There is a article http://pages.iu.edu/~nqle/MA_EVP.pdf of NAM Q.LE state a conjecture of when the eigenvalue of Monge-Ampere equation will arrive the maximum.

It divide into two part:

Conjecture 1 1. Among all bounded open convex sets in $\mathbb R^n$ having a fixed positive volume, the n-dimensional regular simplex (that is the interior of the convex hull of (n + 1) equally spaced points in $\mathbb R^n$) has the smallest Monge-Ampere eigenvalue. 2. Among all open bounded centrally symmetric convex sets in $\mathbb R^n$ having a fixed positive volume, the n-dimensional cube has the smallest Monge-Ampère eigenvalue.

The story is more or less easier for the minimum, the minimum should arrive with the ball $B_{n-1}$ due to the Brunn-Minkowski inequality, which claim the eigenvalue is a convex function on convex body space.

I need explain the basic set first, we consider the Dirichlet eigenvalue problem of Monge-Ampere equation, $$\det(D^2u)=\lambda |u|^{n}\ in\ \Omega$$ $$u=0 \ on \ \partial \Omega$$ This equation is natural to consider due to a rescaling argument, just consider rescaling $u_{\epsilon}(x)=\epsilon\cdot u(x)$, then $\det(D^2 u_{\epsilon})={\epsilon}^n\det(D^2u)=\epsilon^n\lambda|u|^n=\lambda|u_{\epsilon}|^n$.

Lions showed there is a unique constant $\lambda>0$ such that the eigenvalue function have a solution. And there is a variational characterization of $\lambda$, $$\lambda(\Omega)=\inf_{u\in C_0^1(\bar \Omega)\cap C^2(\Omega), u \ convex, u|_{\partial \Omega}=0}\frac{\int_{\Omega}(-u) \cdot \det(D^2u)dx}{\int_{\Omega}(-u)^{n+1}dx}$$

The point is, I think this $\lambda$ should have a geometric characterization, related to some quality of convex body $\Omega$ itself, I do not know if it is right, but I expect the following result:

Conjecture 2 There is a geometric characterization of $\lambda$, that is, for convex body $\Omega$, $$\lambda=\inf_{T }\mu_{n-1}( T(\partial\Omega))$$ Where $T$ runs in all affine map fix the original point and the Lebesgue measure, i.e. under a orthogonal basis, the determination of the matrix of $T$ is 1.

Motivation Let me explain why I believe this conjecture could be true. we know, for equation $\det(D^2u)=\lambda|u|^n$, look it locally, by area formula with the map $F:\mathbb R^n\to \mathbb{TR}^n$, under a basis, $F:(e_1,...,e_n)\to u_{e_1},...,u_{e_n}$, any subset $A\subset \Omega$ we have $$\int_{A} u\cdot \det(D^2u)dx\overset{area \ formula}=\int_{\nabla A}ude$$ where $\nabla A=\{\nabla u(x),x\in A\}$. So the geometric meaning of locally solution of $\det(D^2u)=\lambda|u|^n\ \in \Omega$ is $\forall A\in \Omega$, $|\nabla A|=\int_A \lambda|u|^n$, so $\lambda=\sup_{A\to \Omega}\inf_u \frac{\int_{\nabla A}ude}{\int_{\Omega}(-u)^{n+1}dx} (*)$.

So what is the meaning of $(*)$ by intuition, it should be the optimal choice of a sires of level set, with the zero level set ecocide with $\partial \Omega$. because at once you fix the fibration of level set, the choice of $u$ is more or less the same. So this eigenvalue characterize the geometry of $\Omega$, but I can not go further, due to three obstacles:

  1. The first one is I do not know what is the universal choice of $u$ to arrive minimum $\lambda$ when fix a fiber of level set, I think it is come from a equal condition of holder inequality, but I can not work it out.
  2. The second one is I do not know what is the best choice of level set fibration, I think it come from the fibration center at the "center of mass of $\Omega$" and then do the most natural one, but I do not know how to proof this.
  3. If we can proof 1,2. Then how to explain the product $\lambda$ by this thing could coincide with the statement of $\lambda$ in conjecture 2.

I wish this could be a toy model of "hear the shape from the spectrum".

May be this approaches is useless. In any case, I wish we can say some thing about it, whatever positive answer or negative answer. I will appreciate to any interesting argument, thank you for your attention!


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.