Geometric analysis of the condition of the convex feasibility problem

نویسندگان

  • Dennis Amelunxen
  • Joachim Hilgert
چکیده

The focus of this paper is the homogeneous convex feasibility problem, which is the following question: Given an m-dimensional subspace of R, does this subspace intersect a fixed convex cone solely in the origin or are there further intersection points? This problem includes as special cases the linear, the second order, and the semidefinite feasibility problems, where one simply chooses the positive orthant, a product of Lorentz cones, or the cone of positive semidefinite matrices, respectively. An important role for the running time of algorithms solving the convex feasibility problem is played by Renegar’s condition number. The (inverse of the) condition of an input is basically the magnitude of the smallest perturbation, which changes the status of the input, i.e., which makes a feasible input infeasible, or the other way round. We will give an average analysis of this condition for several classes of convex cones, and one that is independent of the underlying convex cone. We will also describe a way of deriving smoothed analyses from our approach. We will achieve these results by adopting a purely geometric viewpoint leading to computations in the Grassmann manifold. Besides these main results about the random behavior of the condition of the convex feasibility problem, we will obtain a couple of byproduct results in the domain of spherical convex geometry. Kurzbeschreibung Den Mittelpunkt dieser Arbeit bildet das homogene konvexe Lösbarkeitsproblem, welches die folgende Frage ist: Gegeben sei ein m-dimensionaler Unterraum des R; schneidet dieser Unterraum einen gegebenen konvexen Kegel nur im Ursprung, oder gibt es weitere Schnittpunkte? Dieses Problem umfasst als Spezialfälle das lineare, das quadratische, und das semidefinite Lösbarkeitsproblem, wobei man als konvexen Kegel den positiven Orthanten, ein Produkt von Lorentzkegeln, bzw. den Kegel der positiv semidefiniten Matrizen wählt. Für die Laufzeit von Algorithmen, welche das konvexe Lösbarkeitsproblem lösen, spielt die Renegarsche Konditionszahl eine wichtige Rolle. Die Kondition einer Eingabe, bzw. ihr Inverses, ist gegeben durch die Größe einer kleinsten Störung, welche den Status der Eingabe von ‘feasible’ zu ‘infeasible’ bzw. von ‘infeasible’ zu ‘feasible’ ändert. Wir werden eine Durchschnittsanalyse dieser Kondition für verschiedene Klassen von konvexen Kegeln angeben, sowie eine, welche unabhängig ist von der Wahl des zugrunde gelegten konvexen Kegels. Wir werden desweiteren einen Weg beschreiben, auf dem auch geglättete Analysen mittels unseres Ansatzes erzielt werden können. Wir erreichen diese Ergebnisse mit Hilfe einer rein geometrischen Sichtweise, welche zu Berechnungen in der Grassmann-Mannigfaltigkeit führt. Neben diesen Hauptergebnissen über das zufällige Verhalten der Kondition des konvexen Lösbarkeitsproblems werden wir auch einige Nebenergebnisse im Bereich der sphärischen Konvexgeometrie erzielen.

برای دانلود رایگان متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

Solving a non-convex non-linear optimization problem constrained by fuzzy relational equations and Sugeno-Weber family of t-norms

Sugeno-Weber family of t-norms and t-conorms is one of the most applied one in various fuzzy modelling problems. This family of t-norms and t-conorms was suggested by Weber for modeling intersection and union of fuzzy sets. Also, the t-conorms were suggested as addition rules by Sugeno for so-called  $lambda$–fuzzy measures. In this paper, we study a nonlinear optimization problem where the fea...

متن کامل

Posynomial geometric programming problem subject to max–product fuzzy relation equations

In this article, we study a class of posynomial geometric programming problem (PGPF), with the purpose of minimizing a posynomial subject to fuzzy relational equations with max–product composition. With the help of auxiliary variables, it is converted convert the PGPF into an equivalent programming problem whose objective function is a non-decreasing function with an auxiliary variable. Some pr...

متن کامل

پیدا کردن یک نقطه درونی نسبی یک چند‌وجهی با استفاده از برنامه‌ریزی خطی: کاربرد در برنامه‌ریزی هندسی

One of the fundamental concepts in convex analysis and optimization is the relative interior of a set. This concept is used when the interior of a set is empty due to the incompleteness of its dimension. In this paper, first, we propose a linear programming model to find a relative interior point of a polyhedral set. Then, we discuss the application of this model to geometric programming. Speci...

متن کامل

Fekete-Szegö Problem of Functions Associated with Hyperbolic Domains

In the field of Geometric Function Theory, one can not deny the importance of analytic and univalent functions. The characteristics of these functions including their taylor series expansion, their coefficients in these representations as well as their associated functional inequalities have always attracted the researchers. In particular, Fekete-Szegö inequality is one of such vastly studied a...

متن کامل

Investigation of geometric parameters of seawalls on the amount of earth subsidence and its horizontal displacement by FLAC 3D software

Seawalls are built for Protecting of beaches against waves and preventing the progression of water to the beach. For a proper understanding about these constructions, a suiTable information about applied loads on these constructions should be existed. One of the important load that applied on these constructions is sea wave. Others loads are included: weight force of the walls, weight force of ...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

عنوان ژورنال:

دوره   شماره 

صفحات  -

تاریخ انتشار 2011