[Octave-bug-tracker] [bug #47415] out of memory negating a permutation matrix

[Carlo de Falco]

Follow-up Comment #27, bug #47415 (project octave):

jwe,

also from the point of view of maintaining compatibility
w.r.t. to -0 and -Inf, the conversion to sparse appears to be more correct in
this case (factorization reorderings) than a conversion to full.

In Matlab 2015b I see the following:

>> A = sprandn (100, 100, .02);
>> [L, U, P, Q, R] = lu (A);
>> 1./(-P(1:2,1:2))

ans =

   Inf   Inf
   Inf   Inf

>> 


In Octave 4.1 the same code returns

>> A = sprandn (100, 100, .02);
>> [L, U, P, Q, R] = lu (A);
>> 1./(-P(1:2,1:2))
ans =

  -Inf  -Inf
  -Inf  -Inf


while the conversion to sparse would lead to the following

>> 1./(-sparse(P(1:2,1:2)))
ans =

   Inf   Inf
   Inf   Inf


honestly the idea of using n^2 instead of ~n memory to store
a lot of '-0's sounds a bit exotic to me, if we really cared
so much about -0 we could add a signbit property to SparseMatrix
as well, but is that really worth the effort??



    _______________________________________________________

Reply to this item at:

  <http://savannah.gnu.org/bugs/?47415>

_______________________________________________
  Message sent via/by Savannah
  http://savannah.gnu.org/