Rough path

Last updated

In stochastic analysis, a rough path is a generalization of the notion of smooth path allowing to construct a robust solution theory for controlled differential equations driven by classically irregular signals, for example a Wiener process. The theory was developed in the 1990s by Terry Lyons. [1] [2] [3] Several accounts of the theory are available. [4] [5] [6] [7]

Contents

Rough path theory is focused on capturing and making precise the interactions between highly oscillatory and non-linear systems. It builds upon the harmonic analysis of L.C. Young, the geometric algebra of K.T. Chen, the Lipschitz function theory of H. Whitney and core ideas of stochastic analysis. The concepts and the uniform estimates have widespread application in pure and applied Mathematics and beyond. It provides a toolbox to recover with relative ease many classical results in stochastic analysis (Wong-Zakai, Stroock-Varadhan support theorem, construction of stochastic flows, etc) without using specific probabilistic properties such as the martingale property or predictability. The theory also extends Itô's theory of SDEs far beyond the semimartingale setting. At the heart of the mathematics is the challenge of describing a smooth but potentially highly oscillatory and multidimensional path effectively so as to accurately predict its effect on a nonlinear dynamical system . The Signature is a homomorphism from the monoid of paths (under concatenation) into the grouplike elements of the free tensor algebra. It provides a graduated summary of the path . This noncommutative transform is faithful for paths up to appropriate null modifications. These graduated summaries or features of a path are at the heart of the definition of a rough path; locally they remove the need to look at the fine structure of the path. Taylor's theorem explains how any smooth function can, locally, be expressed as a linear combination of certain special functions (monomials based at that point). Coordinate iterated integrals (terms of the signature) form a more subtle algebra of features that can describe a stream or path in an analogous way; they allow a definition of rough path and form a natural linear "basis" for continuous functions on paths.

Martin Hairer used rough paths to construct a robust solution theory for the KPZ equation. [8] He then proposed a generalization known as the theory of regularity structures [9] for which he was awarded a Fields medal in 2014.

Motivation

Rough path theory aims to make sense of the controlled differential equation

where the control, the continuous path taking values in a Banach space, need not be differentiable nor of bounded variation. A prevalent example of the controlled path is the sample path of a Wiener process. In this case, the aforementioned controlled differential equation can be interpreted as a stochastic differential equation and integration against "" can be defined in the sense of Itô. However, Itô's calculus is defined in the sense of and is in particular not a pathwise definition. Rough paths give an almost sure pathwise definition of stochastic differential equations. The rough path notion of solution is well-posed in the sense that if is a sequence of smooth paths converging to in the -variation metric (described below), and

then converges to in the -variation metric. This continuity property and the deterministic nature of solutions makes it possible to simplify and strengthen many results in Stochastic Analysis, such as the Freidlin-Wentzell's Large Deviation theory [10] as well as results about stochastic flows.

In fact, rough path theory can go far beyond the scope of Itô and Stratonovich calculus and allows to make sense of differential equations driven by non-semimartingale paths, such as Gaussian processes and Markov processes. [11]

Definition of a rough path

Rough paths are paths taking values in the truncated free tensor algebra (more precisely: in the free nilpotent group embedded in the free tensor algebra), which this section now briefly recalls. The tensor powers of , denoted , are equipped with the projective norm (see Topological tensor product, note that rough path theory in fact works for a more general class of norms). Let be the truncated tensor algebra

where by convention .

Let be the simplex . Let . Let and be continuous maps . Let denote the projection of onto -tensors and likewise for . The -variation metric is defined as

where the supremum is taken over all finite partitions of .

A continuous function is a -geometric rough path if there exists a sequence of paths with finite total variation such that

converges in the -variation metric to as . [12]

Universal limit theorem

A central result in rough path theory is Lyons' Universal Limit theorem. [1] One (weak) version of the result is the following: Let be a sequence of paths with finite total variation and let

denote the rough path lift of .

Suppose that converges in the -variation metric to a -geometric rough path as . Let be functions that have at least bounded derivatives and the -th derivatives are -Hölder continuous for some . Let be the solution to the differential equation

and let be defined as

Then converges in the -variation metric to a -geometric rough path .

Moreover, is the solution to the differential equation

driven by the geometric rough path .

Concisely, the theorem can be interpreted as saying that the solution map (aka the Itô-Lyons map) of the RDE is continuous (and in fact locally lipschitz) in the -variation topology. Hence rough paths theory demonstrates that by viewing driving signals as rough paths, one has a robust solution theory for classical stochastic differential equations and beyond.

