![MathJax Logo](/templates/jsp/_style2/_tandf/pb2/images/math-jax.gif)
Abstract
The purpose of this paper is to establish a stochastic differential equation for the Donsker delta measure of the solution of a McKean–Vlasov (mean-field) stochastic differential equation.
If the Donsker delta measure is absolutely continuous with respect to Lebesgue measure, then its Radon–Nikodym derivative is called the Donsker delta function. In that case it can be proved that the local time of such a process is simply the integral with respect to time of the Donsker delta function. Therefore we also get an equation for the local time of such a process.
For some particular McKean–Vlasov processes, we find explicit expressions for their Donsker delta functions and hence for their local times.
1. Introduction
The Donsker delta function of a random variable or a stochastic process arises in many studies, including quantum mechanical particles on a circle [Citation7], financial markets with insider trading as in [Citation10] and in [Citation3] for financial markets with singular drift. It has also been used as a tool to determine explicit formulae for replicating portfolios in complete and incomplete markets [Citation9].
Moreover, the Donsker delta function is also of interest because it can be regarded as a time derivative of the local time. Therefore, explicit expressions for the Donsker delta function lead to explicit formulae of the local time.
For example, if we let B be a Brownian motion defined on a filtered probability space , then the Donsker delta function
of a Brownian motion B at the point x can be regarded as the time derivative of the local time
of B. More precisely, we have
Such an integral exists as an element of the Hida space
of stochastic distributions, see Section 2.3.
In [Citation6], the authors use white noise theory to obtain an explicit solution formula for a general stochastic differential equation (SDE), and this is used to find an expression for the Donsker delta function for the solution of an SDE. Subsequently this was also extended to SDEs driven by Lévy noise in [Citation8].
The main result of the current paper is that the Donsker delta measure of a McKean–Vlasov process (see below) always satisfies a certain Fokker–Planck type SPDE in the sense of distributions. Moreover, we use this to find explicit formulae for the Donsker delta functions for McKean–Vlasov processes, and hence their local times, in specific cases.
Let be the solution of a McKean–Vlasov SDE, i.e. a mean-field stochastic differential equation, of the form (using matrix notation),
We call X a McKean–Vlasov process.
Here the σ-algebra denotes the filtration generated by Z and
, Z is a random variable which is independent of the σ-algebra generated by
and such that
Definition 1.1
Define to be regular conditional distribution of
given
generated by the Brownian motion B. This means that
is a Borel probability measure on
for all
and
for all functions g such that
.
Since we consider only a one-dimensional Brownian motion , we will show that the regular conditional distribution of
given the filtration
can be identified with the Donsker delta measure in the sense of distribution. See details in Section 3.1
2. Preliminaries
In this section, we review some basic notions and results that will be used throughout this work.
2.1. Radon measures
A Radon measure on is a Borel measure which is finite on compact sets, outer regular on all Borel sets and inner regular on all open sets. In particular, all Borel probability measures on
are Radon measures.
In the following, we let
be the set of deterministic Radon measures.
be the uniform closure of the space
of continuous functions with compact support.
If we equip with the total variation norm
, then
becomes a Banach space, and it is the dual of
. See Chapter 7 in Folland [Citation5] for more information.
If is a finite measure, we define
(1)
(1) to be the Fourier transform of μ at y.
In particular, if is absolutely continuous with respect to Lebesgue measure
with Radon–Nikodym derivative
, so that
with
, we define the Fourier transform of m at y, denoted by
or
, by
We let
denote the set of all random measures
such that
for each given
.
2.2. The Schwartz space of tempered distributions
We recall now some notions from white noise analysis.
be the Schwartz space of rapidly decreasing smooth real functions on
. It is a Fréchet space with respect to the family of seminorms:
where
,
is a multi-index with
and
is the space of tempered distributions. It is the dual of
.
2.3. The Hida space ![](//:0)
of stochastic distributions
We restrict ourselves to the white noise probability space , where
is the Borel σ-algebra and the probability P is the probability measure on
defined in virtue of the Bochner–Minlos–Sazonov theorem.
Let denote the set of all finite multi-indices
,
, of non-negative integers
.
