sumdotpack

This small package computes sums and dots product using quadruple precision accumulation or such that allmost full double precision is achieved whatever the condtion number of the computation may be. There are 4 routines qsum, fpsum, qdot & fpdot plus 2 mains for testing purpose.
If symbolic toolbox is available results using symbolic computation is given in the test programs.

This routines look like some Xblas modules and are usually to be used to perform iterative refinement in linear algebra problems. Used mainly for pedagogical purpose.