MathGroup Archive 2000

[Date Index] [Thread Index] [Author Index]

Search the Archive

PentadiagonalSolve?

  • To: mathgroup at smc.vnet.net
  • Subject: [mg22241] PentadiagonalSolve?
  • From: Kurt Taretto <Kurt.Taretto at ipe.uni-stuttgart.de>
  • Date: Sat, 19 Feb 2000 01:33:40 -0500 (EST)
  • Organization: Comp.Center (RUS), U of Stuttgart, FRG
  • Sender: owner-wri-mathgroup at wolfram.com

Solving a system of linear equations with a matrix with 5 diagonals is
done pretty well by LinearSolve.  But since I work with large matrices,
and all of them have only 5 nonzero diagonals, I don't have to use
memory for the nonzero entries in order to give LinearSolve a "good and
square" matrix.
Does anybody know how could I, for example, do some transformation from
5 to 3 diagonals and then use TridiagonalSolve?  I would
appreciate any help.  Thanks!



  • Prev by Date: AW: Method of Characteristics
  • Next by Date: RE: A trig identity that Mathematica doesn't know?
  • Previous by thread: Re: AW: Method of Characteristics
  • Next by thread: Re: PentadiagonalSolve?