priority of --maxtime and --force-convergence

[QianXiaobo]

Hi developers,

I found that --maxtime can't make any limitation to --force-convergence. I think --maxtime should be priority to --force-convergence to kill the process when sometimes cannot achieve convergence very long time.

I don't know what I say is correct or not, you could think about it, or ignore it...

Best wishes,
Xiaobo

[tjstruck]

Hi @QianXiaobo, you are correct.

--maxtime is a property of the optimization package dadi-cli uses. Each optimization dadi-cli starts gets the --maxtime assigned to it, so you are correct that --force-convergence would override it. If convergence isn't reached, --force-convergence would launch more optimizations with the --maxtime assigned to them.

It probably wouldn't be too hard to change --maxtime to be for the dadi-cli inference subcommands running, but I'll wait to hear what other devs think (whether --maxtime should stay the time limit of the individual optimization or change it to be the time limit for the subcommand).