Problem Note 34545: A floating point zero divide error might occur in SAS 9.1.3 or SAS 9.2 for ALGORITHM=SIMPLEX in PROC QUANTREG
PROC QUANTREG with the ALGORITH=SIMPLEX option, which is the default, might produce a floating point zero divide error. The source of the error is that for the data at hand, the underlying linear program is very sensitive to small numerical perturbations and zero divide errors might occur at different quantiles on different hardware configurations. This error was initially reported in the experimental SAS 9.1.3 release, but may also occur in SAS9.2.
The log file may show the following error messages and note:
ERROR: Floating Point Zero Divide.
ERROR: Termination due to Floating Point Exception
NOTE: The SAS System stopped processing this step because of errors.
In some cases, placing all binary independent variables in the CLASS statement may circumvent the problem.
A Hot Fix is avaible for this problem; see the Hot Fix tab.
Operating System and Release Information
SAS System | SAS/STAT | z/OS | 9.1 TS1M3 | 9.2 TS2M0 |
Microsoft® Windows® for 64-Bit Itanium-based Systems | 9.1 TS1M3 | 9.2 TS2M0 |
Microsoft Windows Server 2003 Datacenter 64-bit Edition | 9.1 TS1M3 | 9.2 TS2M0 |
Microsoft Windows Server 2003 Enterprise 64-bit Edition | 9.1 TS1M3 | 9.2 TS2M0 |
Microsoft Windows XP 64-bit Edition | 9.1 TS1M3 | 9.2 TS2M0 |
Microsoft® Windows® for x64 | 9.1 TS1M3 | 9.2 TS2M0 |
Microsoft Windows 2000 Advanced Server | 9.1 TS1M3 | |
Microsoft Windows 2000 Datacenter Server | 9.1 TS1M3 | |
Microsoft Windows 2000 Server | 9.1 TS1M3 | |
Microsoft Windows 2000 Professional | 9.1 TS1M3 | |
Microsoft Windows NT Workstation | 9.1 TS1M3 | |
Microsoft Windows Server 2003 Datacenter Edition | 9.1 TS1M3 | 9.2 TS2M0 |
Microsoft Windows Server 2003 Enterprise Edition | 9.1 TS1M3 | 9.2 TS2M0 |
Microsoft Windows Server 2003 Standard Edition | 9.1 TS1M3 | 9.2 TS2M0 |
Microsoft Windows XP Professional | 9.1 TS1M3 | 9.2 TS2M0 |
Windows Vista | 9.1 TS1M3 | 9.2 TS2M0 |
64-bit Enabled AIX | 9.1 TS1M3 | 9.2 TS2M0 |
64-bit Enabled HP-UX | 9.1 TS1M3 | 9.2 TS2M0 |
64-bit Enabled Solaris | 9.1 TS1M3 | 9.2 TS2M0 |
HP-UX IPF | 9.1 TS1M3 | 9.2 TS2M0 |
Linux | 9.1 TS1M3 | 9.2 TS2M0 |
OpenVMS Alpha | 9.1 TS1M3 | 9.2 TS2M0 |
Tru64 UNIX | 9.1 TS1M3 | 9.2 TS2M0 |
*
For software releases that are not yet generally available, the Fixed
Release is the software release in which the problem is planned to be
fixed.
PROC QUANTREG with the ALGORITH=SIMPLEX option, which is the default, might produce a floating point zero divide error.
Type: | Problem Note |
Priority: | high |
Topic: | Analytics ==> Regression
|
Date Modified: | 2009-02-03 16:32:02 |
Date Created: | 2009-01-19 17:11:56 |