[Ipopt] unconstrained step size

[Sean Hardesty]

What Ipopt parameters or parts of the problem formulation control the size
of the initial step that Ipopt takes in the absence of constraints or
Hessian information (I am using the BFGS option)? Ipopt is taking very
small steps on my problem: substantial reduction in the objective function
could be found in the negative gradient direction if the step size were a
couple orders of magnitude larger. I thought at first that the way I had
formulated my constraints was a problem, but I found that Ipopt still takes
tiny steps even without any constraints. It also updates mu. I have to
admit I've never really thought about this issue - not having constraints
or a Newton step, it seems there isn't a natural way to select a step
length.

    Thanks,
    Sean Hardesty
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://list.coin-or.org/pipermail/ipopt/attachments/20180715/48e0f844/attachment.html>