PR#162 introduced outcome_space and had a lengthy discussion of what "outcomes" should be. In conclusion, outcome should be hashable, but as detailed as possible. So, NOT integers.
Currently we encodings as part of the public API. These are not ready yet and we have not concluded on what the encoding API should be. Hence, Encodings must be removed from all docs. Encoding code can stay internally but shouldn't be relied on too much until the devs have agreed on an API and this has been designed and implemented fully.
In conclusion this leads some outstanding todos after #162 :
[x] Remove all referneces to encodings from documentation
[x] Remove encoding references from Dispersion
[x] Fix tests for Dispersion
[x] add outcome space for Dispersion
[x] Implement outcome_space for SymbolicAnything as vector or tuple of elements, not just integers. Done in #172.
[x] Ensure that SymbolicAnything follow the declared probability estimator API as described in the Dev Docs. Done in #172.
[x] fix tests for SymbolicAnything`. Done in #172.
[x] For TransferOperator, fix tests, add outcome space and ensure that it follows the probability estimator API, as above.
PR#162 introduced
outcome_space
and had a lengthy discussion of what "outcomes" should be. In conclusion, outcome should be hashable, but as detailed as possible. So, NOT integers.Currently we encodings as part of the public API. These are not ready yet and we have not concluded on what the encoding API should be. Hence, Encodings must be removed from all docs. Encoding code can stay internally but shouldn't be relied on too much until the devs have agreed on an API and this has been designed and implemented fully.
In conclusion this leads some outstanding todos after #162 :
Dispersion
Dispersion
Dispersion
outcome_space
forSymbolicAnything
as vector or tuple of elements, not just integers. Done in #172.SymbolicAnything
follow the declared probability estimator API as described in the Dev Docs. Done in #172.TransferOperator
, fix tests, add outcome space and ensure that it follows the probability estimator API, as above.