Separation of variables as a systematic tool for finding near-harmonic functions.
The goal of defining preharmonic functions and producing a means of upgrading them to actual harmonics is to simplify the problem of finding solutions to Δφ=0 on a manifold (M,g). Inside the space of preharmonic functions, the actual harmonics have infinite codimension, so we’ve in some sense taken care of an infinite number of degrees of freedom. Unfortunately this still leaves infinitely many more, as the preharmonics are also infinite codimension in the smooth functions on M. The goal of this note is to illustrate one technique of finding preharmonic functions, by guessing a simple ansatz, and plug this into the preharmonic condition to get some equations to solve.
We illustrate this technique on the Euclidean plane, where we begin in cartesian coordinates to obscure the rotational symmetry. Our nice ansatze are that there is a preharmonic function that is additively separable (so g(x,y)=a(x)+b(y)) or multiplicatively separable (so g(x,y)=a(x)b(y)). Using the preharmonic condition, we find explicit solutions of both of these forms, and use them to derive the fundamental solution of the Laplacian. This is of note because the fundamental solution itself is neither additively nor multiplicatively separable. Thus, this technique gives strictly more power to the ansatz-guesser than trying to work with the Laplacian directly.
The Setup
Recall the general theory. A function F:M→R is preharmonic if and only if it has the same level sets as a true harmonic function. With some work this reduces to the following equation: F is preharmonic if and only if
∥∇F∥2d(ΔFdF)=ΔFd(∥∇F∥2dF)
Then, if F is preharmonic function, write ∥∇F∥2ΔF=h∘F for some h:R→R. Then f∘F is harmonic, for f:R→R below:
f(x)=∫e−∫h(x)dxdx
Euclidean Geometry
For this example we work in the Euclidean plane E2. We define coordinates by choosing a point O∈E2 and two orthogonal geodesics vert and horiz through O. We then define the following two functions E2→R:
X(−)=dist(−,vert)Y(−)=dist(−,horiz)
The differentials dX,dY are unit length as X,Y are distance functions. The level sets of these functions are equidistant curves to the vertical and horizontal geodesics, which themselves are orthogonal geodesics (since E2 is flat). Thus, vol=dX∧dY and putting these facts together yields
⟨dX,dY⟩=0∣dX∣=∣dY∣=1
⋆dX=dY⋆dY=−dX
From this its straightforward to compute the Laplacians of X,Y. We do X, the Y case is analogous.
ΔX=⋆d⋆dX=⋆d(⋆dX)=⋆d(dY)=⋆(d2Y)=⋆0=0
These facts together with the two-variable chain rule for the Laplacian let us compute Δg(X,Y) for a function g:R2→R (the cross term 2g12⟨∇X,∇Y⟩ drops out because dX and dY are orthogonal):
Δg(X,Y)=g11(X,Y)∥dX∥2+g22(X,Y)∣dY∣2+g1(X,Y)ΔX+g2(X,Y)ΔY=g11(X,Y)(1)+g22(X,Y)(1)+g1(X,Y)(0)+g2(X,Y)(0)=g11(X,Y)+g22(X,Y)
(Of course, we already know this that in Cartesian coordinates Δ=∂x2+∂y2 but…)
Additively Separable
Here we propose to find a preharmonic function g(X,Y) which is additively separable meaning there are two functions a,b:R→R such that
g(X,Y)=a(X)+b(Y)
The preharmonic condition on g(X,Y) requires we compute Δg and ∥∇g∥2=∥dg∥2 so we begin with those:
dg(X,Y)=d(a(X)+b(Y))=d(a(X))+d(b(Y))=a′(X)dX+b′(Y)dY
Then as dX,dY form an orthonormal basis at each point,
∥dg∥2=⟨dg,dg⟩=(a′)2∥dX∥2+2a′b′⟨dX,dY⟩+(b′)2∥dY∥2=(a′)2+(b′)2
And, using our formula for the laplacian chain rule,
Δg(X,Y)=a′′+b′′
The preharmonic condition on g requires that
∥dg∥2[dΔg∧dg]=Δg[d∥dg∥2∧dg]
The easiest way to satisfy this equation would be for both sides to constantly equal zero. And, since we are looking for simple solutions, let’s impose that!
