In this thesis we study radial weights on Rn. We study two radial weights with different exponent sets. We show that they are both 1-admissible by utilizing a previously shown sufficient condition, for radial weights to be 1-admissible, together with some results connecting exponent sets and Ap weights. Furthermore applying a similar method on a more general radial weight, we manage to improve the previously shown sufficient condition for radial weights to be 1-admissible. Finally we show for one of these two weights that even though it is 1-admissible, whether or not it belongs to some class Ap depends both on the value of p and on the dimension n. Additionally, both of these weights as well as another simple weight are, at least in some dimensions n, not A1 even though they are 1-admissible.

No description available.

Kapaciteter kan vara till stor nytta, bland annat då partiella differentialekvationer ska lösas. Kapaciteter är dock i många fall väldigt svåra att beräkna exakt, speciellt i viktade rum. Vad som istället kan göras är att försöka uppskatta kapaciteterna, vilket för ringar runt en fix punkt kan utföras med hjälp av fyra olika exponentmängder, \underline{Q}_0, \underline{S}_0, \overline{S}_0 och \overline{Q}_0, som beskriver hur vikten beter sig i närheten av denna punkt och i viss mån ger rummets lokala dimension. För att kunna dra nytta av exponentmängderna är det bra att veta vilka kombinationer av dessa som kan förekomma. För att få fram nya kombinationer använder vi olika sätt att mäta volym av klot med varierande radier. Dessa mått är definierade genom olika vikter. Det har tidigare funnits ett fåtal exempel på hur olika kombinationer av exponentmängderna kan se ut. Variationerna består av hur avstånden är i förhållande till varandra och om ändpunkterna tillhör mängderna eller inte. I denna rapport har vi tagit fram ytterligare fem nya kombinationer av mängderna, bland annat en där \underline{Q}_0 är öppen. / Capacities can be of great benefit, for instance when solving partial differential equations. In most cases, capacities can be difficult to calculate exactly, in particular on weighted spaces. In these cases, it can be sufficient with an estimation of the capacity instead. For annuli around a given point, the estimation can be done using four exponent sets \underline{Q}_0, \underline{S}_0, \overline{S}_0 and \overline{Q}_0, which describe how the weight behaves in a neighbourhood of that point and in some sense define the local dimension of the space. To be able to use the exponent sets, it is useful to know which combinations of them can exist. For this we use various measures, which are a way to measure volumes of balls with varying radii in Rn. These measures are defined by different weights. Earlier, there existed a few examples giving different combinations of exponent sets. The variations consist in their relationship to each other and if their endpoints belong to the set or not. In this thesis we present five new combinations of the exponent sets, amongst them one where \underline{Q}_0 is open.

Nous prouvons essentiellement, à partir du formalisme adopté dans les articles [Che] et [CK1], un théorème de di fférentiation de type Calderòn pour les applications des espaces de Hajlasz fondés sur des espaces métriques PI et à valeurs dans des espaces de Banach RNP. Grâce à toutes les techniques développées pour le théorème précédent, nous pouvons -par la suite- a ffaiblir la condition d'appartenance à un espace de Hajlasz surcritique (par rapport à la dimension homogène de l'espace métrique ambiant) en une condition d'intégrabilité locale sur la constante de Lipschitz ponctuelle supérieure. Nous montrons que ces théorèmes de di fférentiation entrent en jeu naturellement pour caractériser les espaces de Hajlasz fondés sur des espaces métriques PI. Ceci débouche sur des critères intégraux, dans la veine de [Br2], pour reconnaitre si des applications mesurables sont constantes ou non dans les espaces métriques PI. En fin, nous discutons certains types d'inégalités de Poincaré locales dépendant du centre et du rayon des boules. Dans ce cadre aff aibli, l'analyse menée précedemment est tout à fait possible mais sous des conditions topologiques et géométriques supplémentaires sur l'espace métrique ambiant. / In this thesis, we essentially prove the Cheeger-differentiability of some Hajlasz-Sobolev functions between PI metric spaces and RNP Banach spaces. Then, we prove a refinement. More precisely, we establish a kind of Rademacher-Stepanov Theorem in the same setting as above but under the simple condition that the upper lipschitz constant is in a Lp space. Then, all these differentiation Theorems are naturally used to give a precise and complete description of the Hajlasz-Sobolev spaces on PI metric spaces in term of an energy integral. This leads to some criteria to detect if a measurable function is constant or not. At the end, we discuss some topological consequences of some weak Poincaré inequalities, we mean that depend of the center and of the radius of the balls involved in these inequalities. In this context, we are able to give some new criteria but the price to pay is to suppose strong topological assumptions on the metric space.

