A quantitative version of the Bianchi-Egnell inequality concerning the stability of the Sobolev inequality is proved for non-negative functions. For the proof we study a flow that interpolates continuously between a function and its symmetric decreasing rearrangement.
introduction and main resultIn [9] Brezis and Lieb posed the question whether it is possible to bound the 'Sobolev deficit'from below in terms of some natural distance from the manifold of optimizers. Here d ≥ 3, 2 * = 2 d/(d − 2) is the 'Sobolev exponent', and, its distributional gradient is a square-summable function and it vanishes at infinity in the sense that |{x ∈ R d : |f (x)| > }| < ∞ for all > 0. Here |A| denotes the Lebesgue measure of a measurable set A. Throughout this paper we deal with real-valued functions. With minor additional effort our arguments can be extended to the case of complex-valued functions.Rodemich [39], Aubin [3] and Talenti [43] (see also [41]) proved that the Sobolev deficit is non-negative. Moreover, it was shown by Lieb [35], Gidas, Ni and Nirenberg [34] and Caffarelli, Gidas and Spruck [14] that the deficit vanishes if and only if the function f is of the form