Failure mode in repetition analysis scripts

[MWSammons]

Hi @cwjames1983 and @JordanHoffmann3 . Amanda Cook has been interested in using the zdm repo for computing some repetition statistics, and wants to rerun the analysis from your repeaters paper. Running through the scripts in the zDM/papers/Repeters/fit_repeating_distributions.py she's running into an issue with the parameters dictionary not containing Wbins parameter key and instead containing a WNbins key. Sifting through the functions and the traceback (which Amanda will include below) I had some trouble disentangling what the issue was. A couple of odd things were that the update_param function ends up doubly defined under the mydata class both here

zdm/zdm/data_class.py

Lines 94 to 102 in 63f79cf

	def update_param(self, param:str, value):
	""" Update the value of a single parameter
	Args:
	param (str): name of the parameter
	value (?): value
	"""
	DC = self.params[param]
	setattr(self[DC], param, value)

and here

zdm/zdm/parameters.py

Lines 418 to 427 in 63f79cf

	def update_param(self, param:str, value):
	# print(self.params)
	DC = self.params[param]
	setattr(self[DC], param, value)
	# Special treatment
	if DC == "cosmo" and param == "H0":
	if self.cosmo.fix_Omega_b_h2:
	self.cosmo.Omega_b = (
	self.cosmo.Omega_b_h2 / (self.cosmo.H0 / 100.0) ** 2
	)

and then the associated self.params dictionary seems to contain distinct keys from what was expected (WNbins rather than Wbins) from where I couldn't pin down. Seems like its pretty close to just working off the shelf and so any assistance in fixing this would be greatly appreciated!

[cookamanda]

Hi all! As promised, the traceback after having removed the second definition of update_param and changing 'Wbins' to 'WNbins'

Traceback (most recent call last):
  File "/arc/home/amandacook/zdm/papers/Repeaters/fit_repeating_distributions.py", line 769, in <module>
    main()
  File "/arc/home/amandacook/zdm/papers/Repeaters/fit_repeating_distributions.py", line 73, in main
    states,names=st.get_states()
                 ^^^^^^^^^^^^^^^
  File "/arc/home/amandacook/zdm/papers/Repeaters/states.py", line 157, in get_states
    state=set_state(psets[ipset],chime_response=True,ischime=True)
          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/arc/home/amandacook/zdm/papers/Repeaters/states.py", line 99, in set_state
    state = loading.set_state(alpha_method=1)
            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/arc/home/amandacook/zdm/zdm/MC_sample/loading.py", line 86, in set_state
    state.update_param_dict(vparams)
  File "/arc/home/amandacook/zdm/zdm/data_class.py", line 83, in update_param_dict
    self.update_param(ikey, params[key][ikey])
  File "/arc/home/amandacook/zdm/zdm/data_class.py", line 102, in update_param
    DC = self.params[param]
         ~~~~~~~~~~~^^^^^^^
KeyError: 'Wscale'

[cwjames1983]

Hi Mawson. Noting officially here two things:
#1: the papers directory is never updated when new versions of the code come out - it's just impossible to go back and redo things. What this means is that it's quite possible that running /papers/..." with newer code causes errors, and I don't intend to maintain it. HOWEVER - happy to help here! I'll check the details here

[cwjames1983]

#2: I see older code used zdm/MCMC/loading. This is deprecated - just go for zdm/loading

#3: I expect there will be more bugs, so please keep pinging me here. This could be a good place for a general "get fit repetition working again!"