CONTROL DIFFUSION PROCESSES WITH LIPSCHITZ CONTINUITY OF DRIFTS

Komang Dharmawan February 2, 2013

Abstract Control diusion processes has been found in a wide eld of applications as in stochastic optimal control and in mathematical nance via the theory of hedging and nonlinear pricing theory for imperfect markets. In this paper we discuss the control diusion process with time and space dependent coecients and local Lipss,,u chitz continuity of the drift. The results show that the controlled process Xt is independent of control u for a constant.

Keywords: Stochastic Dierential Equations, Lipschitz continuity, Control Diusion Process

Introduction

Given a bounded Borel subset U Rn , we denote by U a set of progressively measurable processes u = (ut , t 0) dened on (, F, F, P) such that P(ut U ) = 1 for all t 0. The elements of U are called admissible control processes. For each control process (ut ) U, we consider a stochastic dierential equation, { dXts,,u = b(t, Xts,,u , ut )dt + (t, Xts,,u , ut )dWt , t s, (1.1) Xs = where Xts,,u Rd , and b : R+ Rd U Rd , and : R+ Rd U Rdn are assigned Lipschitz continuous functions for each u Rn . We interpret Xt = Xt () as the state of the system at time t. By a pathwise solution of this equation, we mean an (Ft )-adapted continuous stochastic process Xts,x,u satisfying t t s,,u s,,u s,,u (1.2) (r, Xr , ur )dWr , 0 s t. Xt =+ b(r, Xr , ur )dr +
s s

If the above equation has a unique solution Xts,,u , the process (Xt ) is called a controlled process. This type of problem appears in many applications in insurance and nance. In insurance, Luo [17] consider an optimal dynamic control problem for an insurance company

Jurusam Matematika FMIPA Universitas Udayana, Bali, E-mail : dharmawan.komang@gmail.com

with opportunities of proportional reinsurance and investment. Liang [16] study optimal proportional reinsurance policy of an insurer with a risk process which is perturbed by a diusion. The closed-form expressions for the policy and the value function are derived in the sense of maximizing the expected utility in the jump-diusion framework. Another papers discussing the application of control diusion process in its application can be seen in [5], [7], [12], [13]. In nancial applications, one may refer to [1], [9], [11], [14].

Some useful facts

In this section, we review some facts which are important in next section. Denition 1. A continuous function v : [0, T ) R R is said to be upper semicontinuous if v(s, x) lim sup v(sn , xn )
n

for any sn [0, T ) and x R whenever sn s and xn x, as n . A continuous function v : [0, T ) R R is said to be lower semicontinuous if v(s, x) lim inf v(sn , xn )
n

for any s, sn [0, T ) and x R whenever sn s and xn x as n . The next lemma is well known. Lemma 1. Let {v ; A} be a family of lower semicontinuous functions. Then v(s, x) = sup v (s, x)
A

is lower semicontinuous. Lemma 2. Gronwall Lemma Suppose that the function F : [0, T ] [0, ) satises conditions T F (t)dt <
0

(2.1)

and F (t) a + b

F (s)ds, tT

tT

(2.2)

where a, b 0. Then F (t) aebt , (2.3) The proof of the Gronwall Lemma is well known, one may refer to Dharmawan [7] or Bouchard [5] for the complete proof. The following is a standing assumption on the functions b and appearing in the control system. Assumption 1. For each T > 0 there exists a constant K > 0 such that for all u U, s T and x, y Rd |b(s, x, u) b(s, y, u)| + |(s, x, u) (s, y, u)| K|x y|. |b(s, x, u)| + |(s, x, u)| K(1 + |x|) 2 (2.4) (2.5)

It is well known, see for example [11], p.158 or [22], that Assumption 1 yields the existence of a unique strong solution (Xts,,u ) to (1.1), for each s > 0, each initial condition , and each u U . Moreover, (Xts,,u ) is continuous on [s, T ]. Various versions of the next results are well known, see for example the monograph of Krylov [15] or Bouchard [5]. Denition 2. Quadratic Variation of Martingales Let n = max(tn tn ) as n . The Quadratic variation of a process (Xt ) is i+1 i i dened as a limit in probability Xt = lim
n i=1

