5
$\begingroup$

Theorem (Michel). A $1$-form on the projective plane is exact if and only if its integral over any projective line is equal to zero.

Is there a simple proof of this result due, I think, to R. Michel ?

I'm guessing there must be a representation theoretic proof (everything is $SL(3;\mathbb{R})$ equivariant) and a complex-geometric proof (where $\mathbb{R}P^2$ is complexified to $\mathbb{C}P^2$ ) and I would appreciate reference for these, but I'm really interested in a simple proof that one could teach to seniors or first year grads.

P.S. I have not been able to get a hold of Michel's papers. Perhaps everything is there and in that case I apologize before hand.

$\endgroup$

1 Answer 1

11
$\begingroup$

There is a simple proof along the following lines: Because the first deRham cohomology group of $\mathbb{RP}^2$ is trivial, a $1$-form on $\mathbb{RP}^2$ is exact if and only if it is closed. Let $\alpha$ be a $1$-form on $\mathbb{RP}^2$ whose integral over every line vanishes, and let $\beta = \pi^*\alpha$ where $\pi:S^2\to \mathbb{RP}^2$ is the standard double cover. Then the integral of $\beta$ over every great circle vanishes and $\beta$ is invariant under the antipodal involution $\iota:S^2\to S^2$, so $d\beta$ is a $2$-form on $S^2$ that is $\iota$-invariant and its integral over every hemisphere must vanish.

Now $d\beta = b\ dA$ where $dA$ is the standard volume form on $S^2$ and we must have $b\circ\iota=-b$ in order for $d\beta$ to be invariant. Now you use the representation theoretic fact that the operation $A:C^\infty(S^2)\to C^\infty(S^2)$ defined by $$ Af(u) = \frac1{2\pi}\int_{v\cdot u\ge0} f\ dA $$ for $u\in S^2$ is $\mathrm{SO}(3)$-equivariant and hence must, on each eigenspace of the Laplacian on $S^2$, be a multiple of the identity (since these eigenspaces are irreducible representations of $\mathrm{SO}(3)$). To complete the proof that $Ab=0$ implies $b=0$ (and hence that $d\beta=0$), one just needs to check that $A$ is nonzero on each $\iota$-odd eigenspace of the Laplacian, and this is straightforward (see below for one proof of this).

Added comment: Here's a simple way to see that $A$ is nonzero on the odd eigenspaces (without having to do any explicit integration). The $d$-th eigenspace $H_d$ of the Laplacian on $S^2$ is simply the restriction to $S^2$ of the harmonic polynomials on $\mathbb{R}^3$ that are homogeneous of degree $d$. Since $A$ acts as a multiple of the identity on each of these spaces, it's enough to show that, when $d=2m{+}1$, $A$ is nonzero on at least one element of $H_d$. To do this, let $x$, $y$, and $z$ be the standard coordinates on $\mathbb{R}^3$ and consider the polynomial $$ f = \mathrm{Re}\bigl((x+i y)^{2m+1}\bigr) = \prod_{k=-m}^m \left( x-\tan\left(\frac{k\pi}{2m{+}1}\right) y\right). $$ Clearly, $f$ belongs to $H_{2m+1}$ and $f(1,0,0)=1\not=0$. Moreover, $f$ is odd with respect to rotation by an angle of $\pi/(2m{+}1)$ about the $z$-axis, and it vanishes on the $2m{+}1$ planes given by the above factors, which divide the sphere into $4m{+}2$ congruent 'sectors' or 'spherical wedges', on half of which $f$ is positive and on half of which $f$ is negative. Now, $Af(1,0,0)$ is the average of $f$ over the hemisphere on which $x\ge0$, and this is composed of $2m{+}1$ of these sectors, which alternate in the sign of $f$. Since there is an odd number of them, their integrals can't cancel out, so $Af(1,0,0)$ cannot be zero. Thus, $A$ is nonzero on the odd harmonics, as was to be shown.

$\endgroup$
1
  • $\begingroup$ That's great. I had once used the trick of passing to the hemisphere transform to look at the cosine transform and it's nice to see that it can be used in this case as well. If we write the integral $Af$ as an integral over the whole sphere of the function $f$ multiplied by a Heaviside function, then we can differentiate again and reduce the problem to knowing the kernel of the spherical Radon transform, which I was planning to explain the students anyway. $\endgroup$ Commented Jan 2, 2013 at 21:09

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .