Assume we have a probability measure $\mu$ over $\mathbb{R}^d$ that is absolutely continuous with respect to Lebesgue measure.

Given $m$ polynomials $p_1,\ldots,p_{m}\in \mathbb{R}[x_1,\ldots,x_d]$ we define a mapping from $m$ points in $\mathbb{R}^d$ to $\mathbb{R}^{m}$ $F(\bar{x}_1,\ldots,\bar{x}_m):\mathbb{R}^{d\times m} \to \mathbb{R}^{m}$ in the following way:

$F(\bar{x}_1,\ldots,\bar{x}_m)=\alpha_1,\ldots,\alpha_{m}$ if and only if for every $\bar{x}_i$ we have $\sum_{n=1}^{m} \alpha_n p_n(\bar{x}_i)+1=0$. (This is almost everywhere well defined.)

The mapping can be thought of as a mapping from $m$ points to an algebraic set: The algebraic set which is the common real zeros of a polynomial $\sum_{n=1}^m\alpha_np_n(x)+1$. For each $\bar{\alpha}$ we will denote the corresponding algebraic set by $V_\bar{\alpha}$.

Now assume that we chose our original polynomials in such a way that almost surely each $V_\bar{\alpha}$ is an irreducible algebraic set, and that we can apply the disintegration theorem and obtain a family of probability measures $\{\mu(\cdot | \alpha)\}_{\alpha}$. Each is almost surely supported on $V_\bar{\alpha}$ (on $V_\bar{\alpha}^m$ actually).

It looks kind of obvious that almost all measures $\mu(\cdot|\alpha)$ are probability measure and if $V$ is another algebraic set such that $V_\bar{\alpha} \nsubseteq V$ then $\mu(V|\bar{\alpha})=0$. (i.e. that proper Zariski closed sets of $V_\bar{\alpha}$ have measure zero). Is that true and how can it be proved?


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.