Examples of rough paths

Brownian motion

Let be a multidimensional standard Brownian motion. Let denote the Stratonovich integration. Then

is a -geometric rough path for any . This geometric rough path is called the Stratonovich Brownian rough path.

Fractional Brownian motion

More generally, let be a multidimensional fractional Brownian motion (a process whose coordinate components are independent fractional Brownian motions) with . If is the -th dyadic piecewise linear interpolation of , then

converges almost surely in the -variation metric to a -geometric rough path for . [13] This limiting geometric rough path can be used to make sense of differential equations driven by fractional Brownian motion with Hurst parameter . When , it turns out that the above limit along dyadic approximations does not converge in -variation. However, one can of course still make sense of differential equations provided one exhibits a rough path lift, existence of such a (non-unique) lift is a consequence of the Lyons–Victoir extension theorem.

Non-uniqueness of enhancement

In general, let be a -valued stochastic process. If one can construct, almost surely, functions so that

is a -geometric rough path, then is an enhancement of the process . Once an enhancement has been chosen, the machinery of rough path theory will allow one to make sense of the controlled differential equation

for sufficiently regular vector fields

Note that every stochastic process (even if it is a deterministic path) can have more than one (in fact, uncountably many) possible enhancements. [14] Different enhancements will give rise to different solutions to the controlled differential equations. In particular, it is possible to enhance Brownian motion to a geometric rough path in a way other than the Brownian rough path. [15] This implies that the Stratonovich calculus is not the only theory of stochastic calculus that satisfies the classical product rule

In fact any enhancement of Brownian motion as a geometric rough path will give rise a calculus that satisfies this classical product rule. Itô calculus does not come directly from enhancing Brownian motion as a geometric rough path, but rather as a branched rough path.

Applications in stochastic analysis

Stochastic differential equations driven by non-semimartingales

Rough path theory allows to give a pathwise notion of solution to (stochastic) differential equations of the form

provided that the multidimensional stochastic process can be almost surely enhanced as a rough path and that the drift and the volatility are sufficiently smooth (see the section on the Universal Limit Theorem).

There are many examples of Markov processes, Gaussian processes, and other processes that can be enhanced as rough paths. [16]

There are, in particular, many results on the solution to differential equation driven by fractional Brownian motion that have been proved using a combination of Malliavin calculus and rough path theory. In fact, it has been proved recently that the solution to controlled differential equation driven by a class of Gaussian processes, which includes fractional Brownian motion with Hurst parameter , has a smooth density under the Hörmander's condition on the vector fields. [17] [18]

Freidlin–Wentzell's large deviation theory

Let denote the space of bounded linear maps from a Banach space to another Banach space .

Let be a -dimensional standard Brownian motion. Let and be twice-differentiable functions and whose second derivatives are -Hölder for some .

Let be the unique solution to the stochastic differential equation

where denotes Stratonovich integration.

The Freidlin Wentzell's large deviation theory aims to study the asymptotic behavior, as , of for closed or open sets with respect to the uniform topology.

The Universal Limit Theorem guarantees that the Itô map sending the control path to the solution is a continuous map from the -variation topology to the -variation topology (and hence the uniform topology). Therefore, the Contraction principle in large deviations theory reduces Freidlin–Wentzell's problem to demonstrating the large deviation principle for in the -variation topology. [10]

This strategy can be applied to not just differential equations driven by the Brownian motion but also to the differential equations driven any stochastic processes which can be enhanced as rough paths, such as fractional Brownian motion.

Stochastic flow

Once again, let be a -dimensional Brownian motion. Assume that the drift term and the volatility term has sufficient regularity so that the stochastic differential equation

has a unique solution in the sense of rough path. A basic question in the theory of stochastic flow is whether the flow map exists and satisfy the cocyclic property that for all ,

outside a null set independent of .

The Universal Limit Theorem once again reduces this problem to whether the Brownian rough path exists and satisfies the multiplicative property that for all ,

outside a null set independent of , and .

In fact, rough path theory gives the existence and uniqueness of not only outside a null set independent of , and but also of the drift and the volatility .

As in the case of Freidlin–Wentzell theory, this strategy holds not just for differential equations driven by the Brownian motion but to any stochastic processes that can be enhanced as rough paths.

Controlled rough path

Controlled rough paths, introduced by M. Gubinelli, [5] are paths for which the rough integral

can be defined for a given geometric rough path .

More precisely, let denote the space of bounded linear maps from a Banach space to another Banach space .

Given a -geometric rough path

on , a -controlled path is a function such that and that there exists such that for all and ,

and

Example: Lip(γ) function

Let be a -geometric rough path satisfying the Hölder condition that there exists , for all and all ,

where denotes the -th tensor component of . Let . Let be an -times differentiable function and the -th derivative is Hölder, then

is a -controlled path.