{dΔg∧dg=0d∥dg∥2∧dg=0
Computing the first equation:
d(Δg)=d(a′′+b′′)=a′′′dX+b′′′dY
So
d(Δg)∧dg=(a′′′dX+b′′′dY)∧(a′dX+b′dY)=a′′′b′dX∧dY+b′′′a′dY∧dX=(a′′′b′−b′′′a′)vol
For the second equation,
d∥dg∥2=d((a′)2+(b′)2)=2a′a′′dX+2b′b′′dY
d∥dg∥2∧dg=(2a′a′′dX+2b′b′′dY)∧(a′dX+b′dY)=2a′a′′b′dX∧dY+2b′b′′a′dY∧dX=(2a′a′′b′−2b′b′′a′)dX∧dY=2a′b′(a′′−b′′)vol
If these are both simultaneously equal to zero, we are looking for real valued functions a,b which satisfy the following system of ODEs:
{a′′′b′−b′′′a′=02a′b′(a′′−b′′)=0
Its easiest to begin with the second equation. So long as a,b are not constants (which leads to the trivially harmonic g=const) we know a′ and b′ are not constantly zero and so we must have a′′−b′′=0, or
a′′(X)=b′′(Y)
But since these are functions of different variables which are always equal, we recognize a favorite trick - each side must individually be constant! Thus, there is some K such that
a′′=b′′=K
Integrating each we get
a(X)=2KX2+C1X+C2
b(Y)=2KY2+C3Y+C4
Since both of these are quadratic, their third derivatives are identically zero and the first equation in the system is automatically satisfied. Thus for any constants K,A,B,C the following is preharmonic
g(X,Y)=K(X2+Y2)+AX+BY+C
This is the sum of a function K(X2+Y2) with a function AX+BY+C which is harmonic on all of E2. Since we are in search of fundamental solutions, (which are undefined at the origin) we can ignore this ‘trivial’ harmonic piece, and normalizing to K=1, we’ve found a nontrivial preharmonic function
g(X,Y)=X2+Y2
Upgrading
We now use the general theory to upgrade this to a harmonic function on E2∖O. Computing,
Δg=2+2=4
∥dg∥2=(2X)2+(2Y)2=4(X2+Y2)
So,
∥dg∥2Δg=X2+Y21
and so defining h(s)=1/s we have h∘g=∥dg∥2Δg. Computing the rescaling via integrating factor,
f(s)=∫e−∫hdsds=∫e−∫sdsds=∫e−logsds=∫elogs1ds=∫s1ds=logs
Thus, f∘g(X,Y) is harmonic on E2∖O:
f∘g(X,Y)=logg(X,Y)=log(X2+Y2)
From here its just a simple rescaling to the fundamental solution!
φ(X,Y)=2π1log(r)=4π1log(X2+Y2)
Note that nowhere did we have to know that the answer was radial. The ansatz was additive separability in cartesian coordinates, and the rotational symmetry fell out of the ODEs on its own.
Multiplicatively Separable
Now we propose to find a preharmonic function g(X,Y) which is multiplicatively separable meaning there are two functions a,b:R→R such that
g(X,Y)=a(X)b(Y)
The preharmonic condition on g(X,Y) requires we compute Δg and ∥∇g∥2=∥dg∥2 so we begin with those:
dg(X,Y)=d(a(X)b(Y))=d(a(X))b(Y)+a(X)d(b(Y))=a′(X)b(Y)dX+a(X)b′(Y)dY
Then as dX,dY form an orthonormal basis at each point,
∥dg∥2=⟨dg,dg⟩=(a′b)2∥dX∥2+2(a′b)(ab′)⟨dX,dY⟩+(ab′)2∥dY∥2=(a′b)2+(ab′)2
And, using our formula for the laplacian chain rule,
Δg(X,Y)=a′′b+ab′′
Again, the preharmonic condition on g requires that
∥dg∥2[dΔg∧dg]=Δg[d∥dg∥2∧dg]
Same as above; the easiest way to satisfy this equation would be for both sides to constantly equal zero. And, since we are looking for simple solutions, let’s impose that!
