On a uniform estimate for the quaternionic Calabi problem (Q375779)

For a hypercomplex structure \((I,J,K)\) with hyper-Hermitian metric \(g\) on a manifold \(M\), denote by \(F_I\), \(F_J\), \(F_K\) the fundamental forms of \(I\), \(J\), \(K\), respectively. Then \(\Omega= F_J+ iF_K\) is a \((2, 0)\)-form with respect to \(I\) and \(g\) is called HKT if \(\partial_I\Omega= 0\). Such manifolds have many properties, similar to those of Kähler manifolds. For example, for any function \(f\) the form \(\widetilde\Omega= \Omega+\partial_I\circ J\circ\overline{\partial_I}f\) also defines a HKT metric if some positivity condition is satisfied. The first author and \textit{M. Verbitsky} [Isr. J. Math. 176, 109--138 (2010; Zbl 1193.53118)] formulated on HKT manifolds a quaternionic version of the Calabi problem: find a function \(f\) for which \(\widetilde\Omega^n\) is a fixed non-vanishing \((2n, 0)\)-form. They conjectured the existence and showed the uniqueness of the solution. Moreover, they proved a \(C^0\) a priori estimate for the solution under the condition that the global holonomy of the Obata connection is contained in \(\mathrm{SL}(n,\mathbb H)\). This requires in particular the existence of a global holomorphic non-vanishing \((2n, 0)\)-form. The paper under review shows similar \(C^0\) a priori estimates under different conditions -- namely the Obata connection is flat, which is equivalent to the condition that the local holonomy is trivial. As an intermediate step they prove a quaternionic version of a theorem due to Gauduchon.