[WIP] performance optimizations #83
Conversation
|
Next step on the list is:
@ian-r-rose I created a branch where I pushed the somewhat messy current status. Maybe you can take a look. |
Only SLB is working for now.
|
@ian-r-rose Can you take a look at the new interface for EOS. Only SLB is (kind of?) working for now. |
|
|
||
| if self.method is None: | ||
| raise AttributeError, "no method set for mineral, or equation_of_state given in mineral.params" | ||
|
|
||
| self.V = self.method.volume(self.pressure, self.temperature, self.params) | ||
| self.gr = self.K_T = self.K_S = self.G = self.C_v = self.C_p = self.alpha = None | ||
| self.gibbs = self.helmholtz = self.S = self.H = None | ||
|
|
There was a problem hiding this comment.
I found this confusing at first (setting all of them to None, that is). Kind of ugly, but it works. No real reason not to use it, though.
|
Overall, the interface to for EquationOfState is now more confusing, but I agree that it is for a good reason, and I do not necessarily see a better way of doing it at the moment. |
|
I assume that when you profiled the number of redundant calls have been drastically reduced? |
|
One thing that I would add: If we do use this interface for the equations of state, I see no reason why volume should have different arguments than every other. That is to say, why not just set that to None in set_state() as well? Then just evaluate it in molar_volume? |
|
I think this PR is obsolete now. Can it be closed? |
|
obsolete, but I would like to look at some of the jit functions and implement this. |
|
see #222 |
No description provided.