Take the 2-minute tour ×
MathOverflow is a question and answer site for professional mathematicians. It's 100% free, no registration required.

Hi, harmonicity in 2d is preserved under mappings that satisfy the Cauchy-Riemann equations.

What about 3D? What conditions should a mapping satisfy to preserve harmonicity?

is there a general characterization a la CR for 3D?

Here is an example of non-trivial such mapping

Let $u(x,y,z)=U(X,Y,Z)$ where $$X=xy+z,~~~~ Y= \frac{\sqrt3}4 (x^2-y^2) - \frac{xy}{2}+z ,~~~~ Z= -\frac{\sqrt3}4 (x^2-y^2) - \frac{xy}{2}+z$$

(I found this example by first assuming $X=xy+z$ then guessing for Y,Z from the overdetermined system that they satisfy... hope it's right...)

share|improve this question
The correct notion is the one of pluriharminicity, see en.wikipedia.org/wiki/Pluriharmonic_function Then pluriharmonicity is preserved by pull-back via holomorphic functions of several complex variables. Of course, none of this makes any sense in $R^3$. –  Misha May 14 '12 at 17:26
Your example does not work. It is true that $\partial_XU$, $\partial_YU$, $\partial_ZU$ are orthogonal to each other. But they have different norms and that prevents the mapping to preserve harmonicity. –  Denis Serre May 15 '12 at 9:00

3 Answers 3

Actually, the correct notion of maps preserving harmonicity is that of 'harmonic morphisms': A map $f:(M,g)\to (N,h)$ between Riemannian manifolds is a harmonic morphism if it pulls back $h$-harmonic functions on $N$ to $g$-harmonic functions on $M$. There are many, many nontrivial examples, and there is a large literature on the subject.

There is an extensive Atlas of Harmonic Morphisms (see http://www.maths.lth.se/matematiklu/personal/sigma/harmonic/atlas.html) that contains a useful bibliography.

share|improve this answer
To expound: the paper "Bent Fuglede, Harmonic morphisms between Riemannian manifolds, Ann. Inst. Fourier (Grenoble) 28 (1978)" is (I think) the first paper where the term 'harmonic morphism' is defined. There is some overlap with the paper by Ishihara in Igor's answer as these were published around the same time. Note that harmonic morphisms are not the same as harmonic maps or maps that commute with the Laplacian ("Bill Watson, Manifold maps commuting with the Laplacian, J. Differential Geometry 8 (1973)"). –  Paul Reynolds May 14 '12 at 22:04
I wrote down a system of overdetermined PDE's that such harmonic morphism must satisfy. But it seems like a large system containing 8 equations; I am at a loss as to how to reduce it... lots of structure though... –  curious May 15 '12 at 3:19
ps: i tried using maple's "casesplit" command but it blew up my computer [4gig]... –  curious May 15 '12 at 3:20


MR0545705 (80k:58045) Ishihara, Tôru A mapping of Riemannian manifolds which preserves harmonic functions. J. Math. Kyoto Univ. 19 (1979), no. 2, 215–229. 58E20 (31C12)

The author shows that a mapping has the title property in dimension at least three if and only if it is either constant or a Riemannian covering up to homothety.. For $\mathbb{R}^n$ this means a composition of translations, rotations, and homothety ($x\rightarrow a x,$ for some $a>0.$)

share|improve this answer

Your "examples" usually don't work. Up to a complex conjugation (= symmetry), a $2$D-mapping $\phi$ that preserves harmonicity is a holomorphic function, via $z=x_1+ix_2$ and $f(z)=\phi_1+i\phi_2$. Now, let us form $\psi(x_1,x_2,x_3)=(\phi(x_1,x_2),x_3)$. What happens is that $$(\partial_1^2+\partial_2^2)(u\circ\phi)=|f'|^2(\partial_1^2u+\partial_2^2u)\circ\phi.$$ From this, you see that $\Delta(u\circ\psi)$ is not proportional to $(\Delta u)\circ\psi$ if $|f'|\ne1$ (notice that if $|f'|\equiv1$, then $f$ is an affine isometry). Hence $\psi$ does not preserve harmonicity. More generally, a function $\phi$ preserves harmonicity ($u$ harmonic implies $u\circ \phi$ is harmonic) if and only if ${\rm D}\phi(x)$ is a similitude, that is the product $\rho(x)R_x$ of some isometry and of a homothety.

On the contrary, a theorem due to Liouville says that a mapping in ${\mathbb R}^3$ that preserves angles must be an affine similitude. This applies in particular to mappings that preserve harmonicity. Edit (after comments below): Liouville's Theorem actually says that a direct $C^4$ conformal map is the composition of an affine similarity and possibly of inversions $x\mapsto a\|x-x_0\|^{-2}(x-x_0)$.

share|improve this answer
@curious Note that $\Delta(u\circ \psi)=|f'|^2((\partial_1^2+\partial_2^2)u+\partial_3^2 u)\circ \psi$, which is not proportional to $(\Delta u)\circ \psi$ if $|f'|^2\neq 1$ and $\partial_3^2u\neq 0$. So Denis Serre is right that your "examples" usually don't work. However, I looked up the theorem due to Liouville in my copy of Marcel Berger's "Géométrie 1", and certain inversions also seem to preserve harmonicity. I guess they are $x \mapsto 1/||x||^n$, but I would have to check. Of course, these inversions are not defined everywhere in $\mathbb R^3$, so the remark is only misleading. –  Thomas Klimpel May 15 '12 at 0:58
The comment above contains frustratingly many typos (wrong parenthesis in the expression for $\Delta (u \circ \psi)$, $x\mapsto 1/||x||^n$ instead of $x\mapsto x/||x||^n$), but the message should be clear nevertheless. –  Thomas Klimpel May 15 '12 at 1:03
ah yes my bad. Thank you for straighting me out, my example was indeed nonsense. Let me check if inversion preserves harmonicity... any other examples? or is it an exhaustive list? –  curious May 15 '12 at 1:58

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.