> On other hand, in 1st, 2th, and 3th year, there are still more women in abstract math classes than men.

This is almost definitely false. While the male/female gender ratio is reasonably close to even for math majors (I think it's 60/40 m/f?), it is nowhere close to even for many heavy math-based disciplines. Engineering and computer science degrees are very male-dominate and the students routinely spill over into math classes.

FWIW, I majored in math at a university that was 10th highest in the student female to male ratio. I counted some of the classes for fun, and I remember counting a slight male majority most of the time. One semester I believe I counted a 57% male population over my math classes.

> I do not think any "who is better at what" can be said, beyond the stereotype "women are bad at math" can be explicitly be stated as false.

This certainly seems to be true. High school females are starting to outperform males in math, while males continue to dominate math graduate school. Reasons are unknown, but in general we can say that "girls aren't bad at math".

When people speak of "abstract math", they generally mean mathematics, because they use the term "math" far too loosely. In a very real sense, a person can be taught to "do math" all the way through Calculus III, Linear Algebra, and even Differential Equations and still not have any real aptitude for mathematics.