(2)
(2) If
, we put
(3)
(3) The family
constitutes an orthogonal basis of
.
is the Hilbert space consisting of all
such that
for numbers
.
The space
equipped with the projective topology is the Hida space of stochastic test functions.
is the Hilbert space consisting of all formal sums
equipped with the norm
The space
equipped with the inductive topology is the Hida space of stochastic distributions. It can be regarded as the dual of
.
2.4. The Donsker delta function
We now recall some basic definitions:
Definition 2.1
Let be a random variable which also belongs to the Hida space
of stochastic distributions. Then a continuous function
(4)
(4) is called a Donsker delta function of Y if it has the property that
(5)
(5) for all (measurable)
such that the integral converges in
The Donsker delta function is related to the regular conditional distribution. The connection is the following: the regular conditional distribution with respect to the σ-algebra of a given real random variable Y, denoted by
, is defined by the following properties:
For any Borel set
,
is a version of
.
For each fixed
,
is a probability measure on the Borel subsets of
.
From the required properties of , we get the following formula:
(6)
(6)
Definition 2.2
We call the Donsker delta measure of the random variable Y and denote it by
.
Comparing this with the definition of the Donsker delta function, we obtain the following representation of the regular conditional distribution:
Lemma 2.3
Suppose is absolutely continuous with respect to Lebesgue measure
on
and that Y is measurable with respect to
. Then the Donsker delta function of Y,
is the Radon–Nikodym derivative of
with respect to Lebesgue measure
, i.e.
(7)
(7)
We will prove in Theorem 3.3 that the Donsker delta function can be regarded as a stochastic distribution in , satisfying a Fokker–Planck type SPDE in the sense of distributions. It can also be represented as an element of the Hida stochastic distribution space
, and as such it can in some cases be expressed explicitly in terms of Wick calculus. For example, if
, we have
(8)
(8) where ⋄ denotes Wick multiplication and
denotes Wick exponential. Note that even though the Donsker delta function can only be represented as a distribution, its conditional expectation can be a real-valued stochastic process. For example, for t<T we have
(9)
(9) For more examples, we refer to e.g. [Citation1] or [Citation9].
3. The Donsker delta equation for McKean–Vlasov processes
3.1. The general multidimensional Fokker–Planck equation
To explain the background for this section, let us recall the general multidimensional situation studied in [Citation2], where is a McKean–Vlasov diffusion, of the form (using matrix notation),
(10)
(10) where B is a multi-dimensional Brownian motion.
Here Z is a random variable which is independent of the σ-algebra generated by and such that
Define the σ-algebra
to be the filtration generated by Z and
.
Let denote the set of all Borel measures on
. We assume that the coefficients
and
are bounded and
-predictable processes for all
, and that b and σ are continuous with respect to t and x for all μ.
One can check that under some assumptions, such as Lipschitz and linear growth conditions, there exists a unique solution of Equation (Equation10(10)
(10) ).
Definition 3.1
Fix one of the Brownian motions, say , with filtration
. We define
to be regular conditional distribution of
given
. This means that
is a Borel probability measure on
for all
and
(11)
(11) for all functions g such that
.
The following version of the stochastic Fokker–Planck integro-differential equation for the conditional law for McKean–Vlasov jump diffusions was proved by Agram and Øksendal [Citation2]. For simplicity we consider only the case without jumps here.
Theorem 3.2
Conditional stochastic Fokker–Planck equation [Citation2]
Let be as in (Equation10
(10)
(10) ) with
and let
be the regular conditional distribution of
given
.
Then for a.a. the conditional law
and it satisfies the following SPDE (in the sense of distributions):
(12)
(12) Here
are the integro-differential operator and the differential operator which are given respectively by
(13)
(13) and
(14)
(14) In the above
denote
and
respectively, in the sense of distributions, and
, and similarly with the other terms.
3.2. The Fokker–Planck equation for the Donsker measure
In [Citation2], the theorem above was proved under the assumption that . However, the proof also works if d = 1 and
. Note that in this case, since
is
-measurable, the identity (Equation11
(11)
(11) ) states that
(15)
(15) for all functions g such that
.
In particular, if we choose d = 1 in the above we get that the conditional law coincides with the Donsker measure, i.e.
