Re: [OctDev] [i,j,v] = find(sparse(1:10,1:10,1:10,11,11))

[Etienne G. <et...@cs...>]

  Viva 2.9.3!

(in fact 2.9.2). Thanks a lot David, I didn't even know there was an
"experimental" distribution of Debian packages.

  find(sparse(...)) now works on Machine 1, I will ask our sysad to
install 2.9.X on Machine 2. I guess that solves it.

  Thanks again,

  Etienne

On Wed, Aug 24, 2005 at 03:47:00PM +0200, David Bateman wrote:
# Etienne Grossmann wrote:
# 
# > Hi all,
# >
# >on two different Debian machines, I get:
# >
# >======================================================================
# >octave:3> [i,j,v] = find (sparse ([1 2 3 2],[2 4 2 5],1:4,30000,50000));
# >error: octave_base_value::array_value(): wrong type argument `sparse'
# >======================================================================
# >
# > Moreover on Machine 1, 
# >
# >======================================================================
# >octave:2> [i,j,v] = spfind(sparse(1:10,1:10,1:10,11,11))
# >panic: Segmentation fault -- stopping myself...
# >attempting to save variables to `octave-core'...
# >save to `octave-core' complete
# >======================================================================
# >
# > Whereas w/ Machine 2, I get:
# >
# >======================================================================
# >octave:5> [i,j,v] = spfind(sparse(1:10,1:10,1:10,11,11))
# >[snip] ... normal output ...
# >======================================================================
# >
# >
# > 
# >
# Strange the octave-forge sparse stuff for me working perfectly for these.
# 
# > I have octave 2.1.71 from debian (octave2.1-headers and octave2.1)
# >and a fresh CVS octave-forge (checkout, autogen.sh, config, make) in
# >both cases.
# >
# > I imagine there's @ least a dispatch issue, but don't know enough
# >about dispatch to find it.
# >
# > I would greatly appreciate if anyone had an explanation/fix.
# >
# > 
# >
# However, they also work perfectly in 2.9.3 and the sparse stuff there... 
# The only thing in the octave-forge sparse stuff that isn't in 2.9.3 is 
# the spinv and the iterative functions. The iterative functions are 
# implemented as dot-m in octave-forge and so you can copy them, and I'm 
# not sure where exactly you'd want to use spinv as the inverse of a 
# sparse matrix is full. So I consider 2.9.3 to be in a good state to be 
# considered as a replacement of the octave-forge sparse code. It is also 
# in debian experimental so you could use pinning to get 2.9.3 under 
# debian if you like...
# 
# The fact is with the new sparse stuff in octave I don't think its likely 
# any bugs in the octave-forge will be tracked down...
# 
# Regards
# David
# 
# -- 
# David Bateman                                Dav...@mo...
# Motorola Labs - Paris                        +33 1 69 35 48 04 (Ph) 
# Parc Les Algorithmes, Commune de St Aubin    +33 1 69 35 77 01 (Fax) 
# 91193 Gif-Sur-Yvette FRANCE
# 
# The information contained in this communication has been classified as: 
# 
# [x] General Business Information 
# [ ] Motorola Internal Use Only 
# [ ] Motorola Confidential Proprietary
# 

-- 
Etienne Grossmann ------ http://www.cs.uky.edu/~etienne