Slightly confused on what's going on in the diagram - why is the solution oscillating? Isn't the gradient pointing to the center?

ShallowDream

Is it possible to vary the tao value?

ml2

Are we saying that the results oscillate around the actual solution and eventually converge to it? Or is it that the change in how much closer it gets to the solution oscillates, but it's always on one side of the solution and doesn't overshoot?

coolpotato

I'm confused as to when oscillation occurs. Wouldn't it just occur around a minimum?

manchas

Is it ever possible to solve the differential equation itself, or is it unsolvable?

Oh_skr

Can we detect such oscillation in GD and smooth out our descent?

Murrowow

If it was on the end of the ellipse, would it then be able to travel in a straight line that we would expect?

Slightly confused on what's going on in the diagram - why is the solution oscillating? Isn't the gradient pointing to the center?

Is it possible to vary the tao value?

Are we saying that the results oscillate around the actual solution and eventually converge to it? Or is it that the change in how much closer it gets to the solution oscillates, but it's always on one side of the solution and doesn't overshoot?

I'm confused as to when oscillation occurs. Wouldn't it just occur around a minimum?

Is it ever possible to solve the differential equation itself, or is it unsolvable?

Can we detect such oscillation in GD and smooth out our descent?

If it was on the end of the ellipse, would it then be able to travel in a straight line that we would expect?