Common Errors in Calling LAPACK Routines

As a first step, the user should always carefully read the leading comments of the LAPACK routine. The leading comments give a detailed description of all input/output arguments and can be viewed in the source code, on the LAPACK webpage, or in this users' guide.

For the benefit of less experienced programmers, we list the most common programming errors in calling an LAPACK routine. These errors may cause the LAPACK routine to report a failure, as described in Section 7.3; they may cause an error to be reported by the system; or they may lead to wrong results -- see also Section 7.4.

• wrong number of arguments
• arguments in the wrong order
• an argument of the wrong type (especially real and complex arguments of the wrong precision)
• wrong dimensions for an array argument
• insufficient space in a workspace argument
• failure to assign a value to an input argument

Some modern compilation systems, as well as software tools such as the portability checker in Toolpack [88], can check that arguments agree in number and type; and many compilation systems offer run-time detection of errors such as an array element out-of-bounds or use of an unassigned variable.