#152 closed defect (fixed)
TOL Estimation seg faults when there is less data than variables.
| Reported by: | manuelb | Owned by: | manuelb |
|---|---|---|---|
| Priority: | highest | Milestone: | |
| Component: | Math | Version: | head |
| Severity: | major | Keywords: | |
| Cc: |
Description
When I execute the given estimation, TOL gives this message:
gsl: svd.c:227: ERROR: svd of MxN matrix, M<N, is not implemented
Default GSL error handler invoked.
And then Seg faults.
Attachments (6)
Change History (11)
comment:1 Changed 21 years ago by
| Owner: | changed from Víctor de Buen Remiro to manuelb |
|---|
comment:2 Changed 21 years ago by
| Status: | new → assigned |
|---|
Changed 21 years ago by
| Attachment: | estimation.tol added |
|---|
comment:3 Changed 21 years ago by
| Resolution: | → fixed |
|---|---|
| Status: | assigned → closed |
The problem is that there are more variables than data.
Eventhough it is statistically wrong, that is actually possible, but must be avoided
because the estimation computes the singular value decomposition using the GSL
modified Golub-Reinsch algorithm. This algorithm is very optimized for N>M, (that
is, many more data fields than variables), and has not even been implemented for
N<M.
A fix for this is on the CVS: Prints an error and aborts the estimation.
Changed 21 years ago by
| Attachment: | estimation.tar.gz added |
|---|
gziped file including all sources and data files
comment:4 Changed 21 years ago by
| attachments.isobsolete: | 0 → 1 |
|---|
comment:5 Changed 19 years ago by
| bug_file_loc: | → http://www.tol-project.org |
|---|
Note: See
TracTickets for help on using
tickets.
TOL source file to test this issue