Anybody--
I am trying to model a BJT using a similar approach as in the examples file. Things were going pretty well until Genesys suddenly started producing "#QNAN" strings appended to the tuned variable quantities.
I have tried rebooting my computer, messing with equations, and even saving as a different file and starting over. Nothing seems to help.
:sighs: So I thought I'd share...
Mark
[file]spice vs linear 05.wsx[/file]
I am trying to model a BJT using a similar approach as in the examples file. Things were going pretty well until Genesys suddenly started producing "#QNAN" strings appended to the tuned variable quantities.
I have tried rebooting my computer, messing with equations, and even saving as a different file and starting over. Nothing seems to help.
:sighs: So I thought I'd share...
Mark
[file]spice vs linear 05.wsx[/file]
Nice bug. We've been trying to find an example for this. Apparently a few of the variables you are optimizing start at a value of 0 and we're calculating an indeterminate gradient which then propagates poorly into the equation set.
You can easily fix this by not optimizing the variables that are zero or setting them to any small but non-zero value. We'll fix this in the optimizer and the equation updater for the next release.
Note that it's ok if they go to zero during the optimization. They just don't like to start at exactly 0.
Mark