You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Summary
Allow more specific versions of the FiniteVolume method to be used (e.g. with linear extrapolation or quadratic extrapolation, use arithmetic mean or harmonic mean when shifting from nodes to edges)
Should we:
Create a new inherited class for each option (e.g. FiniteVolumeLinearExtrapolation) etc
Allow for such options to be applied to specific symbols (e.g. pybamm.surf(c_s, "linear"))
Create a generator class of spatial methods and allow for a dictionary of options to be passed to the spatial method
change the __init__ of spatial methods so that only a dict of options are taken in upon initialisation and then when the object passed to Discretisation we call a second set_up function that has the same functionality as the current __init__ method.
I am most in favor of 4. but 2. probably requires the least changes to the code.
Thoughts?
Motivation
Following from discussion in pull request #707 .
The text was updated successfully, but these errors were encountered:
Right, I see that issue that there might be a few different options which makes 1 a bad idea. Agreed that 4 sounds like the best option. It should only require a change to one line in discretisation, and a few relevant tests? We could even make mesh an optional input to __init__, (__init__ calls set_up directly if a mesh is provided) so that existing functionality still works, but this might be confusing
Summary
Allow more specific versions of the FiniteVolume method to be used (e.g. with linear extrapolation or quadratic extrapolation, use arithmetic mean or harmonic mean when shifting from nodes to edges)
Should we:
FiniteVolumeLinearExtrapolation
) etcpybamm.surf(c_s, "linear")
)options
to be passed to the spatial method__init__
of spatial methods so that only a dict ofoptions
are taken in upon initialisation and then when the object passed toDiscretisation
we call a secondset_up
function that has the same functionality as the current__init__
method.I am most in favor of 4. but 2. probably requires the least changes to the code.
Thoughts?
Motivation
Following from discussion in pull request #707 .
The text was updated successfully, but these errors were encountered: