On Sep 11, 2014, at 10:06 AM, Grey Gordon <[email protected]> wrote: > An alternative would be to map the problem into double precision. For > instance, if you need quad precision for some length n vector, you could use > a length 2n double precision vector. Then the 1:n section of the vector would > contain the first decimal places and the n+1:2n section of the vector could > contain the last decimal places (or some similar solution).
A good way to do this is called “double-double” arithmetic, and you can find lots of libraries that implement this for you. _______________________________________________ NLopt-discuss mailing list [email protected] http://ab-initio.mit.edu/cgi-bin/mailman/listinfo/nlopt-discuss
