Failure of hypoellipticity for the canonical solution to \({\bar\partial_b}\) on some nilpotent Lie groups (Q944246)

For \(n\geq 2\) let \(\Sigma_n\) be the analytic submanifold of \(\mathbb C^n\) \[ \Sigma_n=\left\{z=(z_1,\dots,z_n)\in\mathbb C^n: \text{Re}\,z_k=(\text{Re}\,z_1)^k,\,\,1\leq k\leq n\right\}. \] It is clear that \(\Sigma_n\) is a closed CR manifold with real dimension \(n + 1\) and CR dimension 1 diffeomorphic to \(\mathbb R\times\mathbb R^n\) by setting \(z_k = x^k + iy_k\), \(k = 1, \dots ,n\), \(x\in\mathbb R\), \(y = (y_1, y_2, \dots , y_n) \in \mathbb R^n\). It is known that \(\Sigma_n\) admits a nilpotent Lie group structure compatible with the holomorphic structure of \(\mathbb C^n\), in the same way as the classical case of the Heisenberg group. In the paper under review, in analogy to the Stone-Neumann theorem, a complete list of all unitary irreducible representations of \(\Sigma_n\) is given. The \(L^2\) analysis of invariant operators on \(\Sigma_n\) is then carried out using the information about its representations. As \(\Sigma_n\) has CR dimension 1, consider the CR tangential vector field, which in coordinates can be written as \[ {\overline L}=\frac{1}{2}\left(\frac{\partial}{\partial x}+ i\sum_{k=1}^nkx^{k-1}\frac{\partial}{\partial y_k}\right). \] Consider \({\overline L}\) as a closed densely defined operator on \(L^2(\Sigma_n)\), the Hilbert space of square integrable complex valued functions with respect to the Lebesgue measure \(dxdy\). For a suitable given \(f\in L^2(\Sigma_n)\) the author is interested in possible hypoelliptic estimates for the equation \({\overline L}u =f,\) with \(u \perp N_{{\overline L}}\) (the null space of \({\overline L}\)). To be more precise, if \(L\) denotes the conjugate operator \[ L=\frac{1}{2}\left(\frac{\partial}{\partial x}- i\sum_{k=1}^nkx^{k-1}\frac{\partial}{\partial y_k}\right), \] we seek an estimate like \[ \| Lu\|_{L^2}\leq C\| {\overline L}u\|_{L^2},\quad\forall\,\, u \perp N_{{\overline L}}.\tag{1} \] The case \(n = 2\) is well known for \(\Sigma_2\) is just the 1-dimensional Heisenberg group in its polarized form, where even a local version of (1) holds, i.e. for a given point \(p\in \Sigma_2\) and any compact set \(K\subset\Sigma_2\) with nonempty interior such that \(p\in K\), there exists another compact set \(K_0\subset K\) with nonempty interior as well and a constant \(C > 0\) so that \(p\in K_0\) and \(\| Lu\|_{L^2(K_0)}\leq C\| {\overline L}u\|_{L^2(K)}\), as long as \(u\perp N_{{\overline L}}\). In analogy with the Euclidean Riesz transform, the operator \(\mathcal R\), sending \({\overline L}u\) to \(Lu\) with \(u \perp N_{{\overline L}}\), is called the CR Riesz transform. Note that the global Kohn's estimate (1) is equivalent to \(\mathcal R\) having a bounded extension to \(L_2(\Sigma_n)\). It is shown that, surprisingly, for all even \(n > 2\) this is not the case -- there is no constant \(C>0\) such that (1) holds \(\forall u\perp N_{{\overline L}}\).