I was looking at this answer which eventually stated this about the proper acceleration (when $\vec v$ is parallel to $\vec a$): $$ \vec \alpha =\gamma^3 \vec a = \gamma^3 \dfrac{d\vec v}{dt}=\dfrac{du}{dt}=\dfrac{d}{dt}\dfrac{d\vec x}{d \tau} $$ Now, I'm programming a relativistic integrator and to do this I used the first and last part of this equation and switched the derivatives around: $$ \vec \alpha= \dfrac{d}{d\tau} \dfrac{d\vec x}{dt} = \dfrac{d\vec v}{d\tau} $$ Now this seems to work as I can recreate the hyperbolic curve obtained from Rindler coordinates if I choose $\alpha$ to be constant. (I have numerically compared this and it works). So to clarfiy I would do this integration step: $dv = \alpha d\tau$, where $dv$ is the change of velocity in the coordinate frame.
Now this all seems fine until we look back at the first equation and use $dt = \gamma d\tau$: $$ \vec \alpha = \gamma^3 \dfrac{d\vec v}{dt}=\gamma^2 \dfrac{d\vec v}{d\tau} \neq \dfrac{d\vec v}{d\tau} $$ This is not equal to the previously acquired formula. What's going on here? Am I not allowed to just convert $d\tau$ to $dt$?