Stochastic Delay Differential Equations have blow-up solutions if and only if their instantaneous counterparts have them
Julius Busse · 2024
Read Stochastic Delay Differential Equations have blow-up solutions if and only if their instantaneous counterparts have them by Julius Busse (2024) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2412-13383.txt.
Introduction
Julius Busse settles when delayed stochastic equations can blow up in this 2024 math paper. A blow-up means a solution races to infinity in finite time, and delay equations, whose rates depend on past states, are harder to judge. Busse proves that for regular equations with one fixed delay, blow-up solutions exist exactly when the same equation without delay has them. The proof leans on a classic comparison theorem from 1977, and the result serves fields from biology to finance.
Catalog record
| CREATOR | Julius Busse |
|---|---|
| YEAR | 2024 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 33 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2412-13383
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Stochastic Delay Differential Equations have blow-up solutions if and only if their instantaneous counterparts have them Julius Busse Abstract Motivated by a recent publication by Ishiwata and Nakata [ 6 ] , we prove that sufficiently regular stochastic delay differential equations (SDDEs) with a single discrete delay have blow up solutions if and only if their undelayed counterparts have them, using a comparison theorem by Ikeda and Watanabe [ 7 ] . This result has applications in mathematical biology and finance. 1 Introduction Delay differential equations appear in many applications, such as in population modeling [ 13 ] , modeling the glucose-insulin system or other processes in the human body [ 11 ] and in control problems in engineering [ 8 ] . We introduce stochasticity to these models. “All biological dynamical systems evolve under stochastic forces, if we define stochasticity as the parts of the dynamics that we either cannot predict or understand or that we choose not to include in the explicit modeling” [ 1 , p.3] . The resulting SDDEs are a growing research interest, both from a purely mathematical perspective [ 16 ] , in mathematical biology [ 5 ] , and also in mathematical finance, where they can be used to model insider trading [ 10 ] . Explosion properties of these SDDEs are of interest because of their applications in mathematical biology . In mathematical biology, there is a research interest in SDE models with non locally Lipschitz noise coefficients [ 15 ] , [ 1 , p.12] . Using the Itô formula, we can transform the models we study onto (0,\infty) , where explosion becomes extinction (see Example 3.9 ) and the noise coefficients are non locally Lipschitz, thus allowing us to gain insight about explosion properties and estimates of these probabilities using our results (more in [ 2 ] ). Non locally Lipschitz noise coefficients are also used in mathematical finance, for example when modelling how interest rates evolve over time [ 3 ] , where a realistic restriction is that interest rates may never be negative. Again this can be answered by transforming models we study onto (0,\infty) using the Itô formula. We combine the comparison theorem by Ikeda and Watanabe as presented in [ 7 ] and the analysis of blow-up solutions for scalar differential equations with discrete delay as described in [ 6 ] to reduce explosion behavior of sufficiently regular SDDE models to that of SDEs. Through an application of Itô’s formula, this can then be applied to extinction properties of delayed population models with non locally Lipschitz noise coefficients [ 2 ] . 