Integral of a controlled path is a controlled path

If is a -controlled path where , then

is defined and the path

is a -controlled path.

Solution to controlled differential equation is a controlled path

Let be functions that has at least derivatives and the -th derivatives are -Hölder continuous for some . Let be the solution to the differential equation

Define

where denotes the derivative operator, then

is a -controlled path.

Signature

Let be a continuous function with finite total variation. Define

The signature of a path is defined to be .

The signature can also be defined for geometric rough paths. Let be a geometric rough path and let be a sequence of paths with finite total variation such that

converges in the -variation metric to . Then

converges as for each . The signature of the geometric rough path can be defined as the limit of as .

The signature satisfies Chen's identity, [19] that

for all .

Kernel of the signature transform

The set of paths whose signature is the trivial sequence, or more precisely,

can be completely characterized using the idea of tree-like path.

A -geometric rough path is tree-like if there exists a continuous function such that and for all and all ,

where denotes the -th tensor component of .

A geometric rough path satisfies if and only if is tree-like. [20] [21]

Given the signature of a path, it is possible to reconstruct the unique path that has no tree-like pieces. [22] [23]

Infinite dimensions

It is also possible to extend the core results in rough path theory to infinite dimensions, providing that the norm on the tensor algebra satisfies certain admissibility condition. [24]

Related Research Articles

<span class="mw-page-title-main">Independence (probability theory)</span> When the occurrence of one event does not affect the likelihood of another

Independence is a fundamental notion in probability theory, as in statistics and the theory of stochastic processes. Two events are independent, statistically independent, or stochastically independent if, informally speaking, the occurrence of one does not affect the probability of occurrence of the other or, equivalently, does not affect the odds. Similarly, two random variables are independent if the realization of one does not affect the probability distribution of the other.

In mathematics, the tensor product of two vector spaces V and W is a vector space to which is associated a bilinear map that maps a pair to an element of denoted

In mathematics, the Lp spaces are function spaces defined using a natural generalization of the p-norm for finite-dimensional vector spaces. They are sometimes called Lebesgue spaces, named after Henri Lebesgue, although according to the Bourbaki group they were first introduced by Frigyes Riesz.

<span class="mw-page-title-main">Floor and ceiling functions</span> Nearest integers from a number

In mathematics and computer science, the floor function is the function that takes as input a real number x, and gives as output the greatest integer less than or equal to x, denoted x or floor(x). Similarly, the ceiling function maps x to the least integer greater than or equal to x, denoted x or ceil(x).

<span class="mw-page-title-main">Fokker–Planck equation</span> Partial differential equation

In statistical mechanics and information theory, the Fokker–Planck equation is a partial differential equation that describes the time evolution of the probability density function of the velocity of a particle under the influence of drag forces and random forces, as in Brownian motion. The equation can be generalized to other observables as well. The Fokker-Planck equation has multiple applications in information theory, graph theory, data science, finance, economics etc.

<span class="mw-page-title-main">Green's theorem</span> Theorem in calculus relating line and double integrals

In vector calculus, Green's theorem relates a line integral around a simple closed curve C to a double integral over the plane region D bounded by C. It is the two-dimensional special case of Stokes' theorem.

<span class="mw-page-title-main">Martingale (probability theory)</span> Model in probability theory

In probability theory, a martingale is a sequence of random variables for which, at a particular time, the conditional expectation of the next value in the sequence is equal to the present value, regardless of all prior values.

In mathematics, the GrassmannianGr(k, V) is a space that parameterizes all k-dimensional linear subspaces of the n-dimensional vector space V. For example, the Grassmannian Gr(1, V) is the space of lines through the origin in V, so it is the same as the projective space of one dimension lower than V.

<span class="mw-page-title-main">Tropical geometry</span> Skeletonized version of algebraic geometry

In mathematics, tropical geometry is the study of polynomials and their geometric properties when addition is replaced with minimization and multiplication is replaced with ordinary addition:

In abstract algebra, in particular in the theory of nondegenerate quadratic forms on vector spaces, the structures of finite-dimensional real and complex Clifford algebras for a nondegenerate quadratic form have been completely classified. In each case, the Clifford algebra is algebra isomorphic to a full matrix ring over , , or , or to a direct sum of two copies of such an algebra, though not in a canonical way. Below it is shown that distinct Clifford algebras may be algebra-isomorphic, as is the case of and , which are both isomorphic as rings to the ring of two-by-two matrices over the real numbers.

