1: \begin{abstract}

2: 

3: Given an approximation to a multiple isolated solution of a polynomial

4: system of equations, we have provided a symbolic-numeric deflation

5: algorithm to restore the quadratic convergence of Newton's method.

6: Using first-order derivatives of the polynomials in the system,

7: our method creates an augmented system of equations which has the

8: multiple isolated solution of the original system as a regular root.

9: 

10: In this paper we consider two approaches to computing the

11: ``multiplicity structure'' at a singular isolated solution.

12: An idea coming from one of them gives rise to our new

13: higher-order deflation method.

14: Using higher-order partial derivatives of the original polynomials,

15: the new algorithm reduces the multiplicity faster than our first method

16: for systems which require several first-order deflation steps.

17: 

18: We also present an algorithm to predict the order of the deflation.

19: 

20: \noindent {\bf 2000 Mathematics Subject Classification.} Primary

21: 65H10.  Secondary 14Q99, 68W30.

22: % 14Q99 Computational aspects in algebraic geometry

23: % 65H10 Systems of nonlinear equations

24: % 68W30 Symbolic Computation and algebraic computation

25: 

26: \noindent {\bf Key words and phrases.} Deflation,

27: isolated singular solutions, Newton's method, multiplicity,

28: polynomial systems, reconditioning, symbolic-numeric computations.

29: 

30: \end{abstract}

31: