Closed GoogleCodeExporter closed 9 years ago
Exposing this for Sparse/CRS matrices seems to be quite messy. As it breaks
Ceres' assumptions about memory ownership and layout. Perhaps a good place to
start is to only expose dense jacobians/hessians. That way memory allocation
and layout is not an issue.
Original comment by sameerag...@google.com
on 2 Jul 2013 at 3:23
and if we are going to do a dense jacobian, then just having one cost function
is the way to go, since the rest of the solver does not care.
I am going to close this issue for now, since the user can basically do this
already for the dense case, and I do not see a compelling argument for the
sparse case yet.
Original comment by sameerag...@google.com
on 11 Aug 2013 at 10:09
Original issue reported on code.google.com by
sameerag...@google.com
on 10 May 2013 at 6:36