1: \begin{abstract}

2: Computing rational minimax approximations can be very challenging

3: when there are singularities on or near the interval of approximation

4: --- precisely the case where rational functions outperform polynomials

5: by a landslide.  We show that far more robust algorithms than

6: previously available can be developed by making use of rational

7: barycentric representations whose support points are chosen in an

8: adaptive fashion as the approximant is computed.  Three variants of

9: this  barycentric strategy are all shown to be powerful:

10: (1) a classical Remez algorithm,

11: (2) a ``AAA-Lawson'' method of iteratively reweighted least-squares, and (3) a differential correction

12: algorithm.  Our preferred combination, implemented in the Chebfun

13: MINIMAX code, is to use (2) in an initial phase and then switch

14: to (1) for generically quadratic convergence.  By such methods we can

15: calculate approximations up to type $(80,80)$ of $|x|$ on $[-1,1]$

16: in standard 16-digit floating point arithmetic, a problem for which

17: Varga, Ruttan, and Carpenter required 200-digit extended precision.

18: 

19: 

20: \end{abstract}

21: