Hello,
I have a problem concerning a circa 600 atom simulation. When trying to apply a force field to correct the bond lengths, the programm slows down to a crawl and seems to be calculating. However, even after hours of letting it work, there is no progress.
I tried running the same simulation with less atoms, which worked, albeit very slow. So, what I suspect is that the assinged RAM oder processor cores simply dont suffice for that size of molecule.
Are there any settings or ways to access this?
Best Wishes
Thomas
It would help to know a bit more about what you’re trying to do.
At the moment, Avogadro uses a single core for minimization and certainly it can be slow on larger molecules.
On the other hand, I’ve definitely minimized large molecules (500-600 atoms) - it’s slow, but it happens.
So I guess I’m not sure what you mean by “there’s no progress” - and how exactly you’re minimizing. The “Auto Optimize” tool, for example will continue to push even if there’s no progress to be made. It’s designed for interactive sculpting - so that’s the point … to continually minimize until it’s turned off.
The Optimize command, on the other hand, will optimize for some fixed number of steps.
But if you can provide an example file or a bit more details on what you’re trying to do, I can help more.
Sorry, I see that the Information I’ve given is a bit sparse.
I’m trying to optimize the geometry using the ghemical fore field. I mostly just constrain the part of the molecule that is supposed to have fixed coordinates, using “fix selcted atoms”, then start the force field via the Auto Optimization tool and wait until the dE-Value has reached zero.
For the molecule in question i also get a “nan” value for both the Energy and dE. I tried romoving some parts of the molecule, to make it easier to calculate. With that, I got a normal Energy and dE- value, as well as beeing able to see the structure optimize. That is what led me to belive that the molecule seems to be to big for the programm.
I also have some bonds at absurd leghts from connecting two molecules, so I’m certain that the geometry is not in an optimal state.
I wouldn’t recommend Ghemical. It’s an attempt to replicate the old Tripos 5.2 force field. We implemented it because there were no other open-source force fields at the time.
IMHO, MMFF94 or UFF re much better choices.
You’re not giving much detail on the molecule itself, but I’d guess that with Ghemical, you may have elements that are not parameterized. MMFF94 is better for “drug-like” molecules, but doesn’t cover everything either. UFF is nice because it covers pretty much the entire periodic table with semi-reasonable parameters.
Alright, turns out the issue had nothing to do with either the force field or the RAM. There was a mistake in the template I used, where two atoms were at the same spot. I have honestly no idea how to even achieve that, but apparently that breaks the simulation. I tried replicating the issue on a new molecule without success.
In case any of you want to look into this: I was dealing with a nanocapsule, that i was trying to select, by double-clicking the molecule, but likely not selecting the lone atom. So when apliing the force field I had a theoreticly movable atom inside a fixed one. I assume that this resulted in infinite energy, displayed as “nan”. And since infinite energy isn’t easy to calculate, it took up all the programs resources and couldn’t process the other bonds.
While being mathematicly correct, maybe it would be beneficial to implement an error message for those events.
1 Like
Glad to hear you figured it out.
Yes, it would be helpful to have a command to show possible issues. My group has a grant that will be implementing some features like this, so I’ll add to the pile.