Real/Expr Package Ver.1.0


README for sample test programs


Graham:
     this example of Ron Graham compares two sums where each sum
     involves 9 square roots. The two sums agree up to 36 absolute digits.


Rump:
     this expression of Rump is supposed to show the unreliability of
     machine floating point calculations. The program simply evaluates
     a polynomial of degree 6.

     This sample program also uses the ``power'' pow() construct.


ORourke:
     O'Rourke computed the value of r(n, k) defined to be the minimum
     achieved by the absolute value of the expression:

          (\sqrt{a_1}+...+\sqrt{a_k}) - (\sqrt{b_1}+...+\sqrt{b_k})

     where a_i, b_i are positive integers at most n. The two
     expressions in this example is supposed to attain the values for
     r(20, 2) and r(20, 3), respectively.


det:
     computes the determinant of an n x n matrix by expanding it into
     the expression of n! terms.

     NOTE:
          It is less obvious how to implement the O(n^3) method of
          Gauss/Bareiss with Real/Expr.


tCompare, tExpr, tReal, tFloat:
     simple tests of the exact comparison operation, Expr, Real, and
     BigFloat.


WEB SITE:      http://simulation.nyu.edu/projects/exact/
BUG REPORT:    ouchi@simulation.nyu.edu
LAST UPDATE:   Dec, 31, 1996

