1: \begin{abstract}

2: Real root finding of polynomial equations is a basic problem in

3: computer algebra. This task is usually divided into two parts:

4: isolation and refinement. In this paper, we propose two algorithms

5: LZ1 and LZ2 to refine real roots of univariate polynomial equations.

6: Our algorithms combine Newton's method and the secant method to

7: bound the unique solution in an interval of a monotonic convex

8: isolation (MCI) of a polynomial, and have quadratic and cubic

9: convergence rates, respectively. To avoid the swell of coefficients

10: and speed up the computation, we implement the two algorithms by

11: using the floating-point interval method in \verb"Maple15" with the

12: package \verb"intpakX". Experiments show that our methods are

13: effective and much faster than the function \verb"RefineBox" in the

14: software \verb"Maple15" on benchmark polynomials.

15: \end{abstract}