Continuous-time Stochastic Control and Optimization with by Huyên Pham

By Huyên Pham

Stochastic optimization difficulties come up in decision-making difficulties lower than uncertainty, and locate a number of purposes in economics and finance. however, difficulties in finance have lately ended in new advancements within the conception of stochastic control.

This quantity presents a scientific therapy of stochastic optimization difficulties utilized to finance by way of featuring the several latest tools: dynamic programming, viscosity suggestions, backward stochastic differential equations, and martingale duality tools. the idea is mentioned within the context of contemporary advancements during this box, with entire and exact proofs, and is illustrated through concrete examples from the realm of finance: portfolio allocation, choice hedging, genuine strategies, optimum funding, etc.

This ebook is directed in the direction of graduate scholars and researchers in mathematical finance, and also will profit utilized mathematicians drawn to monetary functions and practitioners wishing to grasp extra concerning the use of stochastic optimization tools in finance.

Show description

Read or Download Continuous-time Stochastic Control and Optimization with Financial Applications PDF

Similar linear programming books

Optimization Theory

"Optimization thought is turning into a an increasing number of vital mathematical in addition to interdisciplinary region, specifically within the interaction among arithmetic and lots of different sciences like laptop technology, physics, engineering, operations study, and so forth. "This quantity offers a accomplished creation into the speculation of (deterministic) optimization on a complicated undergraduate and graduate point.

Trust-region methods

This is often the 1st finished reference on trust-region tools, a category of numerical algorithms for the answer of nonlinear convex optimization equipment. Its unified therapy covers either unconstrained and limited difficulties and stories a wide a part of the really good literature at the topic.

Convex analysis

On hand for the 1st time in paperback, R. Tyrrell Rockafellar's vintage examine provides readers with a coherent department of nonlinear mathematical research that's particularly suited for the examine of optimization difficulties. Rockafellar's idea differs from classical research in that differentiability assumptions are changed via convexity assumptions.

Hybrid Dynamical Systems : Modeling, Stability, and Robustness

Hybrid dynamical platforms convey non-stop and prompt adjustments, having good points of continuous-time and discrete-time dynamical platforms. jam-packed with a wealth of examples to demonstrate suggestions, this ebook provides a whole thought of strong asymptotic balance for hybrid dynamical platforms that's appropriate to the layout of hybrid regulate algorithms--algorithms that function good judgment, timers, or combos of electronic and analog parts.

Extra resources for Continuous-time Stochastic Control and Optimization with Financial Applications

Sample text

Dx ϕ(x) + tr(σ(t, x)σ (t, x)Dx2 ϕ(x)), 2 ϕ ∈ C 2 (Rn ). 12). 12), v(t, x) a (real-valued) function of class C 1,2 on T × Rn and r(t, x) a continuous function on T × Rd , we obtain by Itˆo’s formula Mt := e− Rt 0 t r(s,Xs )ds v(t, Xt ) − e− Rs 0 r(u,Xu )du 0 t = v(0, X0 ) + e − Rs 0 r(u,Xu )du ∂v + Ls v − rv (s, Xs )ds ∂t Dx v(s, Xs ) σ(s, Xs )dWs . 20) 0 The process M is thus a continuous local martingale. 22) n where f (resp. g) is a continuous function from [0, T ] × Rn (resp. Rn ) into R. We also assume that the function r is nonnegative.

The objective is to maximize over control processes the gain function J, and we introduce the associated value function: v(t, x) = sup α∈A(t,x) J(t, x, α). 2 Controlled diffusion processes 39 • Given an initial condition (t, x) ∈ [0, T ) × Rn , we say that α ˆ ∈ A(t, x) is an optimal control if v(t, x) = J(t, x, α). ˆ • A control process α in the form αs = a(s, Xst,x ) for some measurable function a from [0, T ] × Rn into A, is called Markovian control. In the sequel, we shall implicitly assume that the value function v is measurable in its arguments.

12) starting at time t. e. Xt = ξ. The uniqueness is pathwise and means that if X and Y are two such strong solutions, we have P [Xs = Ys , ∀t ≤ s ∈ T] = 1. This solution is square integrable: for all T > t, there exists a constant CT such that E sup |Xs |p ≤ CT (1 + E[|ξ|p ]). t≤s≤T This result is standard and one can find a proof in the books of Gihman and Skorohod [GS72], Ikeda and Watanabe [IW81], Krylov [Kry80] or Protter [Pro90]. 12) starting from ξ at time t. When t = 0, we simply write X ξ = X 0,ξ .

Download PDF sample

Rated 4.45 of 5 – based on 35 votes