A package with linear solvers, preconditioners, and nonlinear solvers. These were developed during my course sequence on iterative solvers offered at Texas Tech in 2020-21.

• Linear solvers
  • Preconditioned GMRES
  • Preconditioned conjugate gradients (PCG)
  • Algebraic multigrid (AMG)
    • The initial version of the AMG code was written by Nick Moore as part of his course project.
  • Unified user interface to iterative and direct solvers
• Preconditioners
  • Incomplete ILU with drop tolerance (based on the numpy interface to SuperLU)
  • Incomplete Cholesky with drop tolerance (based on the numpy interface to SuperLU)
  • Algebraic multigrid (AMG)
• Nonlinear solvers
  • Newton's method
    • Inexact solves of linear subproblems with adaptive tolerance
    • Optional reuse of preconditioner for multiple linear subproblems
    • Line search to find sufficient decrease