On the Caffarelli–Kohn–Nirenberg inequalities (Q4956116)

From MaRDI portal

!

This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:

scientific article; zbMATH DE number 1452569
Language Label Description Also known as
default for all languages
No label defined
    English
    On the Caffarelli–Kohn–Nirenberg inequalities
    scientific article; zbMATH DE number 1452569

      Statements

      On the Caffarelli–Kohn–Nirenberg inequalities (English)
      0 references
      0 references
      0 references
      1 February 2001
      0 references
      best constants
      0 references
      existence
      0 references
      extremal functions
      0 references
      properties
      0 references
      Cafarelli, Kohn, and Nirenberg proved \[ \left( \int_{\mathbb{R}^N} |x|^{-bp} |u|^p dx \right)^{2/p} \leq C_{a,b} \left( \int_{\mathbb{R}^N} |x|^{-2a}|\nabla u|^2 dx \right), \] \noindent where \(N \geq 3, - \infty < a < \frac{N-2}{2}, a \leq b \leq a + 1\), and \(p = 2N/(N - 2 +2(b - a))\). The authors are interested in best constants, existence of extremal functions, and properties of extremal functions for \(a\) negative. These questions have been treated for nonnegative \(a\) by a variety of authors. The authors give four results mainly for the case \(-\infty < a < 0\), \(a \leq b \leq a + 1\). Let \[ S(a, b) = \inf_{u \in \mathcal D^{1,2}_a(\mathbb{R}^N) \setminus \{0\}} E_{a,b} (u) \] be the best embedding constant, where \[ E_{a,b} (u) = \left( \int_{\mathbb{R}^N} |x|^{-2a} |\nabla u|^2 dx \right) \times \left( \int_{\mathbb{R}^N}|x|^{-bp}|u|^p dx \right)^{-2/p}. \] and \(\mathcal D^{1,2}_a(\mathbb{R}^N)\) is the completion of \(C^{\infty}_0(\mathbb{R}^N)\) with respect to the inner product: \[ (u,v) = \int_{\mathbb{R}^N} |x|^{-2a} \nabla u \cdot \nabla v dx. \] The results are: \(S(a,b)\) is continuous in the region described above, \(S(a, a+1) = [\frac{N - 2 - 2a}{2}]^2\) and \(S(a, a+1)\) is not achieved, while \(S(a, a) = S(0,0)\) (the best Sobolev constant) and \(S(a,a)\) is not achieved. Next, for \(a<b<a+1\), \(S(a, b)\) is always achieved, for \(b - a \in (0,1)\) fixed, as \(a \rightarrow - \infty\), \(S(a, b)\) is strictly increasing and \(S(a, b) = [(N - 2 - 2a)/2]^{2(b-a)} [S_p(\mathbb{R}^N) + o(1)]\), where \[ S_p(\mathbb{R}^N) = \inf_{u \in H^1(\mathbb{R}^N) \setminus \{0\}} \left( \int_{\mathbb{R}^N} |\nabla u|^2 + u^2 dx \right) \times \left( \int_{\mathbb{R}^N} |u|^p dx \right)^{-2/p}. \] Third, there is a function \(h(a)\) defined for \(a \leq 0\), satisfying \(h(0) = 0\), \(a < h(a) < a+1\) for \(a \leq 0\), and \(a + 1 - h(a) \rightarrow 0\) as \(-a \rightarrow \infty\), such that for any \((a,b)\) satisfying \(a<0\) and \(a < b < h(a)\) the minimizer for \(S(a,b)\) is nonradial. Finally, for \(a<b<a+1\), the minimizer of \(S(a,b)\), possibly after a dilation \(u(x) \rightarrow \tau^{(N - 2 - 2a)/2} u(\tau x)\) satisfies the ``modified inversion'' symmetry: \(u(|x|^{-2} x) = |x|^{N - 2 - 2a} u(x)\). The authors remark that they have given complete results for \(N = 1\) and partial results for \(N = 2\) in the preprint. One remaining open problem is the exact form of the curve \(h(a)\) where the bifurcation of the minimizers of \(S(a, b)\) from radial to nonradial occurs.
      0 references

      Identifiers