Take the 2-minute tour ×
Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

Let $P$ and $Q$ be two equivalent probability measures, and let $dP=ZdQ$ for some $Z$.

If I know the probability density of a random variable $X$ under measure $P$, how can I calulate the density of $X$ under measure $Q$? In other words, if I know that $$P(X\le x)=\int\limits_{-\infty}^x f(s)\, ds,$$ for all real $x$, is it possible to find a function $g$ such that $$Q(X\le x)=\int\limits_{-\infty}^x g(s)\, ds.$$

share|improve this question

1 Answer 1

up vote 3 down vote accepted

The definition of the Radon-Nykodym derivative $\dfrac{\mathrm dQ}{\mathrm dP}=U$ (your $Z$ being $Z=1/U$) is that, for every event $A$, one should have $Q(A)=\mathbb E_P(U;A)$. When $A=[X\leqslant x]$, this is $Q(X\leqslant x)=\mathbb E_P(U;X\leqslant x)$ hence one looks for some density function $g$ such that, for every $x$, $$ \mathbb E_P(U;X\leqslant x)=\displaystyle\int_{-\infty}^xg(t)\mathrm dt. $$ In the special case when $U=\varphi(X)$, calling $f_X$ the density of $X$, one gets $$ Q(X\leqslant x)=\mathbb E_P(\varphi(X);X\leqslant x)=\displaystyle\int_{-\infty}^x\varphi(t)f_X(t)\mathrm dt, $$ hence the function $g=\varphi f_X$ is a density (nonnegative, integrates to $1$) and solves your question.

In the general case, calling $f_{U,X}$ the density of the joint distribution of $(U,X)$, the solution is $$ g(x)=\int_{0}^{+\infty} uf_{U,X}(u,x)\mathrm du. $$ This specializes to $g=\varphi f_X$ when $U=\varphi(X)$ and, at the other extreme, to $g=f_X$ when $U$ and $X$ are independent. However, these cases are the exception rather than the rule since, in general, $g$ does not involve $f_X$ only but the joint distribution of $(U,X)$.

share|improve this answer

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.