diehl et al., 2013).

A more abstract mean-field extension is the theory of random controlled rough paths in (Delarue et al., 2022). There the jet of a controlled object is indexed by Lions forests in a coupled Hopf algebra, and rough integration and composition by smooth functions on the Wasserstein space are both shown to be closed on this class. This suggests that the controlled rough path paradigm is flexible enough to accommodate law-dependent expansions, though the combinatorics become substantially more elaborate (Delarue et al., 2022).

A common misconception is that controlled rough paths are only a level-2 device or only a convenient notation for $X$36. The higher-order theory in (Boedihardjo et al., 2020), the arbitrary-$X$37 combinatorial lift in (Cass et al., 2021), and the general Hopf-algebraic formulation in (Zhu et al., 27 Sep 2025) show otherwise. Another common misconception is that controlled rough paths are inseparable from the shuffle algebra of signatures. The Hopf-algebraic and branched formulations show that geometric, branched, and planarly branched theories share the same analytic mechanism once the correct combinatorial algebra is identified (Zhu et al., 27 Sep 2025, Varzaneh et al., 2022).

Taken together, these developments present controlled rough paths as a general analytic principle: a path is controlled when its local behavior can be transported by the increment structure of a rough driver, with a remainder regular enough for sewing. In the classical geometric case this is the Gubinelli decomposition; in branched and Hopf-algebraic settings it becomes a character-based transport identity; in Sobolev and càdlàg settings the ambient topology changes; and in mean-field settings the jet is indexed by Lions forests. The unifying content is the same: rough integration, composition, stability, and RDE well-posedness are consequences of a controlled expansion with a higher-order remainder (Zhu et al., 27 Sep 2025).