Remove global variables file and split it into multiple settings files

[BhargavaKrishnaSreepathi]

Global variables file is to be phased out. Transfer all the variables to the corresponding settings files in the respective folders

[BhargavaKrishnaSreepathi]

@daren-thomas: To resolve this issue, we can still save the relevant values from the globalvar.py into settings.py file which is local to sub folder like say optimization right? So remove the variables corresponding to optimization from globalvar.py and paste them into optimization_settings.py inside cea\optimization folder. Does this sound ok?

[daren-thomas]

you can. but what you really want to do is move them to the configuration file. see some of the pull requests i've posted that work in this direction. moving the variables to an optimization_settings.py file could be a valid first step, though. like a stepping stone...

[jimenofonseca]

@daren-thomas do you think we can close this issue now?

[jimenofonseca]

@daren-thomas ??

[daren-thomas]

@JIMENOFONSECA no. I don't think so. The issue is still open!

PR #1675 addresses a lot of the low hanging fruit, but there are still some parts of the optimization code and the neural network code that rely on this module.

[jimenofonseca]

@daren-thomas @BhargavaKrishnaSreepathi do you think we can close this issue now? if not, which is the plan of action?

[daren-thomas]

@JIMENOFONSECA It is still being used extensively in the metamodel/nn_generator scripts - what is the status of those? do they work? should they be moved to legacy if they don't?

Other places to fix:

• cea.demand.occupancy_model.main: uses gv.date_start (this can be fixed, right, @gabriel-happle?)
• cea.optimization.optimization_main.main: used in multiple places...
• cea.demand.calibration.subset_calibrator.subset_calibrator.ss_calibrator - @JIMENOFONSECA is this yours?
• cea.supply.supply_system_simulation.main - basically, because optimization still uses gv
• some other places that we can probably remove...
• conclusion: we're close. one last push?

[gabriel-happle]

@daren-thomas I will take a look at the first point and fix it.

[daren-thomas]

So there are still some places we need to look at:

• cea/analysis/mcda.py (@JIMENOFONSECA, I think this is your script?)
• cea.analysis.uncertainty.Individual_Evaluation (@shanshanhsieh, this should be an easy fix, I think the variables used are mainly in the config file - except "num_tot_buildings"?)
• cea/demand/subset_calibrator/ - (this folder still has some usages in it - I'm hesitant to change anything in here since I think most stuff will change...)
• cea/optimization/optimization_main.py:63 - (@shanshanhsieh, the optimization code uses a gv.num_tot_buildings - as well as some other small things. could you please take a look at that?)

[daren-thomas]

we can delete the cea/demand/subset_calibrator/ folder

[jimenofonseca]

@daren-thomas we take over this, so you can focus on the plots and the dashboard

[jimenofonseca]

2120 took down gv from the optimization_main.py and mcda.py scripts. the subset calibration should be taken out

[daren-thomas]

Daren is going to review it, @JIMENOFONSECA and @shanshanhsieh think it's done.

[daren-thomas]

STATUS UPDATE:

• class Thermal_Network takes a gv parameter in its constructor
  • how to test if removing that will create a problem?
  • btw: the class name should be ThermalNetwork and why is it defined in thermal_network_costs?
  • @shanshanhsieh can you check this out quickly?
• the following packages are old and will never ever work again without being updated:
  • cea.demand.calibration.subset_calibrator
  • cea.demand.metamodel.nn_generator
  • these need to be updated to work again or removed from the codebase
  • @JIMENOFONSECA you probably know more about this than anyone else, right?
• I'm creating a PR with the obvious fixes, the above remain as they might alter/change/remove functionality.