On the density of the Ward Ansätze in the space of anti-self-dual Yang Mills solutions (Q753173)

Many of the familiar nonlinear equations of mathematical physics (for example the Bogomolny equation, the sine and sinh-Gordon equations, the Toda-lattice equation, the stationary axial symmetric Einstein and Einstein-Maxwell equations, the KdV equation, the nonlinear Schrödinger equation, and others) are symmetry reductions of the anti-self-dual Yang- Mills (ASDYM) equations for various gauge groups. The ASDYM equations fall into the class of differential equations which are usually described as integrable or solvable [see \textit{R. S. Ward}, Philos. Trans. R. Soc. Lond., A 315, 451-457 (1985; Zbl 0579.35078)]. Many solution generating techniques have been developed for these equations, including the matrix- valued Sparling equation, the use of Lax pairs, and the twistor construction where twistor-data is freely prescribed and the solution of a classical Riemann-Hilbert (RH)-problem is required in order to obtain the anti-self-dual field. In the reviewed paper, it is shown how arbitrary matrix-valued data for the twistor construction can be approximated by data which can then be triangularized; the explicit solution of this approximate data can then be given by quadratures. This exact solution from the approximate data, as this data approaches the exact data, converges to the solution of the exact data. In Section II, the authors review the ASDYM equations and describe the notation. The twistor construction for generating solutions is briefly introduced and the properties (gauge freedoms) of this construction is described. These gauge freedoms are extensively used in the approximation and triangularization scheme of Sections IV and VI. Section III contains an explicit method for obtaining the solutions of the ASDYM equations via RH splitting (twistor construction) for triangular data (for an arbitrary gauge group). In Section IV, the procedure for approximating arbitrary SL(2,C) data which can then be explicitly triangularized is described; the solution to this triangularized data can then be obtained by employing the method given in Section III. In Section V, the solutions to the approximate data are shown to converge to the solutions of the exact data as the approximate data converge to the exact data. Finally the procedure described in Section IV is generalized to arbitrary SL(n,C) data in Section VI.