In mathematics, a pairing function is a process to uniquely encode two natural numbers into a single natural number.

A stochastic differential equation (SDE) is a differential equation in which one or more of the terms is a stochastic process, resulting in a solution which is also a stochastic process. SDEs have many applications throughout pure mathematics and are used to model various behaviours of stochastic models such as stock prices, random growth models or physical systems that are subjected to thermal fluctuations.

<span class="mw-page-title-main">Baker's map</span> Chaotic map from the unit square into itself

In dynamical systems theory, the baker's map is a chaotic map from the unit square into itself. It is named after a kneading operation that bakers apply to dough: the dough is cut in half, and the two halves are stacked on one another, and compressed.

In the theory of stochastic processes, filtering describes the problem of determining the state of a system from an incomplete and potentially noisy set of observations. While originally motivated by problems in engineering, filtering found applications in many fields from signal processing to finance.

In multilinear algebra, the tensor rank decomposition or the decomposition of a tensor is the decomposition of a tensor in terms of a sum of minimum tensors. This is an open problem.

The Brownian motion models for financial markets are based on the work of Robert C. Merton and Paul A. Samuelson, as extensions to the one-period market models of Harold Markowitz and William F. Sharpe, and are concerned with defining the concepts of financial assets and markets, portfolios, gains and wealth in terms of continuous-time stochastic processes.

In cryptography, Learning with errors (LWE) is a mathematical problem that is widely used in cryptography to create secure encryption algorithms. It is based on the idea of representing secret information as a set of equations with errors. In other words, LWE is a way to hide the value of a secret by introducing noise to it. In more technical terms, it refers to the computational problem of inferring a linear -ary function over a finite ring from given samples some of which may be erroneous. The LWE problem is conjectured to be hard to solve, and thus to be useful in cryptography.

In mathematics, a cyclic polytope, denoted C(n,d), is a convex polytope formed as a convex hull of n distinct points on a rational normal curve in Rd, where n is greater than d. These polytopes were studied by Constantin Carathéodory, David Gale, Theodore Motzkin, Victor Klee, and others. They play an important role in polyhedral combinatorics: according to the upper bound theorem, proved by Peter McMullen and Richard Stanley, the boundary Δ(n,d) of the cyclic polytope C(n,d) maximizes the number fi of i-dimensional faces among all simplicial spheres of dimension d − 1 with n vertices.

In the geometry of numbers, the Klein polyhedron, named after Felix Klein, is used to generalize the concept of continued fractions to higher dimensions.

In differential geometry, a fibered manifold is surjective submersion of smooth manifolds YX. Locally trivial fibered manifolds are fiber bundles. Therefore, a notion of connection on fibered manifolds provides a general framework of a connection on fiber bundles.

