### Current browse context:

math.NA

### Change to browse by:

### References & Citations

# Mathematics > Numerical Analysis

# Title: Average-case analysis of the Gaussian Elimination with Partial Pivoting

(Submitted on 22 May 2022 (v1), last revised 15 Jun 2022 (this version, v2))

Abstract: The Gaussian Elimination with Partial Pivoting (GEPP) is a classical algorithm for solving systems of linear equations. Although in specific cases the loss of precision in GEPP due to roundoff errors can be very significant, empirical evidence strongly suggests that for a {\it typical} square coefficient matrix, GEPP is numerically stable. We obtain a (partial) theoretical justification of this phenomenon by showing that, given the random $n\times n$ standard Gaussian coefficient matrix $A$, the {\it growth factor} of the Gaussian Elimination with Partial Pivoting is at most polynomially large in $n$ with probability close to one. This implies that with probability close to one the number of bits of precision sufficient to solve $Ax = b$ to $m$ bits of accuracy using GEPP is $m+O(\log n)$, which improves an earlier estimate $m+O(\log^2 n)$ of Sankar, and which we conjecture to be optimal by the order of magnitude. We further provide tail estimates of the growth factor which can be used to support the empirical observation that GEPP is more stable than the Gaussian Elimination with no pivoting.

## Submission history

From: Konstantin Tikhomirov [view email]**[v1]**Sun, 22 May 2022 22:12:07 GMT (57kb)

**[v2]**Wed, 15 Jun 2022 00:33:14 GMT (57kb)

Link back to: arXiv, form interface, contact.