Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]


Groups > comp.soft-sys.math.maple > #253

Re: LinearSolve, BackwardSubstitute(GaussianElimination) - different answers?

From acer <maple@rogers.com>
Newsgroups comp.soft-sys.math.maple
Subject Re: LinearSolve, BackwardSubstitute(GaussianElimination) - different answers?
Date 2011-09-25 20:35 -0700
Organization http://groups.google.com
Message-ID <15947162.1966.1317008101833.JavaMail.geo-discussion-forums@yqjc18> (permalink)
References <ea40abbb-3b88-4778-9381-983cb4019475@i28g2000yqn.googlegroups.com>

Show all headers | View raw


In the `LinearSolve` case, debugging can reveal that only the first two entries of the third column are being passed to the external solver. The third entry of the third column (which is the RHS of the augmented system) is passed as the value zero. And so if the external solver happens to use the second and third rows of `a` (instead of, say, the first and second) then that incorrect solution is found.

In other words, LinearSolve is computing the solution to this problem instead:

restart:
with(LinearAlgebra):
a:=<<3100|6400|23610>,<250|360|0>>;
                               [3100  6400  23610]
                          a := [                 ]
                               [ 250   360      0]

evalf(LinearSolve(a));
                               [-17.56115702]
                               [            ]
                               [ 12.19524793]


It looks like a coding error in LinearSolve, possibly made after computing the rank of `a` (which is 2, for your example).

Back to comp.soft-sys.math.maple | Previous | NextPrevious in thread | Next in thread | Find similar


Thread

LinearSolve, BackwardSubstitute(GaussianElimination) - different answers? lasalle25 <betty.n.love@gmail.com> - 2011-09-25 17:46 -0700
  Re: LinearSolve, BackwardSubstitute(GaussianElimination) - different answers? acer <maple@rogers.com> - 2011-09-25 20:35 -0700
  Re: LinearSolve, BackwardSubstitute(GaussianElimination) - different answers? Peter Pein <petsie@dordos.net> - 2011-09-26 11:48 +0200
    Re: LinearSolve, BackwardSubstitute(GaussianElimination) - different answers? Ray Vickson <RGVickson@shaw.ca> - 2011-11-04 15:03 -0700

csiph-web