Partial differential equations. A unified Hilbert space approach (Q550501): Difference between revisions

The authors study partial differential equations from the standpoint of Sobolev chains and Sobolev lattices. Let \(C\subseteq H\oplus H\) be a densely defined, closed linear operator in a Hilbert space \(H\) with \(0\) in the resolvent set \(\rho(C)\). Then the triple \((H_1(C), H,H_{-1}(C))\), denoting \(H_1(C)= D(C)\) with norm \(|C\cdot|_H\), and \(H_{-1}(C)= D(C^*)\) with norm \(|C^*\cdot|_H\), is called a short Sobolev chain. \(D(C^n)\) equipped with the norm \(|C^n\cdot|_H\), is denoted by \(H_n(C)\). \(H_{-n}(C):=H_n(C^*)^*\). The family \((H_n(C))_{n\in\mathbb{Z}}\) is called the long Sobolev chain associated with \(C\). \(H_{-\infty}(C)= \bigcup_{k\in\mathbb{Z}}H_k(C)\), and \(H_\infty(C)=\bigcap_{k\in\mathbb{Z}} H_k(C)\). Let \(D= 2^{-1/2}(m-\partial)\), \(D^*= 2^{-1/2}(m+\partial)\). Starting from the base space \(L_2(\mathbb{R},\) \(\exp(-2\nu x) dx)\), the authors define \(D_\nu:= \exp(\nu m)D\exp(-\nu m)\), \(D^*_\nu:\exp(\nu m)D^*\exp(-\nu m)\), and give the Sobolev chain associated with \(|D_\nu|\), \(\nu\in\mathbb{R}\), in \(\mathbb{R}\). They also define the Sobolev lattice for the family \((H_\alpha(C); \alpha\in\mathbb{Z}^{n+1})\) with \(C= (C_i)_{i=0,\dots, n}\), and obtain the relation \((\partial_\nu\pm 1)^k= L^*_\nu(im\pm 1)^k L_\nu\), by using Fourier-Laplace transform \(L_\nu\). They treat linear PDEs with constant coefficients in \(H_{-\infty}(\partial_\nu+ e)\) and in \(H_{-\infty}(|D_\nu|)\) in Chapter 3. Although evolution equations of mathematical physics (MP) are properly treated in Chapter 5, we find many examples appearing in MP, and fundamental solutions of them, in Chapter 3. They treat ``Causality'' in 3.1.10 and 4.2.2. That is, let \(\nu\) be a direction in \(\mathbb{R}^{n+1}\). For a linear functional \(f\) on \(C\degree_\infty(\mathbb R^{n+1})\) define \[ \text{supp}_{\nu} f:=\mathbb{R}\setminus\bigcup \{I\mid(I\text{ open in }\mathbb{R})\wedge (f= 0\text{ in }[I]_{\nu}+ \{\nu\}^\perp)\}. \] Let \(W\) be a mapping from a subset of linear functionals on \(C\degree_\infty(\mathbb{R}^{n+1})\) with values in such subset. \(W\) is causal in direction \(\nu\) if \[ \inf\text{supp}_{\nu(0)}(f- g)\leq \inf\text{supp}_{\nu(0)}(W(f)- W(g)). \] In Chapter 6, they treat evolutionary problems with material laws. Let \(A\) be a skew self-adjoint operator. They aim to find \(U,V\in H_{-\infty}(\partial_\nu+ \nu)\otimes H\) such that, for a given \(f\in H_{-\infty}(\partial_\nu+\nu)\otimes H\), \(\partial_0 V+ AU= f\) and \(V= M(\partial^{-1}_0)U\) hold. \(M(\partial^{-1}_0):= L^*_\nu M(1/(im_0+ \nu)) L_\nu,\) \(\nu> 1/(2r)\), is forward causal, if \(\{M(z);\;z\in B_{C(r,r)}\}\) is a holomorphic family of uniformly bounded linear operators on \(H\). \(B_{C(r,r)}\to [i\mathbb{R}]+ [R_{>1/(2r)}]\), \(z\to 1/z\), is a bijection for \(r\in\mathbb{R}_{> 0}\).