Computer Science > Numerical Analysis
[Submitted on 30 May 2018 (v1), last revised 13 Feb 2024 (this version, v35)]
Title:Good News for Polynomial Root-finding
View PDF HTML (experimental)Abstract:Univariate polynomial root-finding has been studied for four millennia and very intensively in the last decades. We approximate all complex roots as well as the roots in a fixed domain of the complex plane. We propose and extensively analyze various novel techniques, which enable significant acceleration of the known algorithms where a polynomial p is given by an oracle, that is, black box subroutine, for its evaluation and does not need to involve its coefficients. Our polynomial root-finders, however, are highly efficient and even nearly optimal where a polynomial given with their coefficients as well. This is the first time when the record breaking and nearly optimal Boolean complexity bound of our algorithm at STOC 1995 was matched with another root-finder. Unlike its deterministic predecessor of 1995, our new root-finders use Las Vegas randomization but run faster for black box input polynomials and are much simpler for implementation. Already in their initial implementation they have at least competed with user's choice package of root-finding subroutines MPSolve, according to extensive numerical experiments with standard test polynomials. Moreover, they significantly superseded MPSolve as degree of p increased. Our auxiliary algorithms and estimates for black box polynomial computations, e.g., for the approximation of root radii, that is, the distances from a complex point to the roots, for compression of a disc on the complex plane without losing the zeros of an input polynomial, for acceleration of bisection root-finders, for detection of output errors of polynomial root-finders, and for fast approximation of both real and nearly real roots as well as our use of randomization for polynomial root-finding can be of independent interest.
Submission history
From: Victor Pan [view email][v1] Wed, 30 May 2018 15:46:07 UTC (22 KB)
[v2] Mon, 17 Dec 2018 22:33:58 UTC (27 KB)
[v3] Mon, 28 Jan 2019 20:30:30 UTC (27 KB)
[v4] Sun, 10 Feb 2019 22:36:16 UTC (28 KB)
[v5] Mon, 8 Apr 2019 14:53:11 UTC (35 KB)
[v6] Thu, 23 May 2019 15:33:44 UTC (35 KB)
[v7] Wed, 29 May 2019 02:45:34 UTC (36 KB)
[v8] Sat, 6 Jul 2019 11:58:08 UTC (44 KB)
[v9] Mon, 23 Sep 2019 18:36:36 UTC (90 KB)
[v10] Mon, 11 Nov 2019 20:36:37 UTC (91 KB)
[v11] Sat, 25 Jan 2020 12:07:51 UTC (92 KB)
[v12] Tue, 4 Feb 2020 23:57:23 UTC (94 KB)
[v13] Fri, 20 Mar 2020 12:20:34 UTC (102 KB)
[v14] Sat, 4 Apr 2020 02:26:36 UTC (97 KB)
[v15] Mon, 18 May 2020 02:09:25 UTC (101 KB)
[v16] Sun, 13 Sep 2020 22:09:35 UTC (88 KB)
[v17] Wed, 16 Dec 2020 19:56:04 UTC (100 KB)
[v18] Sun, 31 Jan 2021 21:53:07 UTC (106 KB)
[v19] Sun, 28 Feb 2021 16:33:30 UTC (107 KB)
[v20] Thu, 1 Jul 2021 03:10:00 UTC (111 KB)
[v21] Wed, 1 Sep 2021 03:42:26 UTC (110 KB)
[v22] Sun, 30 Jan 2022 20:46:40 UTC (130 KB)
[v23] Tue, 1 Mar 2022 00:44:57 UTC (136 KB)
[v24] Wed, 30 Mar 2022 13:20:21 UTC (141 KB)
[v25] Fri, 29 Apr 2022 17:21:12 UTC (145 KB)
[v26] Mon, 30 May 2022 21:45:52 UTC (143 KB)
[v27] Wed, 29 Jun 2022 23:31:23 UTC (144 KB)
[v28] Wed, 31 Aug 2022 17:59:52 UTC (133 KB)
[v29] Wed, 31 May 2023 16:16:57 UTC (114 KB)
[v30] Mon, 12 Jun 2023 03:08:56 UTC (120 KB)
[v31] Fri, 30 Jun 2023 17:51:53 UTC (126 KB)
[v32] Mon, 31 Jul 2023 15:46:11 UTC (106 KB)
[v33] Thu, 31 Aug 2023 17:49:06 UTC (107 KB)
[v34] Sat, 30 Sep 2023 17:47:37 UTC (108 KB)
[v35] Tue, 13 Feb 2024 18:23:39 UTC (109 KB)
References & Citations
Bibliographic and Citation Tools
Bibliographic Explorer (What is the Explorer?)
Litmaps (What is Litmaps?)
scite Smart Citations (What are Smart Citations?)
Code, Data and Media Associated with this Article
CatalyzeX Code Finder for Papers (What is CatalyzeX?)
DagsHub (What is DagsHub?)
Gotit.pub (What is GotitPub?)
Papers with Code (What is Papers with Code?)
ScienceCast (What is ScienceCast?)
Demos
Recommenders and Search Tools
Influence Flower (What are Influence Flowers?)
Connected Papers (What is Connected Papers?)
CORE Recommender (What is CORE?)
arXivLabs: experimental projects with community collaborators
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.