Uniqueness of weighted Sobolev spaces with weakly differentiable weights (Q1762338)

From MaRDI portal
Revision as of 07:22, 1 February 2024 by Import240129110113 (talk | contribs) (Added link to MaRDI item.)
scientific article
Language Label Description Also known as
English
Uniqueness of weighted Sobolev spaces with weakly differentiable weights
scientific article

    Statements

    Uniqueness of weighted Sobolev spaces with weakly differentiable weights (English)
    0 references
    0 references
    23 November 2012
    0 references
    In order to study the weighted \(p\)-Laplacian \[ - \text{div} \big[ w | \nabla u|^{p - 2} \nabla u \big] = fw, \] which is the Gâteaux derivative of the convex functional \[ E^{\mu}_0: u \mapsto \frac1p \int | \nabla u|^p d \mu, \] by methods of the calculus of variations, one looks for minimizers of the functional \[ E^{\mu}_f: u \mapsto E^{\mu}_0 - \int f u \, d \mu. \] The mimimizers depend on the energy space chosen, but one wants to include the \(C^{\infty}_0\) functions (compact support) in the space. One sets \(H^{1, p}(\mu)\) to be the completion of \(C^{\infty}_0\) with respect to the Sobolev norm \[ \| u \|_{1, p, \mu} = \left( \| \nabla u \|_{L^p(\mu; R^d)}^p + \| u \|^P_{L^p(\mu)} \right)^{\frac1p}, \] which is the strong weighted Sobolev space. Let \(V\) be another weighted Sobolev space such that \(V \subset L^p(\mu)\) is densely and continuously embedded, \(V\) admits a linear gradient-operator \(\nabla^V: V \mapsto L^p(\mu, \mathbb R^d) \) that respects \(\mu\)-classes, \(V\) is complete in the Sobolev norm, and \(C^{\infty}_0 \subset V\), \(\nabla u = \nabla^Vu, \mu\)-a.e. for \(u \in C^{\infty}_0\). Then we know that \(H^{1, p}_0(\mu) \subset V\), but if the inclusion is proper, Lavrentév's phenomenon occurs if \[ \min_{u \in V} E_f(u) < \min_{ u \in H^{1,p}(\mu)} E_f(u). \] In order to prevent this possibility, one wants to know when \(H^{1,p}(\mu) = V\). For \(w \equiv 1\), this is known as the Meyers-Serrin theorem and briefly denoted by \(H = W\). For \(p = 2\), it is known as ``Markov uniqueness''. It has also been studied when \(p = 2\) for a weighted Sobolev space. The author gives a condition (\textbf{Diff}) that implies this result for \(p = 1\). Let \(\phi = w^{\frac1p}\), and for \(1 \leq p < \infty\), \textbf{Diff} is the condition \[ \phi \in W^{1, p}_{\text{loc}}(dx), ~\beta = p \frac{ \nabla \phi}{\phi} \in L^q_{\text{loc}}(\mu; \mathbb R^d), \] where \(q\) is the conjugate index to \(p\). The author's main result is: Theorem 1.1. If the condition (\textbf{Diff}) holds with \(p = 1\) and, additionally, \[ \nabla \phi \in L^{\infty}_{\text{loc}}(dx, \mathbb R^d), \] then \(C^{\infty}_0(\mathbb R^d)\) is dense in \(V^{1, p}(\mu)\) and, in particular, \[ H^{1,p}(\mu) = V^{1,p}(\mu). \] The author also gives other sufficient conditions, a weighted Poincaré inequality, and discusses a parabolic evolution equation with the weighted \(p\)-Laplacian.
    0 references
    \(H=W\)
    0 references
    weighted Sobolev spaces
    0 references
    smooth approximation
    0 references
    density of smooth functions
    0 references
    Poincaré inequality
    0 references
    \(p\)-Laplace operator
    0 references
    nonlinear Kolmogorov operator
    0 references
    weighted \(p\)-Laplacian evolution equation
    0 references
    nonlinear degenerate parabolic equation
    0 references
    Meyers-Serrin theorem
    0 references

    Identifiers