Felix Schmidt

Hi Oscar, thanks for checking! I had started implementing a new forward pass but got stuck. The method requires computing a cut intercept adjustment, cf:  The 'norm' part $\sqrt{(1+||c_t...

I understand! That makes sense! I'll close this issue! I suppose if you pertubed the states towards the centre you'd end up with a stabilisation/gravity centre-like approach where the stability...

> It's not obvious that we even want to do this. Regularizing the forward pass to "nice" trajectories means that we add fewer cuts to extreme parts of the state...

Will do! Thank you! I recall that at some point before adding `RegularizedForwardPass` you had tried out different bundle methods like level set / quadratic regularisation /trust region, right? If...

on it :)