An Assessment of the Efficiency of Computer Algebra Systems in the Solution of Scientific Computing Problems

Computer algebra systems (CASs) have become an important tool for the solution of scientific computing problems. With the increasing number of general purpose CASs, there is now a need for an assessment of the efficiency of these systems. We discuss some peculiarities associated with the analysis of CPU time efficiency in CASs, and then present results from three specific systems (Maple Vr5, Mathematica 4.0 and MuPAD 1.4) on a sample of intermediate size problems. These results show that Maple Vr5 is generally the speediest on our examples. Finally, we formulate some requirements for developing a comprehensive test suite for analyzing the efficiency of CASs.