(Xtn Xtn )2 . i1 i

(2.6)

If (Xt ) is a martingale, then (Xt2 ) is a submartingale. By compensating Xt2 by an increasing process, it is possible to make it into a martingale. The process which compensates Xt2 to form a martingale turns out to be the quadratic variation of process Xt . Theorem 1. If (Xt ) is a local martingale, then X, Xt exists. Moreover Xt2 X, Xt is a local martingale. Theorem 2. Burkholder-Davis-Gundy For every p 1, there exist two constants cp and Cp such that, for all continuous local martingales M vanishing at zero, [ ] [ ] cp E M, M p/2 E [(M )p ] Cp E M, M p/2 where Mt = supst |Ms |.

Results

In this section we prove some results. The results here are not really new, but the proofs are my original works. Another version of the proofs can be seen in [5]. Theorem 3. Let be an Fs -measurable random variable and for p 2 such that E||p < . Then there exists a constant K(T, p) > 0 which is independent of u such that for all 0 s t T, (3.1) E|Xts,,u |p KE(1 + ||p ). Proof. We dene the stopping times inf{t [s, T ]; |Xts,,u | n}, n 1, n = T, if {t [s, T ]; |X s,,u | n} = t

(3.2)

The stopping times n are well dened since the process Xts,,u is continuous in t [s, T ]. Then following (1.2) we have tn tn s,,u s,,u s,,u Xtn = + b(r, Xr , ur )dr + (r, Xr , ur )dWr , 0 s t T. (3.3)
s s

Invoking the Burkholder-Davis-Gundy inequalities 2 we obtain [ E|Xtn |

p

3 +

p1

E|| + 3
p

p1

E
s

tn s,,u b(r, Xr , ur ) dr p/2

1 p1 3 2

[
s

tn

[ ] s,,u E tr( )(r, Xr , ur ) dr

(3.4)

Using the Jensen inequality and Assumption 1 we nd that [ tn ] s,,u p p1 p p1 p s,,u p E|Xtn | 3 E|| + (3T ) K E (1 + |Xr | )dr s [ tn ] 1 p1 p 1 s,,u p 6 T2 E + (1 + |Xr | )dr . 2 s Therefore, there exists a constant k > 0 such that T ( ) p p1 p s,,u E|Xtn | 3 E|| + kE 1 + |Xrn |p dr.
s s,,u The function gn (t) = E|Xtn |p is integrable on [0, T ] by denition of n and t p1 p gn (t) 3 E|| + k (1 + gn (r)) dr t s. s

(3.5)

(3.6)

(3.7)

Therefore, by the Gronwalls inequality (Lemma 2) to (3.7) ( ) gn (t) 3p1 E||p + T ekt t [s, T ]. Applying the Fatou Lemma in order to pass with n in the above inequality we conclude the proof. Theorem 4. Let Xtn be the solution of the stochastic dierential equation t t s,xn ,u n s, n ,u s, n Xt = + b(r, Xr , ur )dr + (r, Xr ,u , ur )dWr , 0st
s s

(3.8)

Let (Xt ) be the solution of (3.3) and assume that for a certain p 2 E (| n |p + ||p ) < , n 1.

Then there exists a constant C(p, K, T ) which is independent of u such that E sup |Xtn Xt |p C(p, K, T )E | n |p .
tT

Proof. We proceed similarly as in the proof of the previous Theorem, hence some details are omitted. Using Lipschitz property of coecients and 2 we obtain E sup |Xtn Xt |p
tT

E| | + 3
n p n p p1

t n [b(r, Xr , ur )

3 3

p1

E sup
stT

b(r, Xr , ur )] dr

s T

n E| | + C1 (p, K)E |b(r, Xr , ur ) b(r, Xr , ur )|p dr s T n + C1 (p, K, T )E (r, Xr , ur ) (r, Xr , ur )p dr s T p1 n p n 3 E| | + C1 (p, K, T )E |Xr Xr |p dr. p1 s

Now we apply the Gronwall inequality (Lemma 2) with gn (t) = E supsut |Xtn Xt |p to obtain E sup |Xtn Xt |p 3p1 E | n |p + C1 (p, K)(T s)eC1 t
tT

C(p, K, T )E | n |p where C(p, K, T ) = max(3p1 , C1 (p, K)(T s)eC1 T ). Theorem 5. Let Xtsn ,,u , where 0 sn t T be a solution of the stochastic dierential equation tT tT sn ,,u sn ,,u s XtT = + , ur )dr + b(r, Xr (r, Xr n ,,u , ur )dWr , sn t T (3.9)
sn sn

Then for all p 2 there exists a constant C(T, p) which is independent of u and such that E sup Xtsn ,,u Xts,,u
stT p

C(T, p)|sn s|p/2

where s = max (s, sn ). Proof. We will prove the Theorem assuming that sn < s. The case of sn > s is completely analogous and omitted. For simplicity we assume also that the drift b = 0. Then we have Xtsn ,,u = +
s n t t s (r, Xr n ,,u , ur )dWr , s,,u (r, Xr , ur )dWr , s

sn t T, s t T.

Xts,,u = +

Let Xtn = Xtsn ,x,u and Xt = Xts,x,u . Then for s t and invoking again 2 we obtain E sup Xtn Xt p
tt

( C1 (T, p)E

s n (r, Xr , ur )2 sn

)p/2 dr + )p/2
n (r, Xr , ur ) (r, Xr , ur )2 dr s

C1 (T, p)E C(T, p)|sn s|

p/2

+ C(T, p)E
s

t n Xr Xr p dr.

Now we apply the Gronwalls inequality (Lemma 2) with g(t) = E supstt Xtn Xt p . We have E sup Xtn Xt p C1 (T, p)|sn s|p/2 + C1 (T, p)(sn s)p/2 eC1 (t s) .
tT

Choose C(T, p) = max(C1 (T, p), C1 (T, P )eC1 (t s) ); then E sup Xtn Xt p C(T, p)|sn s|p/2 .
tT

Theorem 6. Let Assumption 1 hold. For each p 1, T > 0, t s2 > s1 > 0, E sup |Xts2 ,x2 ,u Xts1 ,x2 ,u |p C1 (|x2 x1 |p + |s2 s1 |p/2 ),
s2 tT

where C1 , is independent of u, s, sn , . Proof. The proof is provided for b = 0. The general case does not lead to any additonal diculties. Let Xt1 = Xts1 ,x1 ,u , Xt2 = Xts2 ,x2 ,u . Then s2 p 2 1 p p1 p p1 1 E sup Xt Xt 3 |x2 x1 | + 3 E (r, Xr , ur )dWr
s2 tt

+3
p1

s1 t 2 ((s, Xs , us ) p

E sup

s2 tt

1 (s, Xs , us ))dWs

s2

Then, using the Burkholder-Davis-Gundy inequality (Theorem 2) we obtain ( E sup

s2 tt

Xt2

Xt1 p

p1

|x2 x1 | + 3
p

p1

Cp E

s2

)p/2
1 |(r, Xr , ur )|2 dr

( +3p1 Cp E

s1 t 2 1 |(s, Xs , us ) (s, Xs , us )|2 ds

)p/2 .

s2

Using the assumptions on and the Jensen inequality we obtain s2 ( p p) 2 1 p p1 p 1 1 E sup Xt Xt 3 |x2 x1 | + c1 |s2 s1 | 2 1 + Xr , us ) dr
tT

+c2 T p/2 E

s1 t 2 1 |Xr Xr |p dr,

s2 2 where c1 , c2 > 0 are constants independent of s1 , s2 , x1 , x2 , u. Let F (t) = E sup Xr 1 Xr p . The above inequality and Theorem 3 yield s2 rt

F (t) 3
p1

|x2 x1 | + c3 |s2 s1 |
p

p/2

+ c4
s2

F (r)dr,

for some positive constants c3 , c4 that are independent of s1 , s2 , u. Moreover, if |x1 | , |x1 | R then c3 , c4 depend on R only but not on specic values of x1 and x2 . By the Gronwall inequality (Lemma 2) we have F (t) 3p1 |x2 x1 |p + c3 |s2 s1 |p/2 + c4 |s2 s1 |p/2 eC4 (ts2 ) . Let C = max(3p1 , C3 , C4 eC1 (ts2) ). Then E sup Xt2 Xt1 p C(|x2 x1 |p + |s2 s1 |p/2 ).
tT

Theorem 3 - 6 are important to show the smoothness of value functions arising in pricing barrier options which appear in non-convex payo functions.

