Approximate computation of an eigenwave from measurements of its amplitudes at a given point (Q404693)

The authors consider the equation \[ \begin{aligned}\varphi^{(2)}+ a_1(t) \varphi^{(1)}+ a_2(t) \varphi= \lambda\varphi,\quad & \varphi\in 0,\\ \sum_{\alpha\in F} C_\alpha(x,y,z)\psi^{(\alpha)}(x,y,z)= -\lambda\psi(x,y,z),\quad & \psi(x,y,z)\not\equiv 0,\end{aligned}\tag{1} \] where \(a_1\), \(a_2\), \(C_\alpha\) are real functions. Assume that \(\cdots< c_{-1}> c_0< c_1<\cdots\) are points of nonintegrability of the functions \(a_1\), \(a_2\), \(a_1\in L^1_{\text{loc}}(c_j, c_{j+1})\), \(a_2- a^{(1)}_1\in L^1_{\text{loc}}(c_j, c_{j+1})\) for all \(j\), which permits to rewrite (1) in the form \[ (\varphi^{(1)}+ a_1(t) \varphi)^{(1)}+ (a_2(t)- a^{(1)}_1(t)) \varphi= \lambda\varphi.\tag{2} \] Under the assumption that the variables in the wave equation can be separated and its coefficients are periodic, the authors develop a classification of seismic eigenwaves. A function \(\varphi\) absolutely continuous on each closed interval is called a solution of (2), and hence of (1), if the function \(\varphi^{(1)}+ a_1\varphi\) is absolutely continuous on each closed interval and (2) holds almost everywhere. Let the functions \(a_j\) be \(T\)-periodic on \((-b,b)\), \(b\geq 2^{-1}T\), and consider the equations \[ u^{(1)}(t)= \varepsilon f(t) u(-t)+ g(t) u(t),\quad t\in (a,b),\tag{3} \] \[ v^{(1)}(t)=\varepsilon f(t)\exp\Biggl(\int^{-t}_t (g(t)- g_0(t))\,dt\Biggr)v(-t)+ g_0(t) v(t),\quad t\in (-b,b),\tag{4} \] where \(\varepsilon\) is equal either to \(1\) or \(-1\), and \(f\in L^1_{\text{loc}}(-b,b)\), \(g\in L^1_{\text{loc}}(-b,b)\). The authors suppose that \(f\in A(-b,b)\), where \(A(-b,b)\) is the set of absolutely continuous functions on the interval \((-b,b)\). Then, if \(u\) and \(v\) are solutions of (3), (4) respectively, the authors obtain the expression of \(u^{(2)}\) and \(v^{(2)}\), in Theorem 1, that admits a converse. Moreover, a function \(u\) is a solution of (3) if and only if \((u(t), u(-t))\) is a solution of the problem \[ \begin{aligned} u^{(1)}_1(t) &=\varepsilon f(t)u_2(t)+ g(t)u_1(t),\\ u^{(1)}_2(t) &= -\varepsilon f(-t)u_1(t)- g(-t) u_2(t),\\ u_1(t) &= u_2(-t).\end{aligned} \] The authors make remarks on the problems given above and use the classification of seismic eigenwaves to answer some questions as to how to specify the type and basic parameters of a wave on the basis of measurements of amplitudes, whether there exist points of chaos, and how to predict them.