On the global integrability of non-negative harmonic functions (Q1774175)

Let \((X,{\mathcal H}^*)\) be a Bauer harmonic space with a Green function \(\Gamma= \{G(x, y)\mid x, y\in X\}\) and a reference measure \(r(dx)\) on \(X\). Let \({\mathcal H}_+(X)\) be the cone of nonnegative harmonic functions on \(X\). Under the following hypothesis \[ (h_n)\subset{\mathcal H}_+(X)\text{ and }\sum^\infty_{n=1} r(h_n)< +\infty\Rightarrow \sum^\infty_{n=1} h_n\in{\mathcal H}_+(X),\tag{R} \] the author proves for any positively homogeneous, increasing map \(N:{\mathcal B}_+\to [0,\infty]\) defined on the cone of nonnegative Borel measurable functions on \(X\) such that, \[ N\Bigl(\sup_{n\geq 1} u_n\Bigr)= \sup_{n\geq 1} N(u_n),\quad (u_n)\subset{\mathcal B}_+, \] holds \(N(s)\leq Cr(s)\) for \(s\) nonnegative and superharmonic (\(C\) a given constant), provided that \(N\) is finite on the cone of nonnegative superharmonic function on \(X\). The above result is also true if we replace in it the the cone of superharmonic function on \(X\) by the cone of nonnegative, \(r\)-integrable, harmonic functions on \(X\). The author remarks that the above-mentioned assumption (R) is necessary for his proof, however the Doob convergence axiom may not be necessary for his proof! It is interesting to note that since Green function and reference measure for \((X,{\mathcal H}^*)\) exist under his settings, we are actually close to the situation in which ``every finite potential could be expressed as a countable sum of continuous potentials'' (Axiom \(\overline D\)) [cf. \textit{K. Janssen}, Math. Z. 141, 185--191 (1975; Zbl 0283.31008), Remark 2.5.a].