Extremal functions for the Moser--Trudinger inequality of Adimurthi--Druet type in $W^{1,N}(\mathbb R^N)$ (1702.07970v3)
Abstract: We study the existence and nonexistence of maximizers for variational problem concerning to the Moser--Trudinger inequality of Adimurthi--Druet type in $W{1,N}(\mathbb RN)$ [ MT(N,\beta, \alpha) =\sup_{u\in W{1,N}(\mathbb RN), |\nabla u|NN + |u|_NN\leq 1} \int{\mathbb RN} \Phi_N(\beta(1+\alpha |u|NN){\frac1{N-1}} |u|{\frac N{N-1}}) dx, ] where $\Phi_N(t) =e{t} -\sum{k=0}{N-2} \frac{tk}{k!}$, $0\leq \alpha < 1$ both in the subcritical case $\beta < \beta_N$ and critical case $\beta =\beta_N$ with $\beta_N = N \omega_{N-1}{\frac1{N-1}}$ and $\omega_{N-1}$ denotes the surface area of the unit sphere in $\mathbb RN$. We will show that $MT(N,\beta,\alpha)$ is attained in the subcritical case if $N\geq 3$ or $N=2$ and $\beta \in (\frac{2(1+2\alpha)}{(1+\alpha)2 B_2},\beta_2)$ with $B_2$ is the best constant in a Gagliardo--Nirenberg inequality in $W{1,2}(\mathbb R2)$. We also show that $MT(2,\beta,\alpha)$ is not attained for $\beta$ small which is different from the context of bounded domains. In the critical case, we prove that $MT(N,\beta_N,\alpha)$ is attained for $\alpha\geq 0$ small enough. To prove our results, we first establish a lower bound for $MT(N,\beta,\alpha)$ which excludes the concentrating or vanishing behaviors of their maximizer sequences. This implies the attainability of $MT(N,\beta,\alpha)$ in the subcritical case. The proof in the critical case is based on the blow-up analysis method. Finally, by using the Moser sequence together the scaling argument, we show that $MT(N,\beta_N,1) =\infty$. Our results settle the questions left open in \cite{doO2015,doO2016}.