I noticed the setting for cache_override in time step problems:
fe = {
'chunk_size' : 1000,
'cache_override' : False,
}
What is its purpose?
On 04/15/10 14:44, freevryheid wrote:
I noticed the setting for cache_override in time step problems:
fe = { 'chunk_size' : 1000, 'cache_override' : False, }
What is its purpose?
In coupled problems (e.g. examples/biot/biot.py) there are usually several terms, that need to compute e.g. strain, or other derived quantities. The caches serve to this purpose: each term can ask a cache instance to provide the data - if the data are not present, they are computed, otherwise they are reused. This simplifies and speeds-up certain things, but may introduce subtle bugs related to (not)updating the right data. To debug it, there is the option 'cache_override'. When it's True, the caches always recompute the data every time they are requested (no reuse).
So if you see a strange problem, try setting it to True, otherwise it's safe to have it False. The value True is the default, when it's not set in 'fe'. In the small stationary example problems shipped with the code it really does not matter.
Also watch for Issue 104 when using solutions from previous time step...
r.
participants (2)
-
freevryheid
-
Robert Cimrman