Statement

On a UCdU\subseteq\mathbb C^d, let HH, SHSH, PSHPSH, and PHPH denote the , , , and functions, respectively. Then

PH=HPSH,HSH,PSHSH.PH=H\cap PSH,\qquad H\subseteq SH,\qquad PSH\subseteq SH.

Equivalently, inside SHSH, the classes HH and PSHPSH overlap exactly in PHPH. When d=1d=1, PSH=SHPSH=SH and PH=HPH=H.

Diagram

The set-theoretic picture for d>1d>1 is

SH  H  PSH,HPSH=PH.\begin{array}{c} SH\\[2pt] \supset\; H\ \cup\ PSH,\\[2pt] H\cap PSH=PH. \end{array}

Neither HPSHH\subseteq PSH nor PSHHPSH\subseteq H holds in general.

Why the intersection is pluriharmonic

For a smooth function, PSHPSH means that the Levi matrix is positive semidefinite, while harmonicity says that its is zero because

Δu=4j=1d2uzjzˉj.\Delta u=4\sum_{j=1}^d \frac{\partial^2u}{\partial z_j\partial\bar z_j}.

A positive-semidefinite with zero trace is zero. Hence a harmonic PSH function has vanishing and is pluriharmonic. The same conclusion holds without smoothness by distributional regularity.

For the inclusions, a harmonic function has zero real Laplacian and hence is subharmonic on the underlying real domain, giving HSHH\subseteq SH. A plurisubharmonic function is subharmonic on that real domain because its Levi matrix is positive semidefinite and four times its trace is the real Laplacian, giving PSHSHPSH\subseteq SH. (Restricting a real-harmonic function to an arbitrary complex line would not, by itself, prove the first inclusion.) Vanishing Levi form is locally equivalent to u=Rehu=\operatorname{Re}h for hh, which is pluriharmonic; mollification gives the distributional version. In one complex dimension four times the Levi trace is the planar Laplacian, so the classes collapse as stated.

Separating examples

On Cd\mathbb C^d, the function u(z)=z12u(z)=|z_1|^2 is PSH but not harmonic. For d2d\ge2, the function v(z)=z12z22v(z)=|z_1|^2-|z_2|^2 is harmonic but not PSH. These examples expose the distinction: subharmonicity controls the trace of the , whereas plurisubharmonicity controls the whole Hermitian matrix.

References
  1. Lars Hörmander, Notions of Convexity, Birkhäuser, 2007. DOI record. Relevant: Chapter 2.
  2. Marek Klimek, Pluripotential Theory, Oxford University Press, 1991. Publisher record.