Closed BenSmithers closed 1 year ago
Same comment as for #733: Please pull/merge the current changes from master, which are fixing the test runs. I'll look at the changed code in more detail, later!
Sorry for leaving this PR sit idle...I'm fine with the propose changes and it can be merged
It still needs to be brought up to speed with master, so the tests won't fail, though.
Summary
These commits will allow
DerivedParam
s to be defined in thecfg
files directly. Much of this is relevant to additions in pull request #715This is very handy if you have a basis change for your nuisance parameters, and don't need as many parameters in the new basis to span the space of all output shapes accessible.
Additions
Edits here are pretty minor. I just added some lines to the
config_parser
to make theDerivedParam
s and configure them appropriately. I also made a minor change to theStage
class so it doesn't raise exceptions when a bunch of extra parameters are there. I removed a print statement too.Usage
For a
DerivedParam
, a list of parameter names theDerivedParam
is derived from needs to be provided, separated by spaces; this is assigned todepends_names
.It should also be passed a
function_file
, which is a pisa-path (or full path) pointing to a json-serialized utils.callable.Funct whose relevantVar
s refer to thoseParam
names theDerivedParam
depends on.Ex:
So, for example, you could have
Param
s theta_34 and theta_24, andDerivedParam
s for U_mu4 and U_tau4 which load function files telling it how to calculate the unitary mixing matrix elements from the mixing angles themselves (or you could do the exact opposite).