A New Augmented Singular Transform and its Partial Newton-Correction Method for Finding More Solutions Academic Article uri icon

abstract

  • 2016, Springer Science+Business Media New York. With the information provided by previously found solutions, an augmented singular transform is introduced in Xie etal. (JCAM 286:145157, 2015) to change the local basin structure of the original problem for finding new solutions. However the old formulation in Xie etal. (JCAM 286:145157, 2015) involves the kernel of an unknown solution to be found or the kernels of all previously found solutions, thus left several theoretical issues unsolved and prevents from further development. In this paper, we derive a new augmented singular transform which changes only the local basin/barrier structure around u= 0 for finding more solutions. Comparing to the old formulation, the new one is much easier to apply and resolves all unsolved theoretical issues left in Xie etal. (JCAM 286:145157, 2015). A corresponding partial Newton-correction method is then designed to solve the augmented problem on the solution set. Mathematical justification of the new formulation, method and its local convergence are established. The new method is first tested on two very different variational problems and then applied to solve a nonvariational nonlinear convection-diffusion equation for multiple solutions, which are, for the first time, numerically computed and visualized with their profile and contour plots. Several interesting phenomena are observed for the first time and open for mathematical verification. Since the new formulation is general and simple, it can also be modified to treat other problems, e.g., quasilinear PDEs, a large system of PDEs with equality constraints, for finding multiple solutions.

published proceedings

  • Journal of Scientific Computing

author list (cited authors)

  • Li, Z., Wang, Z., & Zhou, J.

citation count

  • 4

complete list of authors

  • Li, Zhaoxiang||Wang, Zhi-Qiang||Zhou, Jianxin

publication date

  • May 2017