This PR fixes the computation of the acceleration. Also, the case of an acceleration close to 0 is now covered; in this case, a linear forecast is used.
I added a TODO that should be resolved by comparing simulation results. Note that the computation of the remaining ressources looks particularly strange. Compare the linear forecast with/without acceleration. If we use the window approach, but measure no acceleration, only the velocity of the second half of the window is used.
This PR fixes the computation of the acceleration. Also, the case of an acceleration close to 0 is now covered; in this case, a linear forecast is used.
I added a TODO that should be resolved by comparing simulation results. Note that the computation of the remaining ressources looks particularly strange. Compare the linear forecast with/without acceleration. If we use the window approach, but measure no acceleration, only the velocity of the second half of the window is used.