(16)
(16) Therefore we get the following Fokker–Planck equation for the Donsker measure:
Theorem 3.3
Assume that is as in (Equation10
(10)
(10) ), but with d = 1.
Then the Donsker delta measure satisfies the following equation (in the sense of distribution):
(17)
(17) where
and
.
4. Local time
In this section, we first recall the definition of local time of a stochastic process :
Definition 4.1
The local time of
at the point y and at time t is defined by
where λ denotes Lebesgue measure on
and the limit is in
.
In the white noise context, the local time can be represented as the integral of the Donsker delta function. More precisely, we have the following result:
Theorem 4.2
The local time of X at the point x and the time t is given by
(18)
(18) where the integration takes place in
(or in
for each ω).
Proof.
For completeness, we give the proof.
By definition of the local time and the Donsker delta function, we have
because the function
is continuous in
(and in
).
Remark 4.3
Note that even though we in general can only say that ,
usually exists as a real-valued stochastic process.
5. Explicit solutions
In this section, we find explicitly the Donsker delta function for some particular McKean–Vlasov processes and accordingly their local time.
Suppose that is absolutely continuous, i.e.
(19)
(19) Then (Equation10
(10)
(10) ) gets the form
(20)
(20) where
and (Equation17
(17)
(17) ) becomes a stochastic partial differential equation (SPDE), as follows:
Theorem 5.1
Suppose (Equation19(19)
(19) ) holds. Then the Donsker delta function
is the solution in
of the following SPDE:
(21)
(21)
(22)
(22)
5.1. Brownian motion
Consider the special case when . Then b = 0 and
and Equation (Equation17
(17)
(17) ) becomes
(23)
(23)
(24)
(24) We can easily verify by Wick calculus that a solution in
of Equation (Equation23
(23)
(23) ) is
(25)
(25) which is in agreement with (Equation8
(8)
(8) ). The details are as follows:
Try
Then
and
and
Collecting the terms we see that
satisfies the Fokker–Planck equation (Equation23
(23)
(23) ) for the conditional law of
.
From white noise theory, we know that
for all random variables X, Y with a finite expectation (independent or not). From this we see that
(26)
(26) In particular, if
(constant)
a.e., then
(27)
(27) which has a singularity at x = z.
5.2. Coefficients not depending on x
The next result shows that, under some conditions, the Donsker delta function can be an ordinary function if the initial value has a density.
Theorem 5.2
Assume that is the solution of the following McKean–Vlasov equation:
(28)
(28) where the coefficients
and
do not depend on x. Suppose that
is a random variable (independent of B) with density
(29)
(29)
(1) | Define | ||||
(2) | The solution |
Proof.
We show that
satisfies Equation (Equation21
(21)
(21) ).
By the Ito formula, we have
(32)
(32) Since
(33)
(33) we see that Equation (Equation32
(32)
(32) ) can be written as
(34)
(34) which is the same as Equation (Equation21
(21)
(21) ).
Since
we conclude by uniqueness that
for all t.
This follows from the definition of the Donsker delta function.
5.2.1. Constant coefficients
As a special case of the case above, suppose that
(35)
(35) where α and β are constants. Then by Theorem 5.2, the Donsker delta function is
(36)
(36)
5.3. Mean-field geometric Brownian motion
Suppose that is a McKean–Vlasov process of the form
(37)
(37) We call this a mean-field geometric Brownian motion. For such processes, we have:
Theorem 5.3
(i) | The Donsker delta function | ||||
(ii) | The solution |
Proof.
The corresponding Fokker–Planck equation for the Donsker delta function
is
(41)
(41) This is a stochastic partial differential equation in
It seems difficult to find directly an explicit solution of this equation. However, we can find the solution
by proceeding as follows:
The solution of (Equation37
(37)
(37) ) is
where
By Theorem 5.2, we know that
where
By definition we have
With
this gives
Hence, substituting
,
From this we deduce that
(42)
(42) is the Donsker delta function of
.
This part follows by the definition of the Donsker delta function.