In the study of the weighted p-Laplace equation, it is often important to acquire good estimates of capacities. One useful tool for finding such estimates in metric spaces is exponent sets, which are sets describing the local dimensionality of the measure associated with the space. In this thesis, we limit ourselves to the weighted Rn space, where we investigate the relationship between exponent sets and Muckenhoupt Ap-weights - a certain class of well behaved functions. Additionally, we restrict our scope to radial weights, that is, weights w(x) that only depend on |x|. First, we determine conditions on α such that |x|α ∈ Ap(μ) for doubling measures μ on Rn. From those results, we develop weight exponent sets - a tool for making Ap-classifications of general radial weights, under certain conditions. Finally, we apply our techniques to the weight |x|α(log 1/|x|)β. We find that the weight belongs to Ap(μ) if α ∈ (-q, (p-1)q), where q = sup Q(μ) is a constant associated with the dimensionality of μ. The Ap-conditions in this thesis are found to be sharp.

This thesis consists of four papers and focuses on function spaces related to first-order analysis in abstract metric measure spaces. The classical (i.e., Sobolev) theory in Euclidean spaces makes use of summability of distributional gradients, whose definition depends on the linear structure of Rn. In metric spaces, we can replace the distributional gradients by (weak) upper gradients that control the functions’ behavior along (almost) all rectifiable curves, which gives rise to the so-called Newtonian spaces. The summability condition, considered in the thesis, is expressed using a general Banach function lattice quasi-norm and so an extensive framework is built. Sobolev-type spaces (mainly based on the Lp norm) on metric spaces, and Newtonian spaces in particular, have been under intensive study since the mid-1990s. In Paper I, the elementary theory of Newtonian spaces based on quasi-Banach function lattices is built up. Standard tools such as moduli of curve families and the Sobolev capacity are developed and applied to study the basic properties of Newtonian functions. Summability of a (weak) upper gradient of a function is shown to guarantee the function’s absolute continuity on almost all curves. Moreover, Newtonian spaces are proven complete in this general setting. Paper II investigates the set of all weak upper gradients of a Newtonian function. In particular, existence of minimal weak upper gradients is established. Validity of Lebesgue’s differentiation theorem for the underlying metric measure space ensures that a family of representation formulae for minimal weak upper gradients can be found. Furthermore, the connection between pointwise and norm convergence of a sequence of Newtonian functions is studied. Smooth functions are frequently used as an approximation of Sobolev functions in analysis of partial differential equations. In fact, Lipschitz continuity, which is (unlike <img src="http://www.diva-portal.org/cgi-bin/mimetex.cgi?%5Cmathcal%7BC%7D%5E1" />-smoothness) well-defined even for functions on metric spaces, often suffices as a regularity condition. Thus, Paper III concentrates on the question when Lipschitz functions provide good approximations of Newtonian functions. As shown in the paper, it suffices that the function lattice quasi-norm is absolutely continuous and a fractional sharp maximal operator satisfies a weak norm estimate, which it does, e.g., in doubling Poincaré spaces if a non-centered maximal operator of Hardy–Littlewood type is locally weakly bounded. Therefore, such a local weak boundedness on rearrangement-invariant spaces is explored as well. Finer qualitative properties of Newtonian functions and the Sobolev capacity get into focus in Paper IV. Under certain hypotheses, Newtonian functions are proven to be quasi-continuous, which yields that the capacity is an outer capacity. Various sufficient conditions for local boundedness and continuity of Newtonian functions are established. Finally, quasi-continuity is applied to discuss density of locally Lipschitz functions in Newtonian spaces on open subsets of doubling Poincaré spaces.