Using ADIFOR and ADIC to provide Jacobians for the SNES component of PETSc

作者: Po-Ting Wu , C.H. Bischof , P.D. Hovland

DOI: 10.2172/567514

关键词: Computational scienceComputer scienceMatrix (mathematics)Automatic differentiationSubroutineNonlinear systemTheoretical computer scienceComponent (UML)Jacobian matrix and determinant

摘要: The solution of large-scale nonlinear problems is important to many areas computational science. SNES component PETSc provides a robust and flexible suite numerical routines for the solving such problems. These generally utilize Jacobian matrix. We present strategy using ADIFOR or ADIC assist in development subroutine computing this illustrate one example programs four different approaches via automatic differentiation.

参考文章(4)
C. Bischof, P. Khademi, A. Mauer, P. Hovland, A. Carle, ADIFOR 2.0 user`s guide (Revision B) Other Information: PBD: Apr 1995. ,(1995) , 10.2172/93483
Satish Balay, William D. Gropp, Lois Curfman McInnes, Barry F. Smith, Efficient Management of Parallelism in Object-Oriented Numerical Software Libraries Modern Software Tools for Scientific Computing. pp. 163- 202 ,(1997) , 10.1007/978-1-4612-1986-6_8
Brett M. Averick, Jorge J. Moré, Christian H. Bischof, Alan Carle, Andreas Griewank, Computing large sparse Jacobian matrices using automatic differentiation SIAM Journal on Scientific Computing. ,vol. 15, pp. 285- 294 ,(1994) , 10.1137/0915020
C. Bischof, P. Khademi, A. Mauer, A. Carle, Adifor 2.0: automatic differentiation of Fortran 77 programs computational science and engineering. ,vol. 3, pp. 18- 32 ,(1996) , 10.1109/99.537089