Re: new snapshot?

[John W. Eaton]

On 25-Feb-2010, John W. Eaton wrote:

| On 25-Feb-2010, Jaroslav Hajek wrote:
| 
| | The ARPACK failures are mysterious. Technically what has changed are
| | the 1D constructors and 1D resize. I scanned through eigs-base.cc and
| | eigs.cc but couldn't spot any problems. I'll try to compile a 64-bit
| | ARPACK and also give it a shot, but right now it's more important for
| | me to get the legends in FLTK backend.
| 
| Don't worry about the ARPACK problems.  I think I found the reason for
| the failure.

I checked in the following change:

  http://hg.savannah.gnu.org/hgweb/octave/rev/b4f67ca318d8

That fixed all the eigs failures for me, but I'm still seeing the
following failure in svds, which uses eigs:

  >>>>> processing /export/home/jwe/src/octave/scripts/sparse/svds.m
    ***** testif HAVE_ARPACK
   [u2,s2,v2,flag] = svds(a,k,0);
   s2 = diag(s2);
   assert(flag,!1);
   assert(s(k:-1:1), s2, 1e-10); 
  !!!!! test failed
  assert (s (k:-1:1),s2,1e-10) expected
     38.060
     38.060
     38.034
     38.034
     38.015
     38.015
     38.004
  but got
     38.060
     38.034
     38.034
     38.015
     38.015
     38.004
     38.004
  maximum absolute error 0.0263523 exceeds tolerance 1e-10

The test is:

  n = 100;
  k = 7;
  a = 
sparse([3:n,1:n,1:(n-2)],[1:(n-2),1:n,3:n],[ones(1,n-2),0.4*n*ones(1,n),ones(1,n-2)]);
  [u,s,v] = svd(full(a));
  s = diag(s);
  [dum, idx] = sort(abs(s));
  s = s(idx);
  u = u(:,idx);
  v = v(:,idx);
  randn('state',42)

  [u2,s2,v2,flag] = svds(a,k,0);
  s2 = diag(s2);
  assert(flag,!1);
  assert(s(k:-1:1), s2, 1e-10); 

If I run this test in a copy of Octave compiled *without* --enable-64,
I sometimes see also the error now.  The problem does not happen every
time I run the test.  So it seems like some kind of memory error, but
valgrind is not telling me anything.

I'm compiling with CXXFLAGS=-ggdb3 CFLAGS=-ggdb3, GCC version 4.4.3
20100108 (prerelease) (Debian 4.4.2-9) on an amd64 system.

Can anyone else duplicate this problem?

jwe