Summary
The ellipsoid method is the first known algorithm which in the sense of computational complexity solves linear and convex programming problems efficiently. Based on a simple computational model we introduce some notions from complexity theory and survey the historical antecessors which led to the development of the ellipsoid method. We discribe the fundamental geometric idea as well as a basic version of the algorithm and prove that this basic version gives an efficient algorithm for linear programming problems. We discuss some modifications of the basic method and outline the algorithmic equivalence of optimization and separation which is a consequence of the ellipsoid algorithm. By means of an example we show how this equivalence has led to a unifying framework for efficient algorithms which allowed to discover further efficiently solvable problems.
Zusammenfassung
Das Ellipsoidverfahren ist das erste bekannte Verfahren, das lineare und konvexe Optimierungsprobleme im Sinne der Komplexitätstheorie effizient löst. Mithilfe eines einfachen Rechnermodells führen wir kurz einige zum Verständnis notwendige Begriffe der Komplexitätstheorie ein und geben anschließend einen Überblick über die algorithmischen Vorläufer, die zur Entwicklung des Ellipsoidverfahrens geführt haben. Wir beschreiben die grundlegende geometrische Idee sowie eine Basisversion des Verfahrens, für die wir den Nachweis führen, daß sie lineare Programme im komplexitätstheoretischen Sinne effizient löst. Wir diskutieren einige Modifikationen der Basisversion und skizzieren die algorithmische Äquivalenz von Optimierung und Separation, die auf der Ellipsoidmethode beruht. Anhand eines Beispiels erläutern wir, wie diese Äquivalenz zu einem einheitlichen Modell effizienter Methoden geführt hat, das es erlaubte, weitere, effizient lösbare Probleme zu entdecken.
Similar content being viewed by others
References
Adler I, McLean R, Provan S (1980) An application of the Khachian-Shor algorithm to a class of linear complementarity problems. Cowles Foundation Discussion Paper No. 459, Cowles Foundation for Research in Economics, New Haven, USA
Agmon S (1954) The relaxation method for linear inequalities. Can J Math 6:382–392
Aho AV, Hopcroft JE, Ullman JD (1974) The design and analysis of computer algorithms. Addison-Wesley, Reading, Mass.
Agkül M (1980) On Shor-Khachian algorithm. University of Waterloo, Dept. of Combinatorics and Optimization, Waterloo, Ontario, Canada
Avis D, Chvatal V (1976) Notes on Bland's pivoting rule. Math Prog Stud 8:24–34
Bachem A (1982) Concepts of algorithmic computation. In: Korte B (ed) Modern applied mathematics — optimization and operations research. North-Holland, Amsterdam Oxford New York, pp 3–49
Borgwardt K-H (1982) The average number of pivot steps required by the simplex method is polynomial. ZOR 26:157–177
Chung SJ, Murty KG (1981) A polynomially bounded algorithm for positive definite symmetric LCPs. In: Burkard RE, Ellinger T (eds) Methods of operations research, vol 40. Athenäum/Hain/Scriptor/Hanstein, Königstein Ts. pp 63–66
Dantzig GB (1979) Comments on Khachian's algorithm for linear programming. Dept. of Operations Research, Tech. Rep. SOL 79-22, Stanford University, Stanford, USA
Ech-Cherif A, Ecker JG (1982) A class of rank-two ellipsoid algorithms for convex programming. Working Paper, Mathematical Sciences Department of Rensselaer Polytechnic Institute
Edmonds J (1970) Submodular functions, matroids and certain polyhedra. Proc. Int. Conf. on Combinatorics (Calgary), Gordon and Breach, New York, pp 69–87
Eremin II (1960, 1965) The relaxation method of solving systems of inequalities with convex functions on the left sides. Dokl Akad Nauk SSR 160(5):994–996; Sov Math Dokl 6:219–221
Ermolev YuM (1966) Methods of solution of nonlinear extremal problems. Kibernetika 2 (No 4):1–17; Cybernetics 2 (No4):1–16
Gács P, Lovász L (1981) Khachian's algorithm for linear programming. In: König H, Korte B, Ritter K (eds) Mathematical programming at Oberwolfach. Math Programming Study, vol 14. North Holland, Amsterdam Oxford New York, pp 61–68
Garey MR, Johnson DS (1979) Computers and intractability — a guide to the theory of NP-completeness. WH Freeman and Company, San Francisco, USA
Godfarb D, Sit WJ (1979) Worst case behavior of the steepest edge simplex method. Discrete Appl Math 1:277–285
Goldfarb D, Todd MJ (1980) Modifications and implementation of the ellipsoid algorithm for linear programming. Tech. Report No. 406, Dept of Comp Sci, Cornell University, Ithaca, NY, USA
Grötschel M, Lovász L, Schrijver A (1981) The ellipsoid method and its consequences in combinatorial optimization. Combinatorica 1:169–197
Jeroslow R (1973) The simplex algorithm with the pivot rule of maximizing criterion improvement. Discrete Math. 4:367–377
Jones PC, Marwil ES (1982) Dimensional reduction variant of the ellipsoid algorithm for linear programming problems. Math Oper Res 7:245–252
Judin DB, Nemirovskii AS (1976, 1977) Evaluation of the information complexity of mathematical programming problems. Ekonom Mat Metody 12:128–142; Matekon 13(2): 3–25
