copper grain boundary system started today

[graeme]

Andreas started a simulation to look at the dynamics in a grain boundary in copper today. This is a somewhat more tricky problem than we have considered in the past because atom in the boundary can see very low diffusion barriers. It will be interesting to see how it goes.

The potential is the same as was being used before in our Mg simulations, except with parameters appropriate for Cu. Hopefully there will be no problems due to the switch. The system is quite a bit smaller, so the time between results will decrease. I have also added some improvements to the code on the client side which will reduce the average number of force calls required to find each saddle point.

We'll be watching the system pretty closely over the next few days, so if problems arrise, we'll be on them quickly. Also, Andreas is in Denmark and I'm in the US so one of us should always be awake.

[Thor]

I would say they are quite a bit smaller

I had 127 results in the lst 30 minutes to about 10 with the "old" test


Thor

[graeme]

Hopefully the server keeps up with the traffic.

[PY 222]

Woot. So this is why I am geting thousands of extra points. I was scratching my head when I saw the stats update.



I'll try to bring in more firepower so you guys can analyze those copper grains faster.

[graeme]

You are a one man supercomputer -- unbelievable!

[PY 222]

Looks like vgriffin has more GHz than me!

Now I need to double my efforts to outproduce him.

[black_civic55]

haha wow my 2 lowly comps running right now are sure truckin with the smaller measurements!!!

[PY 222]

I see that the amount of measurements have been decreasing over time since this morning when the change to copper grain was made.

Any idea what is up?

[graeme]

Yeah, it's a little hard to tell what's going on because we don't have good tools (yet) to see how the atoms are moving in this bulk simulation. But by watching the results come in, the server does seem to get into states which are hard to get out of. In these states, many of the searches are failing. I'm going to leave it running for a while, and perhaps get Andreas to tweak a few parameters tomorrow, so that the searches are more conservative.