skip to main content
OSTI.GOV title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: Refining a triangulation of a planar straight-line graph to eliminate large angles

Conference ·
OSTI ID:10165881

Triangulations without large angles have a number of applications in numerical analysis and computer graphics. In particular, the convergence of a finite element calculation depends on the largest angle of the triangulation. Also, the running time of a finite element calculation is dependent on the triangulation size, so having a triangulation with few Steiner points is also important. Bern, Dobkin and Eppstein pose as an open problem the existence of an algorithm to triangulate a planar straight-line graph (PSLG) without large angles using a polynomial number of Steiner points. We solve this problem by showing that any PSLG with {upsilon} vertices can be triangulated with no angle larger than 7{pi}/8 by adding O({upsilon}{sup 2}log {upsilon}) Steiner points in O({upsilon}{sup 2} log{sup 2} {upsilon}) time. We first triangulate the PSLG with an arbitrary constrained triangulation and then refine that triangulation by adding additional vertices and edges. Some PSLGs require {Omega}({upsilon}{sup 2}) Steiner points in any triangulation achieving any largest angle bound less than {pi}. Hence the number of Steiner points added by our algorithm is within a log {upsilon} factor of worst case optimal. We note that our refinement algorithm works on arbitrary triangulations: Given any triangulation, we show how to refine it so that no angle is larger than 7{pi}/8. Our construction adds O(nm+nplog m) vertices and runs in time O(nm+nplog m) log(m+ p)), where n is the number of edges, m is one plus the number of obtuse angles, and p is one plus the number of holes and interior vertices in the original triangulation. A previously considered problem is refining a constrained triangulation of a simple polygon, where p = 1. For this problem we add O({upsilon}{sup 2}) Steiner points, which is within a constant factor of worst case optimal.

Research Organization:
Sandia National Labs., Albuquerque, NM (United States)
Sponsoring Organization:
USDOE, Washington, DC (United States)
DOE Contract Number:
AC04-76DP00789
OSTI ID:
10165881
Report Number(s):
SAND-93-1392C; CONF-931167-1-Extd.Abst.; ON: DE93016510
Resource Relation:
Conference: 34. meeting of the Institute of Electrical and Electronics Engineers Foundation of Computer Science,Palo Alto, CA (United States),3-5 Nov 1993; Other Information: PBD: 13 May 1993
Country of Publication:
United States
Language:
English