What does it mean if PDF of a random variable comes out to be negative?

50 Views Asked by At

Suppose I have random variable $X \sim \exp(1)$, then we have a random variable Y such thath

$$Y = 1+ e^{-sX}$$

When I find the PDF for $Y$, it came out to be

$$f_Y(y) = -e^{x(s-1)}$$


Steps to find the PDF Let $F_Y(y)$ be the CDF of new variable $Y$, $$F_Y(y) = P[Y \leq y] = P[1+ e^{-sX} \leq y] = P\Large[\small X \geq \frac{-\ln(y-1)}{s}\Large]$$

$ P\Large[\small X \geq \frac{\ln(y-1)}{s}\Large]$ is the Complimentary CDF of $X \sim \exp(1)$ and equals to $\Large e^{-\frac{\ln(y-1)}{s}}$

$$F_Y(y) = \Large e^{-\frac{\ln(y-1)}{s}}.$$

Now, if I take derivate of $F_Y(y)$, I get the PDF as above to be a negative value.

If now I replace $y = 1 + e^{-sx}$, then I get the above result.

1

There are 1 best solutions below

0
On BEST ANSWER

$P[X \geq \frac {-\ln (y-1)} s] =e^{\ln (y-1) /s}$ and not $e^{-\ln (y-1) /s}$; the derivative of this is $\frac 1 {s(y-1)} e^{\ln (y-1) /s}$ for $y >1$.

[$P{(X>t)}=e^{-t}$. Put $t=\frac {-\ln (y-1)} s$ in this].