A control problem with fuel constraint and Dawson-Watanabe superprocesses (Q389072)

From MaRDI portal
scientific article
Language Label Description Also known as
English
A control problem with fuel constraint and Dawson-Watanabe superprocesses
scientific article

    Statements

    A control problem with fuel constraint and Dawson-Watanabe superprocesses (English)
    0 references
    0 references
    17 January 2014
    0 references
    The author is concerned with the problem of minimizing the functional \[ \mathrm{E}_{0,z}\Biggl[\int^T_0|\dot x(t)|^p \eta(Z_\tau)\,dt+ \int^T_0 |x(t)|^p A(dt)\Biggr] \] over adapted and absolutely continuous strategies \(x(t)\) satisfying \(x(0)= x_0\) and \(x(T)= 0\). Here \(p\in[2, \infty]\), \(\eta\) is a strictly positive function, and \(A\) a nonnegative additive functional of the time-inhomogeneous Markov process \(Z\) with \(Z_0= z\) a.s. \([\mathrm{P}_{0,z}]\). This control problem is related to the monotone follower problems with fuel constraint introduced by \textit{V. E. Benes}, \textit{L. A. Shepp} and \textit{H. S. Witsenhausen} [Stochastics 4, 39--83 (1980; Zbl 0451.93068)]. It also occurs in mathematical finance when looking into strategies minimizing the cost of liquidating a given amount of stock within a certain interval of time, see [\textit{R. Almgren}, SIAM J. Financ. Math. 3, No. 1, 163--181 (2012; Zbl 1256.49031)]. The author solves his problem using the log-Laplace transforms of \(J\)-functionals (as introduced by \textit{E. B. Dynkin} [Probab. Theory Relat. Fields 90, No. 1, 1--36 (1991; Zbl 0727.60095)]) of superprocesses with not necessarily homogeneous branching parameters. The solution is related to the solution of quasilinear parabolic PDEs of the form \[ v_t- (p- 1) \eta^{1/(p-1)} v^{1+ 1/(1-p)}+ a+ L_t v= 0,\;v(T, z)= \infty, \] where \(L_t\) is the generator of \(Z\). This is the type of equations solved by \textit{E. B. Dynkin} [Ann. Probab. 20, No. 2, 942--962 (1992; Zbl 0756.60074)] by means of superprocesses. Clearly, the direct probabilistic approach followed here is more elegant than the classical, often cumbersome approach via the HJB equation. As a byproduct, the author obtains sharp bounds on the blow-up behavior of the log-Laplace functionals, which are of interest in themselves.
    0 references
    optimal stochastic control
    0 references
    fuel constraint
    0 references
    optimal trade execution
    0 references
    superprocess
    0 references
    \(J\)-functional
    0 references
    log-Laplace equation
    0 references

    Identifiers

    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references