2 Mathematical Basics 2.1 Basics of SDEs 2.1.1 Elementary definitions (SDEs) For basic notions from probability theory, including Brownian motion, we refer to [ 9 ] . Analogously to the definition of a strong solution given earlier in [ 9 , p.285] , we can also define the concept of a strong solution up to an explosion time. Thus, we arrive at the following definition. Definition 2.1 (Strong local solutions of SDEs) . A strong solution up to an explosion time of the SDE \,\mathrm{d}x(t)=b(x(t),t)\,\mathrm{d}t+\sigma(x(t),t)\,\mathrm{d}W(t) (1) on the given probability space (\Omega,\mathcal{F},\mathbb{P}) with filtration (\mathcal{F}_{t})_{t\geq 0} and with respect to the fixed Brownian motion W on (\Omega,\mathcal{F},\mathbb{P}),(\mathcal{F}_{t})_{t\geq 0} and initial condition x_{0} (a \mathcal{F}_{0} -measurable random variable) is a process x=\{x(t):0\leq t<\infty{}\} with sample paths that are continuous in the extended real numbers \mathbb{R}\cup\{-\infty{},+\infty{}\} and with the following properties. 1. x is adapted to the filtration (\mathcal{F}_{t})_{t\geq 0} , \mathbb{P}[x(0)=x_{0}]=1 . 2. Define S_{n}:=\inf\{t\geq 0:|x(t)|>n\} (possibly \infty{} ), then we have \mathbb{P}\left[\int_{0}^{t\land{}S_{n}}|b(x(s),s)|+\sigma^{2}(x(s),s)\,\mathrm{d}s<\infty{}\right]=1 (2) for all 0\leq t<\infty{} , and we have 3. \displaystyle x(t\land{}S_{n})= \displaystyle x(0)+\int_{0}^{t}b(x(s),s)1_{\{s\leq S_{n}\}}\,\mathrm{d}s (3) \displaystyle+\int_{0}^{t}\sigma(x(s),s)1_{\{s\leq S_{n}\}}\,\mathrm{d}W(s) (4) for all t\geq 0 , n\geq 1 almost surely. We refer to S=\lim_{n\to\infty{}}S_{n} (5) as the explosion time for x . We say x explodes or x blows up at time S . Remark 2.2 . If explosion occurs towards +\infty{} , we denote the corresponding explosion time by S^{+} or T_{+\infty} , and if it occurs towards -\infty{} , we denote it by S^{-} or T_{-\infty} . Similarly, in the case of x_{0}>0 , we call the time (S_{0}=)T_{0}:=\lim_{n\to\infty{}}\inf\{t\geq 0:x(t)\notin(\frac{1}{n},\infty{})\} the extinction time of x . 2.1.2 Existence and uniqueness results (SDEs) The following definition is taken from [ 12 , p.250] Definition 2.3 ((Locally) Lipschitz functions) . A function f:\mathbb{R}\cup\{-\infty{},\infty{}\}\times\mathbb{R}_{+}\to\mathbb{R} is Lipschitz if there exists a constant k such that 1. |f(x,t)-f(y,t)|\leq k|x-y| , for each t>0 , x,y\in\mathbb{R}\cup\{-\infty{},\infty{}\} , and 2. t\mapsto f(t,x) is right continuous with left limits, for each x\in\mathbb{R}\cup\{-\infty{},\infty{}\} . Similarly, f is called locally Lipschitz if for every compact K\subset{}\mathbb{R}\cup\{-\infty{},\infty{}\} there exists a k such that |f(x,t)-f(y,t)|\leq k|x-y| for all x,y\in K and all t>0 . The following result is mentioned in [ 12 , p.254] . Theorem 2.4 (Existence and uniqueness of SDE solutions). Given an SDE \,\mathrm{d}x(t)=b(x(t),t)\,\mathrm{d}t+\sigma(x(t),t)\,\mathrm{d}W(t), (6) where b and \sigma are locally Lipschitz as defined above. Then there exists a unique strong solution up to an explosion time T . 