References
[1] Avellaneda, M., Levy, A., and Paras, A. 1995. Pricing and hedging derivatives in markets with stochastic volatilities. Applied Mathematical Finance. 2, p.73-88. [2] Bardi, M. and I. Capuzo Dolcetta. 1997. Optimal control and viscosity solutions to Hamilton-Jacobi-Bellman equations. Birkhauser Boston. [3] Barles G. and Perthame. 1988. Exit time problems in optimal control and vanishing viscosity method. SIAM journal in Control and Optimisation. 26. p.1133-1148. [4] Borkar, V.S. 1989. Optimal Control of Diusion Processes, Pitman Research Notes, No. 203, Longman Sci. and Tech. Harlow, UK. [5] Bouchard, B. 2007. Introduction to stochastic control of mixed diusion processes, viscosity solutions and applications in nance and insurance. First Draft. Universite Paris-IX Dauphine, Ceremade and CREST [6] Davis M.H.A. and J.M.C. Clark. 1994. A note on super-replicating strategies. Phil. Trans. R. Soc. London A 347, p.485-494. [7] Dharmawan, K. 2005.Superreplication Method for Multi-Asset Barrier Oprions. PhD. Thesis. University of New South Wales Library. Sydney, Australis. [8] El-Karoui N. and M.C. Quenez. 1995. Dynamic programming and the pricing of contingent claim in an incomplete market. SIAM, Journal of Control and Optimisation. 33. p.29-66. [9] El-Karoui, S. Peng, and M.C. Quenez. 1997. Backward Stochastic Dierential Equations. Mathematical Finane, Vol. 7, No. 1. p.1-71 [10] El-Karoui N., M. Jeanblanc-Pique, S.E. Shreve. 1998. Robustness of the Black-Schole Formula. Math. Finance 8(2), p.93-126. [11] Fleming, W.H. and H.M. Soner. 1993. Controlled Markov Processes and Viscosity Solution. Springer-Verlag, New-York. [12] Hipp C. and M. Plum. 2001. Optimal investment for insurers. Preprint. [13] Hojgaard B. and M. Taksar. 2001. Optimal risk control for a large corporation in the presence of returns on investments. Finance and Stochastics, 5, 527-547. [14] Karatzas, I. 1996. Lectures on mathematics of nance. CRM monograph series, ISSN 1065-8599; v.8 [15] Krylov, N.V. 1990. Control Diusion Processes. Springer-Verlag, New York. [16] Liang Z. 2007. Optimal Proportional Reinsurance for Controlled Risk Process which is Perturbed by Diusion. Acta Mathematicae Applicatae Sinica (English Series) Volume 23, Number 3 (2007), 477-488,

[17] Luo S.2010. Ruin Minimization for Insurers with Borrowing onstraints. North America Actuarial Journal, Vol. 12, No. 2 [18] Oksendal, B. 1995. Stochastic Dierential Equations. Fourth Edition, Springer Verlag, New York. [19] Soner, Halil M. 1997. Controlled Markov Processes, Viscosity Solutions and Applications to mathematical Finance. in Viscosity solution and Applications. Eds. I.C. Dolcetta and P.L. Lions. p.134-185. [20] Soner M. and N. Touzi. 2002. Dynamic Programming for Stochastic Target Problems and Geometric Flows. Journal of the European Mathematical Society, 4, 201-236. [21] Taksar M. 2000. Optimal risk/dividend distribution control models. Applications to insurance. Math. Methods Oper. Res., 1, 1-42. [22] Yong J. and X.Y, Zhou. 1999. Stochastic Control : Hamiltonian Systems and HJB Equations Springer-Verlag, New York.