References

  1. 1 2 Lyons, Terry (1998). "Differential equations driven by rough signals". Revista Matemática Iberoamericana. 14 (2): 215–310. doi:10.4171/RMI/240. ISSN   0213-2230. S2CID   59183294. Zbl   0923.34056. Wikidata   Q55933523.
  2. Lyons, Terry; Qian, Zhongmin (2002). System Control and Rough Paths. Oxford Mathematical Monographs. Oxford: Clarendon Press. doi:10.1093/acprof:oso/9780198506485.001.0001. ISBN   9780198506485. Zbl   1029.93001.
  3. Lyons, Terry; Caruana, Michael; Levy, Thierry (2007). Differential equations driven by rough paths, vol. 1908 of Lecture Notes in Mathematics. Springer.
  4. Lejay, A. (2003). "An Introduction to Rough Paths". Séminaire de Probabilités XXXVII. Lecture Notes in Mathematics. Vol. 1832. pp. 1–59. doi:10.1007/978-3-540-40004-2_1. ISBN   978-3-540-20520-3. S2CID   12401468.
  5. 1 2 Gubinelli, Massimiliano (November 2004). "Controlling rough paths". Journal of Functional Analysis . 216 (1): 86–140. doi:10.1016/J.JFA.2004.01.002. ISSN   0022-1236. S2CID   119717942. Zbl   1058.60037. Wikidata   Q56689330.
  6. Friz, Peter K.; Victoir, Nicolas (2010). Multidimensional Stochastic Processes as Rough Paths: Theory and Applications. Cambridge Studies in Advanced Mathematics. Cambridge University Press.
  7. Friz, Peter K.; Hairer, Martin (2014). A Course on Rough Paths, with an introduction to regularity structures. Springer.
  8. Hairer, Martin (7 June 2013). "Solving the KPZ equation". Annals of Mathematics . 178 (2): 559–664. arXiv: 1109.6811 . doi:10.4007/ANNALS.2013.178.2.4. ISSN   0003-486X. JSTOR   23470800. MR   3071506. S2CID   119247908. Zbl   1281.60060. Wikidata   Q56689331.
  9. Hairer, Martin (2014). "A theory of regularity structures". Inventiones Mathematicae. 198 (2): 269–504. arXiv: 1303.5113 . Bibcode:2014InMat.198..269H. doi:10.1007/s00222-014-0505-4. S2CID   119138901.
  10. 1 2 Ledoux, Michel; Qian, Zhongmin; Zhang, Tusheng (December 2002). "Large deviations and support theorem for diffusion processes via rough paths". Stochastic Processes and their Applications . 102 (2): 265–283. doi:10.1016/S0304-4149(02)00176-X. ISSN   1879-209X. Zbl   1075.60510. Wikidata   Q56689332.
  11. Friz, Peter K.; Victoir, Nicolas (2010). Multidimensional Stochastic Processes as Rough Paths: Theory and Applications (Cambridge Studies in Advanced Mathematics ed.). Cambridge University Press.
  12. Lyons, Terry; Qian, Zhongmin (2002). System Control and Rough Paths. Oxford Mathematical Monographs. Oxford: Clarendon Press. doi:10.1093/acprof:oso/9780198506485.001.0001. ISBN   9780198506485. Zbl   1029.93001.
  13. Coutin, Laure; Qian, Zhongmin (2002). "Stochastic analysis, rough path analysis and fractional Brownian motions". Probability Theory and Related Fields. 122: 108–140. doi: 10.1007/s004400100158 . S2CID   120581658.
  14. Lyons, Terry; Victoir, Nicholas (2007). "An extension theorem to rough paths". Annales de l'Institut Henri Poincaré C. 24 (5): 835–847. Bibcode:2007AIHPC..24..835L. doi: 10.1016/j.anihpc.2006.07.004 .
  15. Friz, Peter; Gassiat, Paul; Lyons, Terry (2015). "Physical Brownian motion in a magnetic field as a rough path". Transactions of the American Mathematical Society. 367 (11): 7939–7955. arXiv: 1302.2531 . doi:10.1090/S0002-9947-2015-06272-2. S2CID   59358406.
  16. Friz, Peter K.; Victoir, Nicolas (2010). Multidimensional Stochastic Processes as Rough Paths: Theory and Applications (Cambridge Studies in Advanced Mathematics ed.). Cambridge University Press.
  17. Cass, Thomas; Friz, Peter (2010). "Densities for rough differential equations under Hörmander's condition". Annals of Mathematics. 171 (3): 2115–2141. arXiv: 0708.3730 . doi:10.4007/annals.2010.171.2115. S2CID   17276607.
  18. Cass, Thomas; Hairer, Martin; Litterer, Christian; Tindel, Samy (2015). "Smoothness of the density for solutions to Gaussian rough differential equations". The Annals of Probability. 43: 188–239. arXiv: 1209.3100 . doi:10.1214/13-AOP896. S2CID   17308794.
  19. Chen, Kuo-Tsai (1954). "Iterated Integrals and Exponential Homomorphisms". Proceedings of the London Mathematical Society. s3-4: 502–512. doi:10.1112/plms/s3-4.1.502.
  20. Hambly, Ben; Lyons, Terry (2010). "Uniqueness for the signature of a path of bounded variation and the reduced path group". Annals of Mathematics. 171: 109–167. arXiv: math/0507536 . doi:10.4007/annals.2010.171.109. S2CID   15915599.
  21. Boedihardjo, Horatio; Geng, Xi; Lyons, Terry; Yang, Danyu (2016). "The signature of a rough path: Uniqueness". Advances in Mathematics . 293: 720–737. arXiv: 1406.7871 . doi: 10.1016/j.aim.2016.02.011 . S2CID   3634324.
  22. Lyons, Terry; Xu, Weijun (2018). "Inverting the signature of a path". Journal of the European Mathematical Society . 20 (7): 1655–1687. arXiv: 1406.7833 . doi:10.4171/JEMS/796. S2CID   67847036.
  23. Geng, Xi (2016). "Reconstruction for the Signature of a Rough Path". Proceedings of the London Mathematical Society. 114 (3): 495–526. arXiv: 1508.06890 . doi:10.1112/plms.12013. S2CID   3641736.
  24. Cass, Thomas; Driver, Bruce; Lim, Nengli; Litterer, Christian. "On the integration of weakly geometric rough paths". Journal of the Mathematical Society of Japan.