# On smooth relaxations of obstacle sets

- 229 Downloads

## Abstract

We consider the problem of avoiding obstacle sets described by finitely many smooth convex inequality constraints, as it frequently occurs in, for example, trajectory or location planning. We present and discuss a general method to relax such sets by the upper level set of a single smooth convex function, covering different smoothing approaches like hyperbolic and entropic smoothing.

Based on error bounds and Lipschitz continuity, special attention is paid to the computations of the maximal geometric approximation error and of a guaranteed safety margin. Our results thus allow to safely avoid the obstacle by obeying a single nonconvex smooth constraint. Numerical results indicate that our technique gives rise to smoothing methods which perform well even for smoothing parameters very close to zero.

## Keywords

Relaxation Error bound Lipschitz continuity Hyperbolic smoothing Entropic smoothing Obstacle problem## Notes

### Acknowledgements

We thank two anonymous referees for their precise and substantial remarks which helped to significantly improve this paper. The first author also wishes to thank Adil Bagirov and Diethard Klatte for bringing the references (Li and Fang 1997) and (Klatte 1998; Klatte and Thiere 1996), respectively, to his attention.

## References

- Bertsekas DP (1982) Constrained optimization and Lagrange multiplier methods. Academic Press, New York MATHGoogle Scholar
- Birbil SI, Fang S-C, Frenk JBG, Zhang S (2003) Recursive approximation of the high dimensional max function. Report No. ERS-2003-003-LIS, Erasmus Research Institute of Management, Erasmus Universiteit Rotterdam Google Scholar
- Chen X (2000) Smoothing methods for complementarity problems and their applications: a survey. J Oper Res Soc Jpn 43:32–47 CrossRefMATHGoogle Scholar
- Grossmann C, Kaplan AA (1979) Strafmethoden und modifizierte Lagrange Funktionen in der nichtlinearen Optimierung. Teubner-Texte zur Mathematik, Leipzig MATHGoogle Scholar
- Guerra Vazquez F, Günzel H, Jongen HTh (2001) On logarithmic smoothing of the maximum function. Ann Oper Res 101:209–220 CrossRefMATHMathSciNetGoogle Scholar
- Hiriart-Urruty J-B, Lemaréchal C (2001) Fundamentals of convex analysis. Springer, Berlin CrossRefMATHGoogle Scholar
- Kennings AA, Markov IL (2002) Smoothening max-terms and analytical minimization of half-perimeter wirelength. VLSI Des 14:229–237 MathSciNetGoogle Scholar
- Klatte D (1998) Hoffman’s error bound for systems of convex inequalities. In: Fiacco AV (ed) Mathematical programming with data perturbations. Marcel Dekker, New York, pp 185–199 Google Scholar
- Klatte D, Thiere G (1996) A note on Lipschitz constants for solutions of linear inequalities and equations. Linear Algebra Appl 224:365–374 CrossRefMathSciNetGoogle Scholar
- Lewis AS, Pang J-S (1998) Error bounds for convex inequality systems. In: Crouzeix JP, Martinez-Legaz JE, Volle M (eds) Generalized convexity, generalized monotonicity: recent results. Kluwer Academic Publishers, Dordrecht, pp 75–110 CrossRefGoogle Scholar
- Li XS, Fang SC (1997) On the entropic regularization method for solving min-max problems with applications. Math Methods Oper Res 46:119–130 CrossRefMATHMathSciNetGoogle Scholar
- Robinson SM (1975) An application of error bounds for convex programming in a linear space. SIAM J Control Optim 13:271–273 CrossRefMATHGoogle Scholar
- Robinson SM (1976) Stability theorems for systems of inequalities, part II: differentiable nonlinear systems. SIAM J Numer Anal 13:497–513 CrossRefMATHMathSciNetGoogle Scholar
- Rockafellar RT (1970) Convex analysis. Princeton University Press, Princeton MATHGoogle Scholar
- Smale S (1987) Algorithms for solving equations. In: Proceedings of the international congress of mathematicians. AMS, Providence, pp 172–195 Google Scholar
- Steuermann P (2012) NLPSOLVER ver. 0.0.1-0002. http://kop.ior.kit.edu/downloads.php
- Sun D, Qi L (1999) On NCP-functions. Comput Optim Appl 13:201–220 CrossRefMATHMathSciNetGoogle Scholar
- Xavier AE, Xavier VL (2011) Solving the minimum sum-of-squares clustering problem by hyperbolic smoothing and partition into boundary and gravitational regions. Pattern Recognit 44:70–77 CrossRefMATHGoogle Scholar
- Ye F, Liu H, Zhou S, Liu S (2008) A smoothing trust-region Newton-CG method for minimax problem. Appl Math Comput 199:581–589 CrossRefMATHMathSciNetGoogle Scholar