I want to find the probability of $P(X<0.5 \;|\; Y>0.25)$ of a joint pdf
$$f_{(X,Y)}(x,y)= \frac 1 x $$
with $0<y<x<1.$
This should be as follows:
$$P(X<0.5\;|\; Y>0.25)=\frac{P(X<0.5 \;,\; Y>0.25)}{P(Y>0.25)}$$
The denominator can be calculated as
$$1-P(Y\leq0.25)=1-\int_{y=0}^{0.25}f_Y(y) dy =1-\int_{0}^{0.25} -\ln y \;dy=1-0.596$$
since $f_Y(y)=\int_{y}^1 1/x \; dx=-\ln(y).$
As for the numerator:
$$P(X<0.5 \;,\; Y>0.25)=\int_{x=0}^{0.5}\int_{y=0.25}^{x}f_{X,Y}(x,y) \; dy dx=\int_{x=0}^{0.5}\Big[y/x\Big]^{x}_{0.25}\;dx\\=\int_{x=0}^{0.5}1-0.25/x\;dx=\Big[x- 0.25 \ln x\Big]^{0.5}_0=0.5-0.25 \ln(0.5)+0.25 \ln(0)$$
but $\ln(0)$ is $-\infty$ and the integral does not converge. What have I done wrong?
I see this works changing the limits of integration to
$$P(X<0.5 \;,\; Y>0.25)=\int_{x=0.25}^{0.5}\int_{y=0.25}^{x}f_{X,Y}(x,y) \; dy dx$$
which makes sense looking at the area of integration, but having factored the relationship between $x$ and $y$ in the limits of the inner integral, I don't understand why we can't run the limits from $0$ to $0.5$ on the outside integral.