Judin DB, Nemirovskii AS (1976, 1977) Informational complexity and efficient methods for the solution of convex extremal problems. Ekonom Mat Metody 12:357–369; Matekon 13(3):25–45
Judin DB, Nemirovskii AS (1977) Informational complexity of strict convex programming. Ekonom Mat Metody 13:550–559
Karp R, Papadimitriou C (1980) On linear characterization of combinatorial optimization problems. Proc. 21st Ann Symp Foundations of Computer Sciences, IEEE. Computer Society Press, Los Angeles, pp 1–9
Khachian LG (1979, 1980) A polynomial algorithm in linear Programming. Dokl Akad Nauk SSR 244:No. 5 (Abstract); Dokl Akad Nauk SSSR 20:51–68; Sov Math Dokl 20 (No 1):191–194
Klee V, Minty GL (1982) How good is the simplex algorithm? In Shisha O (ed) Inequalities III. Academic Press, New York, pp 159–175
König H, Pallaschke D (1981) On Khachian's algorithm and minimal ellipsoids. Numer Math 36:211–223
Korte B, Schrader R (1981) A note on convergence proofs for Shor-Khachian-methods. In: Auslender A, Oettli W, Stoer J (eds) Optimization and optimal control. Lecture Notes in Control and Information Sciences, vol 30. Springer, Berlin Heidelberg New York, pp 51–57
Kozlov MK, Tarasov SP, Khachian LG (1979) Polynomial solvability of convex quadratic programming. Dokl Akad Nauk SSSR 248 (No 5); Sov Math Dokl 20 (No 5)
Krol J, Mirman B (1980) Some practical modifications of the ellipsoid method for LP problems. Arcon Inc, Boston, Mass, USA
Lawler, EL (1975) Matroid intersection algorithms. Math Programming 9:31–56
Levin A (1965) On an algorithm for the minimization of convex functions. Dokl Akad Nauk SSSR 160 (Sov Math Dokl 6:286–290)
Mityagin BS (1969) Two inequalities for volumes of convex bodies. Mat Zametki 5:99–106;Math Notes 5:61–65
Motzkin T, Schoenberg IZ (1954) The relaxation method for linear inequalities. Can J Math 6:393–404
Padberg M, Rao MR (1980) The russian method for linear inequalities and linear optimization. New York Univ., Graduate School of Business Administration
Padberg M, Rao MR (1980) The russian method for linear inequalites. II: approximate arithmetic. New York Univ., Graduate School of Business Administration
Padberg M, Rao MR (1980) The russian method and integer programming. Graduate School of Business Administration, New York Univ., New York
Pickel PF (1979) Some improvement to Khachiyan's algorithm in linear programming. Polytechnic Institute of New York, Farmingdale, NY, USA
Polyak BT (1967) A general method of solving external problems. Dokl Akad Nauk SSSR 174(1):33–36; Sov Math Dokl 8:593–597
Polyak PT (1978) Subgradient methods: a survey of soviet research. In: Lemarechal C, Mifflin R (eds) Nonsmooth optimization. Pergamon Press, Oxford New York Frankfurt
Shor NZ (1968) The rate of convergence of the generalized gradient descent method. Kibernetika 4 (no 3):98–99; Cybernetics 4 (No 3):79–80
Shor NZ (1970) Utilization of the operation of space dilatation in the minimization of convex functions. Kibernetika 6:6–12; Cybernetics 6:7–15
Shor NZ (1970) Convergence rate of the gradient descent method with dilatation of the space. Kibernetika 8 (No 2): 80–85; Cybernetics 8:102–108
Shor NZ (1975) Convergence of a gradient method with space dilatation in the direction of the difference between two successive gradients. Kibernetika 11 (No 4):48–53; Cybernetics 11:564–570
Shor NZ (1977) Cut-off method with space extension in convex programming problems. Kibernetika 13 (No 1):94–95; Cybernetics 13:94–96
Shor NZ (1977) New development trends in non-differentiable optimization. Kibernetika 13 (No 6):87–91; Cybernetics 13:881–886
Shor NZ, Gamburd PR (1971) Certain questions of convergence of generalized gradient descent. Kibernetika 8 (no 6): 82–84; Cybernetics 8:1033–1036
Shor NZ, Gershovich VI (1979) Family of algorithms for solving convex programming problems. Kibernetika 15 (No 4):62–67; Cybernetics 15:502–507
Shor NZ, Zhurbenko NG (1971) A minimization method using the operation of extension of the space in the direction of the difference of two successive gradients. Kibernetica 7 (No 3):51–59;Cybernetics 7:450–459
Smale S (1982) The problem of the average speed of the simplex method. In: Bachern A, et al. (eds) Mathematical programming, Bonn 1982. The State of the Art. Springer, Berlin Heidelberg New York
Todd MJ (1982) Minimum volume ellipsoid containing part of a given ellipsoid Math Operat Res 7:253–261
Traub JF, Wozniakowski H (1982) Complexity of linear programming. OR Lett 1:59–62
Yamnitsky B, Levin LA (1982) An old linear programming algorithm runs in polynomial time. Paper presented at the Silver Jubilee Conference, Waterloo, Ont., Canada
Zadeh N (1973) A bad network problem for the simplex method and other minimum cost flow algorithms. Math Programming 5:255–266
Author information
Authors and Affiliations
Rights and permissions
About this article
Cite this article
Schrader, R. The ellipsoid method and its implications. OR Spektrum 5, 1–13 (1983). https://doi.org/10.1007/BF01720281
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/BF01720281