[Octave-bug-tracker] [bug #36656] deleting rows from sparse matrices leads to failed assertion sometimes

[David Bateman]

Follow-up Comment #6, bug #36656 (project octave):

Jordi,

Deep or not I think I understand the code. Yes my comment was a throw away
that I wrote rapidly and I misread cidx as ridx. So taking your comment as a
challenge, here is my analysis of the situation

In Sparse<T>::delete_elements when idx_j is a colon then the creation of both
tmpl and tmpu will first call index(idx_i,idx_j) before first falling back to
index(idx_i) as nc==1 and j is a colon. This calls the code


 Sparse<T> (idx_dims(0), idx_dims(1), nz ? data(0) : T ());


to create the index vector. As idx_i (as for all index vectors) is a column
vector, then the index that is returned to create tmpu and tmpl in
delete_elements is also a column vector and we were expecting it to be a row
vector. 

The basic problem is that when nr=nc==1 then index(idx_j) can't know the shape
of the desired return vector and so guessing its a column vector is
reasonable. Whereas index(idx_i,idx_j) knows what shape it should be. The fix
is in index(idx_i,idx_j) to transpose retval if nc=nr==1 and j is a colon.

I still think its a good idea to make the change I suggested previously in any
case to avoid issues if the vectors data and ridx are not defined. So my
proposed fix if


diff --git a/liboctave/Sparse.cc b/liboctave/Sparse.cc
--- a/liboctave/Sparse.cc
+++ b/liboctave/Sparse.cc
@@ -1583,6 +1583,10 @@
     {
       // It's actually vector indexing. The 1D index is specialized for
that.
       retval = index (idx_i);
+      // If nr == 1 then the vector indexing will return a column vector!!
+      if (nr == 1)
+        retval.transpose ();
+        
     }
   else if (idx_i.is_scalar ())
     {
diff --git a/liboctave/Sparse.h b/liboctave/Sparse.h
--- a/liboctave/Sparse.h
+++ b/liboctave/Sparse.h
@@ -81,14 +81,7 @@
           c[i] = 0;
       }
 
-    SparseRep (octave_idx_type nr, octave_idx_type nc) : d (0), r (0), c (new
octave_idx_type [nc+1]), nzmx (0),
-      nrows (nr), ncols (nc), count (1)
-      {
-        for (octave_idx_type i = 0; i < nc + 1; i++)
-          c[i] = 0;
-      }
-
-    SparseRep (octave_idx_type nr, octave_idx_type nc, octave_idx_type nz) :
d (new T [nz]),
+    SparseRep (octave_idx_type nr, octave_idx_type nc, octave_idx_type nz=0)
: d (new T [nz]),
       r (new octave_idx_type [nz]), c (new octave_idx_type [nc+1]), nzmx
(nz), nrows (nr),
       ncols (nc), count (1)
       {



D.


    _______________________________________________________

Reply to this item at:

  <http://savannah.gnu.org/bugs/?36656>

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