Strongly \(P\)-positive operators and explicit representations of the solutions of initial value problems for second-order differential equations in Banach space (Q1304685)
From MaRDI portal
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | Strongly \(P\)-positive operators and explicit representations of the solutions of initial value problems for second-order differential equations in Banach space |
scientific article |
Statements
Strongly \(P\)-positive operators and explicit representations of the solutions of initial value problems for second-order differential equations in Banach space (English)
0 references
11 September 2000
0 references
The author considers the following initial value problems: \[ {d^2x(t)\over dt^2}+ Ax= 0,\quad t\in (0,T],\quad x(0)= x_0,\quad x'(0)= 0,\tag{1} \] \[ {d^2x(t)\over dt^2}+ {k\over t} {dx(t)\over dt}+ Ax= 0,\quad t\in (0,T],\quad x(0)= x_0,\quad x'(0)= 0,\tag{2} \] where \(x:\mathbb{R}\to E\) is a vector-valued function and \(A\) is a densely defined closed operator with domain \(D(A)\) in a Banach space \(E\). To extend some of his own recent results, the author uses in this paper a class of strongly \(P\)-positive operators \(A\) whose spectral set is situated inside a parabola, and thus he gives a representation of the exact solution for which the representations from previous papers are special cases. More precisely, using the generating function of the Laguerre polynomials \(L^{(\alpha)}_n(t)\) \[ \exp\Biggl(-{wt\over 1-w}\Biggr)= \sum^\infty_{n= 0} L^{(\alpha)}_n(t) w^n(1- w)^{\alpha+ 1},\quad|w|< 1, \] assuming that \(A\) is strongly \(P\)-positive (i.e. has the spectral set situated inside of a parabola and the resolvent \((z- A)^{-1}\) satisfies a certain estimation) and \(x_0\in D(A^\sigma)\) with \(\sigma> 7/4\), then it is shown that the solution to problem (1) can be represented in the form \[ x(t)= e^{-\delta t} \sum^\infty_{n= 0} (L^{(0)}_n(t)- L^{(0)}_{n- 1}(t))u_n\quad (\delta>0),\tag{3} \] where the sequence of vectors \(\{u_n\}\) is defined by \[ u_{n+ 1}= 2(A+ \delta(\delta- 1)I)(A+ (\delta- 1)^2I)^{-1} u_n- (A+ \delta^2I)(A+ (\delta- 1)^2I)^{-1} u_{n- 1},\quad n\geq 1, \] \[ u_0= x_0,\quad u_1= (A- \delta(\delta- 1)I)(A+ (\delta- 1)^2I)^{-1} x_0. \] For \(\sigma\in\left({3\over 4},{7\over 4}\right)\), \(x(t)\) defined by (3) is a generalized solution. Moreover, the truncated sum \[ x^N(t)= e^{-\delta t} \sum^N_{n= 0} (L^{(0)}_n(t)- L^{(0)}_{n- 1}(t))u_n \] is an approximate solution to (1), i.e. \[ \sup_{t\in [0,T]} \|x^N(t)- x(t)\|\leq cN^{-(\alpha- 3/4-\varepsilon)} \|A^\sigma x_0\|, \] where \(\varepsilon\) is an arbitrarily small positive number and \(c\) is a constant independent of \(N\) and \(x_0\). In the case of the initial value problem (2), the author introduces an operator Bessel-function as a part of the solution and gives a representation of this function and of the solution to (2). Thus, the solution to (2) has, in the same hypotheses, the representation \[ x(t)= e^{-\delta t} \sum^\infty_{p=0} [\ell_p(t)- \ell_{p-1}(t)] u_p\quad (\delta> 0),\tag{4} \] with \[ \ell_p(t)= {1\over 2} \sum^p_{m=0}\widetilde\beta(p,m; k)L^{(0)}_{p- m}(t), \] \[ \widetilde\beta(p,m; k)= {p\choose m} \sum^m_{i= 0} {m\choose i}(-1)^i B\Biggl({p- m+ i+1\over 2},{k\over 2}\Biggr), \] \(B(\mu,\nu)= \Gamma(\mu)\Gamma(\nu)/\Gamma(\mu+ \nu)\) is the beta function and \(\Gamma\) is the gamma function. For \(\sigma\in\left({3\over 4},{7\over 4}\right)\) the function (4) is a generalized solution. A natural approximation to the solution \(x(t)\) in this case is the truncated sum \(x^N(t)= \sum^N_{p= 0}\ell_p(t) u_p\).
0 references
initial value problems
0 references
strongly \(P\)-positive operators
0 references
explicit representations of solutions
0 references
operator Bessel-function
0 references
0 references
0 references