Running time becomes very long when using a different database

[Chuangye-Wang]

It takes much longer time to run the multicomponent (NiCrAl) precipitation example using my database than the provided NiCrAl.tdb. What could be the reasons for making the calculaiton efficiency very different? Do you have any suggestions on how to make the simulation faster?

[nury12n]

There's a few things you could try:

• Since this issue was brought up before kawin 0.3.0, you could try updating kawin to see if that helps since there were quite a few changes in how the models are solved
• You could use the thermodynamics class directly rather than the surrogate, using model.setThermodynamics(therm) and skipping the surrogate section in the example. If the surrogates don't match the underlying thermodynamics that well, it's possible to get non-physical values which can affect the simulation.
• If you want to use the surrogates, you could try increasing the density of points or expanding the composition range it's trained on

If neither of these options work, it would be helpful if you're able to share the TDB file. It's possible that the thermodynamic models in the TDB file result in largely different driving forces compared to the TDB file used in the example which can affect how the precipitation model is solved. Being able to look at the TDB file could give some insight into why you're seeing the time differences.