Appion

Did a search in leginon/tomography and found this:

# leastsq function gives improper parameters error if too many input data array are empty
      # This happens if 6 or more preceeding tilt series have much lower end tilt angle
      # than the current. We will ignore them.
      if cos_tilts.size < 1:
        args_list.pop()
        parameters.pop()

Will have to run simulation to find how we still get there with this code that suppose to prevent it. My guess is that it is related to my separation of the tilts when implementing the alternating tilt scheme.

We just got this same error at the same point during data collection. Leginon collected 1 tilt sequentially, returned to 0, collected the 0 degree and 1 degree tilt, and then gave this "improper input parameter" error.

If I remember correctly I used to see this error sometime after a tilt-series fails. Predictive tracking will use previous tilt-series as models for the current tilt-series​. If one of those previous tilt-series did not collect in full, or possibly if you changed the range or increment after collecting some, then there is a greater chance of seeing this error. This error occurs more frequently if you have a failed tilt-series in the first 5-10 tilt-series in the session. If there is a failed tilt-series after about 10-15 well collected ones, then there is less chance of seeing this error due to less reliance on the broken tilt-series I guess.

To fix it try first clicking the green reset model button two circular green arrows) at the top-right of the Tomography node after a failed tilt-series. If this doesn't fix it, then try killing the application, restarting it, and then resetting the model again. Also make sure your alpha went back to 0.

Let me know if that helps.