Application of the method of \(R\)-functions to integration of differential equations with partial derivatives (Q1968502)

The \(R\)-functions method (RFM) is developed in this article and applied to the integration of boundary value problems of the form \[ Au =f\text{ in }\Omega \tag{1} \] \[ L_iu= \varphi_i\text{ on }\partial \Omega_i\;(i= 1,2, \dots,m),\tag{2} \] where \(\partial\Omega_i\) \((i=1,2, \dots,m)\) is a covering of the boundary \(\partial\Omega\) of the domain \(\Omega,f\) and \(\varphi_i\) are known functions. The geometric information is taken into account by the functions \(\omega\) and \(\omega_i\) selected from elementary functions so that \(\partial\Omega\) and \(\partial\Omega_i\) are sets of their zeros. A solution \(u\) of the problem (1),(2) is represented as \[ u=B(\varphi, \omega, \omega_i)+ \varphi_0, \tag{3} \] where \(\varphi_0\) is a known function, \(B\) is an operator dependent on the form of the boundary \(\partial\Omega\) and its parts \(\partial \Omega_i\) \((i=1,\dots,m)\), so that for any choice of an uncertain component \(\Phi\) from some set \(M\) formula (3) satisfies the boundary conditions (2). If there is a possibility of such a choice of the uncertain component which transforms (3) into a solution of the problem with any given accuracy, then it is called the general structure of solution (GSS) of the problem considered. In order to construct the functions \(\omega\) and \(\omega_i\) appearing in formulas of the form (3), the following \(R\)-functions are used in most cases: \[ x \wedge_\alpha y={1 \over 1+\alpha} \bigl(x+y- \sqrt{x^2+ y^2-2 \alpha xy} \bigr), \] \[ x\vee_\alpha y={1\over 1+\alpha} \bigl(x+y+ \sqrt{x^2+ y^2-2 \alpha xy} \bigr), \] \[ \overline x=-x,\;x \sim y={xy \over x+y},\;x\overset {n}\sim y={xy\over \root n\of {x^n+y^n}}, \] where \(\alpha=\alpha (x,y)\), \(-1<\alpha\leq 1\), \(\alpha= 0;1; 0;q\) or \(\alpha= (1+x^2 +y^2)^{-1}\). The authors give GSSs, for some boundary value problems. Thus, for the Dirichlet boundary value problem \(u=\varphi_i\) on \(\partial\Omega_i\) the proper GSS can be represented in the form \[ u={\omega_1^{-1} \varphi_1+ \cdots+ \omega_m^{-1} \varphi_m\over \omega_1^{-1}+ \cdots+ \omega^{-1}_m}+\omega \Phi, \] for the Neumann boundary value problem \({\partial u\over\partial \nu}=\varphi\) on \(\partial \Omega\) a proper GSS can be represented in the form \[ u=\varphi_1-\omega\left( {\partial\omega \over\partial x}{\partial \Phi_1\over \partial x}+{\partial \omega\over \partial y}{\partial \Phi_1\over \partial y}+{\partial \omega \over\partial z}{\partial \Phi_1\over \partial z}\right)+ \omega^2 \varphi_2 +\omega\varphi \] and for problem with mixed boundary conditions of the form \(u=\varphi\) on \(\partial\Omega_1\) \({\partial u\over\partial \nu}=\psi\) on \(\partial \Omega_2\) a GSS can be represented in the form \[ u=\omega_1 \Phi_1+ {\omega_1\omega_2 \over\omega_1 +\omega_2} \bigl[\psi+ \omega_2 \Phi_2-D_1^{(2)} (\omega_1\Phi_1) -D_1^{(2)}\varphi \bigr]+ \varphi, \] where \[ D_1^{(2)}= {\partial\omega_2 \over\partial x}{\partial \over\partial x}+ {\partial \omega_2\over \partial y}{\partial \over\partial y}+ {\partial \omega_2\over \partial z}{\partial \over\partial z}. \]