Opened 11 years ago
Closed 11 years ago
#533 closed defect (fixed)
Floating point problems
| Reported by: | Juergen Reuter | Owned by: | ohl |
|---|---|---|---|
| Priority: | P3 | Milestone: | v2.2.0 |
| Component: | vamp | Version: | 2.1.1 |
| Severity: | normal | Keywords: | |
| Cc: |
Description
One of the integrations tests leads (with quadruple precision) to
|=============================================================================| | It Calls Integral[fb] Error[fb] Err[%] Acc Eff[%] Chi2 N[It] | |=============================================================================| At line 3036 of file vamp.f90 (unit = 14, file = 'integrations_4.r1.vg') Fortran runtime error: Bad value during floating point read
Change History (3)
comment:1 Changed 11 years ago by
comment:2 Changed 11 years ago by
F*ing problem is double_fmt of vamp :((((((( Grmpf ... can you do something against this, pronto!?
comment:3 Changed 11 years ago by
| Resolution: | → fixed |
|---|---|
| Status: | new → closed |
Allowing for exponents |ae|<104-1 instead of |ae|<103-1 in VAMP output. Doesn't hurt. And solves the quadruple problem (maybe defining two different output formats there (double_fmt and quad_fmt) and switching. But for now it's good to go. The problem was there were exponents greater than 999 for quadruple precision integration in the output of the VAMP grid files, such that the * entries failed reading them in again. Closing for now.
Note: See
TracTickets for help on using
tickets.
Ok, giving up.... no Bock, no time, no sonst was .... do what you like!