Q regarding Integration

[ReyChiquito]

Im trying to understand a "simple" identity.

I have the problem \ddot{y}+g(t,y)=0
with y(0)=y_{0} and \dot{y}(0)=z_{0}

What i need to prove is that the solution can be expresed in the form
y(t)=y_{0}+tz_{0}-\int_{0}^{t}(t-s)g(s,y(s))ds

Integrating twice is clear that
y(t)=y_{0}+tz_{0}-\int_{0}^{t}\{\int_{0}^{s}g(\tau,y(\tau))d\tau\}ds

Now the book goes
\int_{0}^{t}\{\int_{0}^{s}g(\tau,y(\tau))d\tau\}ds =\int_{0}^{t}\{\int_{\tau}^{t}ds\}g(\tau,y(\tau))d \tau

???????????????????? How do i prove that? should i do a change of variable? use fubini?

Im lost, pls help.

[mathman]

It is not any change of variables. If you look at the domain of integration in the (s,tau) plane, you will see it is a triangle. All the book is describing is how the triangle is expressed when you switch the order of integration. Fubini's theorem says you can do the switch.

[ReyChiquito]

Ohhhhhhhhhhhhh.... i know i should have named the tread "stupid Q about integration".
Thx a lot dood. I sure need to review my calc notes.