Improving productivity and reliability of labor-intensive high-performance software is an important scientific problem with obvious economic ramifications. This research is testing the feasibility of a radically new compilation and meta-compilation methodology to improve productivity and reliability for high-performance algorithmic software in the C programming language. The research combines partial evaluation for deeply-nested SETL data structures, data structure and algorithm selection techniques, and transformational techniques from relational query optimization. This is leading to a statically typed variant of SETL from which it is expected large-scale C programs can be generated at a high translation rate.***
