Updated log files given changes in the printing message

changed milestone to %Towards python version 1.0

mentioned in merge request !135 (closed)

In this PR !135 (closed), I encountered a different behaviour for each example. I used the latest devel branch to run each exampel (8005a7e8). We need to investigate the reason of these changes.

By reviewing the code history, I found these changes between the master and devel branch (8b2371c4):

• Reading of SRDF for default posture (i.e. half-sitting).
• Changed the alpha values self.alphas = [4**(-n) for n in range(10)] -> self.alphas = [ 2**(-n) for n in range(10) ].

Then I run all the examples and record the logs files (see log-comparison.tar.gz). From these results I notice:

1. Behaviour is changed due to alpha values, except bipedal case since we never decrease the step length.
2. My log files are different to these ones 895e9720.
3. New alpha values (inside devel) are better than before one (confirming https://gepgitlab.laas.fr/loco-3d/crocoddyl/merge_requests/113).

My thesis is:

• Point 2 might be explained as using an old talos-data repository which doesn't include armature. It's a reasonable explanation since it already happens (see discussion https://gepgitlab.laas.fr/loco-3d/crocoddyl/issues/126#note_3317).
• Salto solution in devel branch is worse than master one, the reason is that new alpha values aren't a good fit for this case.

Conclusions:

1. We should change manually the alpha values for the salto example.
2. We should update the log file for the salto example after doing 1.
3. We should merge this PR !135 (closed) after doing 3.

To confirm your hypothesis could you do as follow for one example (for example the jump):

• specify at the beginning of the example file the armature you want, to match the armature that was set before.
• specify the alphas in the ddp solver to match the ones we had before.
• specify the half-sitting reference position to match the one we had before.

With these 3 changes, the modified example should produce the same log.

Solved in !139 (merged). I will close it

closed