2.2 Basics of SDDEs 2.2.1 Elementary definitions (SDDEs) For the definition of a solution of an SDDE and for the corresponding existence and uniqueness results we refer to [ 16 ] . Fix a probability space (\Omega,\mathcal{F},\mathbb{P}) together with a filtration (\mathcal{F}_{t})_{0\leq t<\infty{}} satisfying the usual conditions. Denote by \mathbb{D}[0,\infty{}) the space of \mathcal{F}_{t} -adapted càdlàg processes \{x(t):0\leq t<\infty{}\} . One can trivially extend the filtration by setting \mathcal{F}_{s}=\mathcal{F}_{0} for -\tau\leq s<0 , for some \tau>0 fixed, and in a similar fashion to the above we write \mathbb{D}[-\tau,\infty{}) for the space of (\mathcal{F}_{t})_{-\tau\leq t<\infty{}} -adapted càdlàg processes \{x(t):-\tau\leq t<\infty{}\} . The space \mathbb{D}[-\tau,0] consists of all càdlàg processes \{x(t):-\tau\leq t\leq 0\} where x(s) is \mathcal{F}_{0} -measurable for all times s\in[-\tau,0] . A stochastic process x=\{x(t):-\tau\leq t<\infty{}\in\mathbb{D}[-\tau,\infty{})\} gives rise to the segment process \{x_{t}:t\geq 0\} , where each segment x_{t} is a \mathcal{F}_{t} -measurable càdlàg process on [-\tau,0] defined by x_{t}(\theta):=x(t+\theta),\quad-\tau\leq\theta\leq 0. (7) The segment process can be regarded as an \mathcal{F}_{t} -adapted \mathbb{D}[-\tau,0] -valued stochastic process, thus where each x_{t} is a \mathbb{D}[-\tau,0] -valued random variable. Let W denote a Brownian motion. We wish to provide a rigorous mathematical interpretation for autonomous equations of the form \begin{cases}\,\mathrm{d}x(t)=a(x_{t})\,\mathrm{d}t+b(x_{t-})\,\mathrm{d}W(t),\quad&\text{for }t\geq 0,\\ x(u)=\Phi(u),\quad&\text{for }u\in[-\tau,0],\end{cases} (8) subject to an initial process \Phi\in\mathbb{D}[-\tau,0] . Equation ( 8 ) is shorthand notation for the integral equation x(t)=\Phi(0)+\int_{0}^{t}\bm{A}_{\Phi}(x)(s)\,\mathrm{d}s+\int_{0}^{t}\bm{B}_{\Phi}(x)(s-)\,\mathrm{d}W(s), (9) t\geq 0 , where 1. the maps \bm{A}_{\Psi},\bm{B}_{\Psi}:\mathbb{D}[0,\infty{})\to\mathbb{D}[0,\infty{}) , for some fixed \Psi\in\mathbb{D} , are defined pathwise for all y=\{y(t):t\geq 0\}\in\mathbb{D}[0,\infty{}) by \displaystyle\bm{A}_{\Psi}(y)(s,\omega)=a(y_{s}^{\Psi}(\omega))\text{ and } (10) \displaystyle\bm{B}_{\Psi}(y)(s,\omega)=b(y_{s}^{\Psi}(\omega)),\ s\geq 0; (11) 2. the (y_{t}^{\Psi})_{t\geq 0} is the segment process of y^{\Psi}\in\mathbb{D}[-\tau,\infty{}) , where y^{\Psi}(s)=\begin{cases}\Psi(s),&-\tau\leq s<0,\\ y(s),&s\geq 0;\end{cases} (12) 3. the functional a:\mathbb{D}[-\tau,0]\to\mathbb{R} and b:\mathbb{D}[-\tau,0]\to\mathbb{R} will be assumed to be locally Lipschitz and continuous. Definition 2.5 (Strong local solution of SDDE) . Let x=\{x(t):-\tau\leq t<\infty{}\} be a stochastic process and S_{\infty}{} be a stopping time. Then (x,S_{\infty}{}) , often abbreviated by x again, is said to be a strong solution up to an explosion time of the equation ( 8 ) on the interval [-\tau,S_{\infty}{}) , if x_{0}=\Phi holds \mathbb{P} -a.s., x^{S_{k}}\in\mathbb{D}[0,\infty{}) , and we have \mathbb{P} -a.s. \displaystyle x(t\land{}T_{k})=\Phi(0) \displaystyle\int_{0}^{t\land{}S_{n}}\bm{A}_{\Phi}(x^{S_{n}})(s)\,\mathrm{d}s+ (13) \displaystyle+ \displaystyle\int_{0}^{t\land{}S_{n}}\bm{B}_{\Phi}(X^{S_{n}})(s-)\,\mathrm{d}W(s), (14) t\geq{0} , where (S_{n})_{n\geq 1} is a