{dΔg∧dg=0d∥dg∥2∧dg=0
Expanding the first
d(Δg)=d(a′′b+ab′′)=(a′′′b+a′b′′)dX+(a′′b′+ab′′′)dY
and wedging with dg=a′bdX+ab′dY,
d(Δg)∧dg=(a′′′b+a′b′′)(ab′)dX∧dY+(a′′b′+ab′′′)(a′b)dY∧dX=[(a′′′b+a′b′′)ab′−(a′′b′+ab′′′)a′b]vol
Repeating with the second equation,
d∥dg∥2=d((a′b)2+(ab′)2)=2a′bd(a′b)+2ab′d(ab′)=2a′b(a′′bdX+a′b′dY)+2ab′(a′b′dX+ab′′dY)=2(a′a′′b2+aa′(b′)2)dX+2((a′)2bb′+a2b′b′′)dY
(d∥dg∥2)∧dg=2(a′a′′b2+aa′(b′)2)(ab′)dX∧dY+2((a′)2bb′+a2b′b′′)(a′b)dY∧dX=2[(a′a′′b2+aa′(b′)2)ab′−((a′)2bb′+a2b′b′′)a′b]vol
Together these result in the following system of equations:
{(a′′′b+a′b′′)ab′−(a′′b′+ab′′′)a′b=0(a′a′′b2+aa′(b′)2)ab′−((a′)2bb′+a2b′b′′)a′b=0
Every solution to this system is a multiplicatively separable preharmonic (though this may not cover all such functions - we are looking at a special case to simplify things after all). As in the additive case, the second equation provides us with the easiest line of attack: factoring out a copy of a′b′ and noting these are zero only in the trivial case where g is constant,
(a′′b2+a(b′)2)a−((a′)2b+a2b′′)b=0
Multiplying through we collect terms with a2 on one side and b2 on the other:
[a′′a−(a′)2]b2=[b′′b−(b′)2]a2
Dividing through gives a separation of variables: thus each side must equal the same constant
a2a′′a−(a′)2=b2b′′b−(b′)2=K
Every term has a product of two copies of a or b (and their derivatives). This suggests proposing an ansatz where a′ is a multiple of a, say a′(X)=u(X)a(X) for some function u (we use u rather than f, which is already spoken for as the upgrading function). This implies
a′′=(ua)′=u′a+ua′=u′a+u2a=(u′+u2)a
Plugging this in,
K=a2a′′a−(a′)2=a2(u′+u2)a2−u2a2=u′
Thus u′=K and so u(X)=KX+C is affine. Looking at the linear case u(s)=Ks for simplicity, we have
a′=ua⟹aa′=u⟹[loga]′=u
a(s)=exp(∫uds)=exp(∫Ksds)=exp(2Ks2)
The same computation with b gives the same function of Y, so our candidate is
g(X,Y)=a(X)b(Y)=exp(2KX2)exp(2KY2)=exp(2K(X2+Y2))
Finally, we need to see that this solution also satisfies the first equation. The quickest route is not to expand the third derivatives but to notice that both quantities appearing in the system are functions of s=X2+Y2 alone. Writing g=eKs/2 and using a′′=K(1+KX2)a,
Δg=a′′b+ab′′=K[2+K(X2+Y2)]g=K(2+Ks)eKs/2
∥dg∥2=(a′b)2+(ab′)2=K2(X2+Y2)g2=K2seKs
Both are functions of s, as is g itself, so all three are constant on the circles s=const. Their differentials are therefore all proportional to ds, and any wedge of two of them vanishes — in particular d(Δg)∧dg=0, which is the first equation.
So it does, and we have our separable preharmonic.
We can see directly from here that X2+Y2 has the same level sets as what we found, and so the multiplicative ansatz has landed on the same function as the additive one, just relabeled. Upgrading proceeds exactly as in the additive case and returns the same log(X2+Y2) — which is the point of the preharmonic formalism: it only ever sees the level sets, so two quite different-looking ansatze that pick out the same foliation are doing the same work.