Newton relaxation too small

Hi there,

I had a problem in running the motion and parameter optimization with error of Newton relaxation too small. Here is the details of the error:
Failed to resolve kinematic constraints. Newton relaxation too small. (final kin. error = 7.328953E-004)
Optimality cond.no. 14 above error tolerance 0.000100, error = 0.000145.
......
Optimality cond.no. 293 above error tolerance 0.000100, error = 0.000354.
Constraint violations for study 'Main.Studies.MotionOptimization' :
ERROR(OBJ.MCH.KIN3) : C:/S..s/Tall/S28/R..g/S..n/M..l/Kinematics.any(81) : MotionOptimization.Kinematics : Kinematic analysis failed in time step 13 : Failed to solve position analysis.

I have looked up the forum, I revised the initial position many times, but the model still failed somewhere (e.g., step 24). I also adjust the starting frame and ending frame in the C3D file, but it still failed. I also tried to adjust the marker position, but I have less confidence when I do so since I have no idea of the principals in adjusting the marker position. It would be greatly appreciated if anyone could help with this model.

Due to the data sharing restriction by the institution, I could not provide the model files. Thanks very much in advance.

Best,

Junfei

Hi JunFei,

I have a couple of questions.

Initial positions will only help for the first frame of the study, after this there is no impact, so unless the error is in first frame changing initial pos will not help.

Marker positions should of course be as good as possible, and connected to the correct underlying segment.

Also note that there needs to be enough markers to determine the motion of the DOF in the human model, if you have few markers in regions of the body you may need to apply an extra driver on a joint.

Hope it helps.

Best regards
Søren

Hi Søren,

Thank you so much for your reply. It did not run one analysis completely. I have successfully run the model for many other subjects but failed with this one after lots of trials. Therefore, I think the constraints should be fine. As for the marker position, should I just adjust the marker position to make the marker close to the C3D data in the initial position? Thank you very much.

Best,

Junfei

Hi Junfei,

If you have been running many subjects with the same setup it would check if there is any chance of a marker dropping out?

I would look if there are any markers which for this subject was placed differently on the human model and then adjust it slightly. Try to look at the model running and spot the largest marker errors then try to reduce these, large errors may lead to numeric issues.

Best regards
Søren

This topic was automatically closed 125 days after the last reply. New replies are no longer allowed.