Pass user-defined parameter to MACRO

[Aprillmylmy]

Dear developer,

I am wondering if it is possible to pass a user-specific parameter (e.g., maxloop=500 or similar constant parameters) into macro explicitly from the Jupyter notebook, instead of making changes from GAMS?

Thanks very much,
Emily

[khaeru]

Hi Emily—thanks for the question!

Yes, this is supported. Please have a look at the following:

• Scenario.init_scalar() or perhaps .init_par()—these methods allow to define a new 0-d or 1+-dimensional parameter, like your "maxloop", so that data can then be stored with .change_scalar() or .add_par(), respectively.
• When scenario.solve() is called, all data (sets and parameters) are written to GDX file. This will include your newly-defined one.
• You will still need to modify the GAMS code to retrieve that value from the GDX file in order to be available as a GAMS variable for use by other code. You should do that around here:

  message_ix/message_ix/model/MACRO/macro_data_load.gms

  Lines 122 to 139 in b572cce

  	$GDXIN '%in%'
  	$IF %macromode% == 'linked' $GOTO macro_data
  	$LOAD node
  	$LOAD year_all = year,type_year,cat_year,time
  	$LOAD duration_period
  	$LOAD commodity,level
  	$LOAD interestrate
  	$LABEL macro_data
  	$LOAD type_node,cat_node
  	$LOAD sector,mapping_macro_sector
  	$LOAD kpvs,kgdp,esub,depr,drate,lotol
  	$LOAD lakl,prfconst
  	$LOAD aeei,grow
  	$LOAD gdp_calibrate,historical_gdp
  	$LOAD demand_MESSAGE,price_MESSAGE,cost_MESSAGE
  	$GDXIN

  
  After having done this once, then afterwards you can use the Python API to change the value and every .solve() call will pass your custom parameter value through to the GAMS code.

I hope that helps! Please mark the answer ✅ if it covered your initial question, and either add a reply or start a new thread if you have others.