Biased simulation in tutorial 5 seems broken #17
Comments
sroet
changed the title
|
Mobile now, so a little hard to read code, but I think this is one where you have to create the shooting strategy for this scheme. Did you include the engine when you created the strategy? It's possible to create a strategy without an engine (defaults to None) -- strongly considering making it required in 2.0 for exactly this reason. Tests, including advanced tests (which runs notebook 5 & the CLI commands) ran 17 hours ago and passed, so I think the code is okay. |
|
Aside: FWIW, the error that matters there is, I think, not the netcdf stuff, but: AttributeError: 'NoneType' object has no attribute 'generate'which says that your engine is set to |
🤦 I did not.
Any reason why we don't already do this? Is there a working use-case in 1.x that does not set the engine at initialization? (I know it is technically an API break, but I don't know of a working use-case for this API) |
|
(This can be closed and moved to an issue in the main OPS repo if you think it is more suited there) |
Early on, we had this idea of setting a "default" engine for a given Python session. The intent was to make it so that users wouldn't need to specify the engine as frequently (if you're setting up strategies manually, it feels like redundant information in most cases). In retrospect, that breaks "Explicit is better than implicit", especially since the implicit value is per-session and won't be retained if you reload simulation objects from a file and continue to create new simulation objects in a new Python process. Since the redundant information is just We do have some examples (focused on showing how to create move schemes) that take advantage of this, but I think the better approach is to clean up the placeholder
I'll close here and open a new issue in openpathsampling/openpathsampling to discuss improving API parameters in general, with this as a first example. |
I am running through these notebooks with a new environment.
When I try to run the:
openpathsampling pathsampling shooting_setup.nc -o biased.nc --scheme biased_shooting -n 500command the code seems to be unable to collect theengineobject (this is not an issue in the regular simulation from the same file)The code errors out in either the backward partial
stack trace
or the forward partial:
stack trace
With a bunch of netcdf errors.
I run with the current conda setup:
conda list
The text was updated successfully, but these errors were encountered: