Cancellations for dispersive PDEs with random initial data
Trees (05C05) Hopf algebras and their applications (16T05) Existence problems for PDEs: global existence, local existence, non-existence (35A01) Uniqueness problems for PDEs: global uniqueness, local uniqueness, non-uniqueness (35A02) Smoothness and regularity of solutions to PDEs (35B65) Heat equation (35K05) Wave equation (35L05) Time-dependent Schrödinger equations and Dirac equations (35Q41) NLS equations (nonlinear Schrödinger equations) (35Q55) Applications of stochastic analysis (to PDEs, etc.) (60H30) White noise theory (60H40) Algebraic structures and computation (60L70)
The authors present a combinatorial formalism for dealing with cancellations in the context of dispersive partial differential equations with random initial data. They start with the Schrödinger equation: \((\partial _{t}-i\Delta )u=i\mu ^{2}\left\vert u\right\vert ^{2}u\), with the initial condition \(u(0,x)=v(x)\), \(x\in \mathbb{T}_{L}^{d}=[0,L]^{d}\). The random initial data \(v\) is given by \(v(x)=\frac{1}{L^{d}}\sum_{k\in \mathbb{Z} _{L}^{d}}v_{k}e^{2\pi ikx}\), \(v_{k}=\sqrt{w_{k}}\eta _{k}\), where \(\mathbb{Z} _{L}^{d}=(L^{-1}\mathbb{Z})^{d}\), \(w:\mathbb{R}^{d}\rightarrow \lbrack 0,+\infty )\) is a given Schwartz function, and the \(\eta _{k}\) are i.i.d. centered complex Gaussian random variables satisfying \(\mathbb{E}(\left\vert \eta _{k}\right\vert ^{2})=1\), \(\mathbb{E}(\eta _{k}\eta _{\ell })=0\), for \( k,\ell \in \mathbb{Z}_{L}^{d}\). The authors rewrite this equation in Duhamel form as \(u(t)=e^{it\Delta v}+i\mu ^{2}\int_{0}^{t}e^{i(t-s)\Delta }\left\vert u(s)\right\vert ^{2}u(s)ds\). In Fourier space, one has: \( u_{k}(t)=e^{-itk^{2}}v_{k}+i\mu ^{2}\sum_{k=-k_{1}+k_{2}+k_{3}}\int_{0}^{t}e^{-i(t-s)k^{2}}\overline{u} _{k_{1}}(s)u_{k_{1}}(s)u_{k_{1}}(s)ds\). The set \(\mathfrak{T}\) of decorated trees consists of elements of the form \(T_{e}^{\mathfrak{f}}\), where \(T\) is a nonplanar rooted tree (a rooted tree where the order of a node's children does not matter) with root node \(\varrho _{T}\), edge set \(E_{T}\), node set \( N_{T}\) and leaves set \(L_{T}\), \(\mathfrak{e}:E_{T}\rightarrow \{\mathfrak{t} _{1},\mathfrak{t}_{2}\}\times \{0,1\}\) are edge decorations (all the edges of \(T_{e}^{\mathfrak{f}}\) are supposed to be connected to leaves decorated by \((\mathfrak{t}_{1},p)\) with \(p\in \{0,1\}\) and all the other edges are decorated by \((\mathfrak{t}_{2},p)\)), and \(\mathfrak{f}:N_{T}\setminus \{\varrho _{T}\}\rightarrow \mathbb{Z}^{d}\) are node decorations satisfying, for every inner node \(u\), the relation \((-1)_{\mathfrak{c}(\mathfrak{e}u)} \mathfrak{f}(u)=\sum_{e=(u,v)\in E_{T}}(-1)^{\mathfrak{c}(e)}\mathfrak{f}(v)\) , where \(\mathfrak{e}(e)=(\mathfrak{t}(e),\mathfrak{c}(e))\) and \(e_{u}\) is the edge outgoing \(u\) of the form \((w,u)\). The first main result connects the two formalisms via decorated trees and words, allowing to switch from decorated trees to words where one can perform the algebraic manipulations required to compute the cancellations: For every decorated tree \(T_{ \mathfrak{e},p}^{\mathfrak{f}}\), \((\Pi T_{\mathfrak{e},p}^{\mathfrak{f} })(t)=(\Pi ^{A}\mathfrak{a}(T_{\mathfrak{e},p}^{\mathfrak{f}}))(t)\), where \( (\Pi T_{\mathfrak{e},p}^{\mathfrak{f}})(t)=\int Q\Pi e\in E_{T}^{2}(-1)^{ \mathfrak{c}(e)}i\mu ^{2}\mathbf{1}_{\{0<t_{e_{-}}<t_{e_{+}}\}}(\exp((-1)^{ \mathfrak{c}(e)}i(t_{e_{-}}-t_{e_{+}})\mathfrak{f}(e_{-})^{2}))\) \(\times \Pi _{z\in E_{T}^{\mathfrak{p}}}\xi _{\mathfrak{f}(z_{-})}^{\mathfrak{c} (z)}(t_{z_{+}})\Pi _{\{x,y\}\in \mathfrak{p}}\mathbb{E}(\xi _{\mathfrak{f} (x_{-})}^{\mathfrak{c}(z)}(t_{x_{+}})\xi _{\mathfrak{f}(y_{-})}^{\mathfrak{c} (y)}(t_{y_{+}})\Pi _{e\in E_{T}^{2}}dt_{e_{-}}\), and \((\Pi ^{A}\mathfrak{a} (T_{\mathfrak{e},p}^{\mathfrak{f}}))(t)\) is an evaluation map on words. The main tool of the proof is the possibility to rewrite any internal edge of a tree as a covariance between two edges: \(e^{i(s-t)k^{2}}=\mathbb{E} (e^{-itk^{2}}\eta _{k}\overline{e^{-isk^{2}}\eta _{k}})\). The authors then consider the case of the 3D cubic wave equation: \((\partial _{t}^{2}+1-\Delta )u=-u^{3}\), with the initial condition \((u,\left\langle \nabla \right\rangle ^{-1}u)\mid _{t=0}=(\varphi ^{\cos},\varphi ^{\sin})\), where \((t,x)\in \mathbb{R}\times \mathbb{T}^{3}\). Given independent copies \( (W_{n}^{\varphi })_{n\in \mathbb{Z}^{3}}\) of the Gaussian process with mean zero and covariance given by \(E(W_{n}^{\varphi }(s)W_{m}^{\varphi }(t))=\delta _{m+n}s\wedge t\), they define the Gaussian process \( W_{s}^{\varphi }\) by \(W_{s}^{\varphi }(x)=\sum_{n\in \mathbb{Z} ^{3}}W_{n}^{\varphi }(s)e^{inx}\). They consider the stochastic heat equation with space-time white noise given by \((\partial _{t}+1-\Delta )u^{\varphi }= \sqrt{2}dW_{t}^{\varphi }\), \(u^{\varphi }(t,x)=\sqrt{2}\sum_{n\in \mathbb{Z} ^{3}}u_{n}^{\varphi }(t)e^{inx}\), with \(u_{n}^{\varphi }(t)=\int_{-\infty }^{t}e^{-(t-s)\left\langle n\right\rangle ^{2}}dW_{n}^{\varphi }(s)\). They also consider the solutions to the linear wave equation: \((\partial _{t}+1-\Delta )v=0\), with the initial condition \((v,\left\langle \nabla \right\rangle ^{-1}v)\mid _{t=0}=(u^{\cos},u^{\sin})\), which can be described explicitly by \(v(t,x)=\cos(t\left\langle \nabla \right\rangle )u^{\cos}(t,x)+\sin(t\left\langle \nabla \right\rangle )u^{\sin}(t,x)\). In the Fourier space, one has \(v_{n}(t)=\cos(t\left\langle n\right\rangle )u_{n}^{\cos}(t)+\sin(t\left\langle n\right\rangle )u_{n}^{\sin}(t)\). They prove a quite similar result in the context, mainly using the identity: \( \frac{\sin((t-t^{\prime })\left\langle n\right\rangle )}{\left\langle n\right\rangle }=-\partial _{t}\mathbb{E}(v_{n}(t)v_{-n^{\prime }}(t^{\prime }))=\partial _{t^{\prime }}\mathbb{E}(v_{n}(t)v_{-n^{\prime }}(t^{\prime }))\).
- A class of growth models rescaling to KPZ
- A theory of regularity structures
- A Wong-Zakai theorem for stochastic PDEs
- Algebraic renormalisation of regularity structures
- An Algebraic Theory of Integration Methods
- Approximations of dispersive PDEs in the presence of low-regularity randomness
- Controlling rough paths
- Differential equations driven by rough signals
- Ecalle's arborification-coarborification transforms and Connes-Kreimer Hopf algebra
- Full derivation of the wave kinetic equation
- Geometric stochastic heat equations
- Geometric versus non-geometric rough paths
- Hopf algebras, renormalization and noncommutative geometry
- Invariant Gibbs measures for the three dimensional cubic nonlinear wave equation
- Nondivergence form quasilinear heat equations driven by space-time white noise
- Poincaré-Dulac normal form reduction for unconditional well-posedness of the periodic cubic NLS
- Quasi-shuffle algebras and renormalisation of rough differential equations
- Ramification of rough paths
- Random tensors, propagation of randomness, and nonlinear dispersive equations
- Renormalising SPDEs in regularity structures
- Renormalization in quantum field theory and the Riemann-Hilbert problem. I: The Hopf algebra structure of graphs and the main theorem
- Resonance-based schemes for dispersive equations via decorated trees
- Solving the KPZ equation
- The arborification-coarborification transform: analytic, combinatorial, and algebraic aspects
- The Hopf algebra of rooted trees, free Lie algebras, and Lie series
This page was built for publication: Cancellations for dispersive PDEs with random initial data
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q7308703)