5.4. An example related to the Burgers equation
Suppose the McKean–Vlasov equation has the form
(43)
(43) where
Then the corresponding FP equation for the Donsker function is
(44)
(44)
(45)
(45) This is a stochastic Burgers equation. It is well known that by using the Cole–Hopf transformation the equation can be transformed into the classical heat equation. The details are as follows: if we introduce a new function
such that
(46)
(46) then we see that the Burgers equation (Equation44
(44)
(44) ) becomes the following equation in ψ:
(47)
(47) Integrating with respect to x this gives
(48)
(48) Now define the function
by
(49)
(49) for some constant γ. Then in terms of φ the above equation gets the form
(50)
(50) This simplifies to
(51)
(51) If we choose
(52)
(52) the equation for φ reduces to the (linear) stochastic heat equation
(53)
(53)
(54)
(54) or, using Ito differential notation,
(55)
(55)
(56)
(56) To find an expression for the solution of (Equation53
(53)
(53) ), define an auxiliary process
by
(57)
(57) where
is an auxiliary Brownian motion with law
and independent of B. Then by the Feynman–Kac formula
(58)
(58) where
denotes expectation with respect to
and
, solves Equation (Equation53
(53)
(53) ). Going back to m, we get
(59)
(59) In particular, setting t = 0 we get
(60)
(60) from which we deduce that
(61)
(61) We summarize what we have proved as follows:
Theorem 5.4
(1) | The Donsker delta function | ||||
(2) | The solution |
5.5. A solution approach based on Laplace and Fourier transforms
Consider the Fokker–Planck equation, with ,
(66)
(66) for the McKean–Vlasov equation (Equation35
(35)
(35) ). If
are constants, this becomes
(67)
(67) If
, the equation can be written as
(68)
(68) Let
(69)
(69) and
(70)
(70) Then
(71)
(71) and
(72)
(72) and
(73)
(73) Hence, applying the Laplace and Fourier transform to (Equation68
(68)
(68) ), we get
or
or
(74)
(74) Put
and
Taking inverse Laplace transform, we get
where
Recall that
(75)
(75) Hence
Therefore
and (Equation75
(75)
(75) ) can be written as
Taking inverse Fourier transform we get, with
We have proved the following:
Theorem 5.5
Suppose α and β are constants and that the Donsker delta measure is absolutely continuous with respect to Lebesgue measure. Then the Donsker delta function of the corresponding McKean–Vlasov process is a solution in
of the following stochastic Volterra equation:
where
Remark 5.6
If we get
For comparison, recall that the density of Brownian motion at t, x (when starting at
) is
Acknowledgments
We are grateful to Frank Proske for helpful comments.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Additional information
Funding
References
- K. Aase, B. Øksendal, and J. Ubøe, Using the Donsker delta function to compute hedging strategies, Potential Anal. 14(4) (2001), pp. 351–374.
- N. Agram and B. Øksendal, Stochastic Fokker–Planck PIDE for conditional McKean–Vlasov jump diffusions and applications to optimal control. To appear in SICON. arXiv preprint arXiv:2110.02193v3, 2021.
- N. Agram and B. Øksendal, A financial market with singular drift and no arbitrage, Math. Financ. Econ. 15(3) (2021), pp. 477–500.
- B.L. Breiman, Probability, Addison-Wesley, 1968.
- G.B. Folland, Real Analysis: Modern Techniques and Their Applications, Wiley, 1984.
- A. Lanconelli and F. Proske, On explicit strong solution of Itô-SDE's and the Donsker delta function, Infin. Dimens. Anal. Quantum Probab. Relat. Top. 7(03) (2004), pp. 437–447.
- A. Lascheck, P. Leukert, L. Streit, and W. Westerkamp, More about Donsker's delta function. arXiv preprint math-ph/0303011, 2003.
- T. Meyer-Brandis and F. Proske, On the existence and explicit representability of strong solutions of Lévy noise driven SDEs with irregular coefficients, Commun. Math. Sci. 4(1) (2006), pp. 129–154.
- G. Di Nunno, B. Øksendal, and F. Proske, (Eds.) Malliavin Calculus for Lévy Processes with Applications to Finance, Springer Berlin Heidelberg, 2009.
- B. Øksendal and E. Engen Røse, A white noise approach to insider trading, in Let Us Use White Noise, 2017, pp. 191–203.