1: \begin{definition}\label{def:srd}

2: Let $U\sim L_k(m,\Sigma)$, with nonsingular $\Sigma$. We say that the map $\phi:\RR\times\Omega\to\RR$ defined by

3: \[

4:   \phi(x,\omega) = \sum_{i=1}^k H_i(\omega)\,\I_{[t_{i-1},t_i)}(x)

5: \] 

6: is a \textit{simple random density}, in which $H=(H_1,\dots,H_k)$ are the \textit{random heights of the steps} of $\phi$, with distribution given by $\mu_H(A)=\mu_{U\mid \SD(U)}(A\mid 1)$, for $A\in\cR^k$, and $\mu_{U\mid \SD(U)}$ is the regular version of the conditional distribution of $U$ given $\SD(U)$ obtained in Theorem \ref{theo:cond.dens}. Hence, for every $A\in\cR^k$, we have

7: \[

8:   \mu_H(A) = \int_A \frac{f_U(h)}{f_{\SD(U)}(1)}\,\I_{\HH_1}(h)\,d\tau_1(h) \, ,

9: \]

10: in which $\tau_1(A)=d_k^{-1} \lambda_{k-1}(\pi(A\cap\HH_1))$ and it holds that $\mu_H(\HH_1)=1$. We use the notation $\phi\sim\Delta(m,\Sigma)$.

11: \end{definition}

12: