gamma method in corr routines about pyerrors HOT 5 CLOSED

It should work, but it would not show the errors, I assume. The question is, whether such one-liner has to work, when this means that one has to call the gamma method without the user knowing it/ been able to prevent this.

from pyerrors.

I think corr.m_eff().some_func().show() should still work just without displaying any y-errors for the individual data points. I see the convenience of this one-liner but I think having to call the gamma_method explicitly leads to cleaner workflows as one should always manually monitor the autocorrelation function and the chosen window and adjust the parameters of the gamma_method accordingly (there can be unpleasant surprises otherwise). The problem with implicit calls to the gamma_method is that manually chosen values for say S would be overwritten (which the user might not expect).

One thing we could think about is giving the show method an additional parameter which triggers running the gamma_method (with standard parameters but explicitly authorized by the user). This would alter your one-liner to

corr.m_eff().some_func().show(run_the_gamma_method_before_plotting=True) 

or something like that (suggestions for a short but meaningful parameter name are appreciated).

from pyerrors.

I think you are right. The show method is really the only method that really needs to have this. I suggest to call it auto_gamma to imply that it uses the default parameters and to have it default to true, so one does not have to set it for every .show.

from pyerrors.

an other solution might be to have a .apply_gamma method, which returns a copy with the gamma method applied. This way one could decide to call it at any time in a one-liner eg.
corr.apply_gamma().some_func_needing_errors().m_eff().apply_gamma().show()

from pyerrors.

auto_gamma sounds like a good solution, I can take care of that. I would prefer to default the parameter to False. Otherwise the following could happen

corr.gamma_method(S=3.0)
corr.show()
print(corr[0].S)
> {'my_ensemble': 2.0}

which could be confusing to a user unfamiliar with the history of this.

We could also introduce a apply_gamma method if that is useful for you. Although I have to say that corr.apply_gamma().some_func_needing_errors().m_eff().apply_gamma().show() is not very readable 😅

from pyerrors.