Steady-State Warm Start

warm_start.py modules

ogzaf.warm_start

Warm-start the steady-state solve from a previously solved state.

OG-Core cold-starts every steady-state solve from constants (household savings and labour at fixed values, the outer unknowns from the initial_guess_* parameters). That seed can sit far from the solution, and the single constants OG-Core accepts for b and n cannot carry the shape of a solved distribution. This module seeds the solver from a state that has already solved: the household matrices and every outer unknown together, so they are mutually consistent, with factor passed directly.

b and n are in model units, so a seed stays valid across modest parameter moves. Regenerate it with python examples/save_warm_start.py after any large recalibration.

Ported from OG-KOR’s ogkor/warm_start.py (itself from OG-JPN).

ogzaf.warm_start.enable(path='/home/runner/work/OG-ZAF/OG-ZAF/ogzaf/data/ss_warm_start.pkl')

Patch SS.SS_initial_guesses to start from the stored seed.

Call once before solving. Does nothing if no seed is stored, so a fresh checkout still runs (slowly) rather than failing.

Parameters:

path (str) – seed written by save_seed()

Returns:

True if a warm start is now active

Return type:

bool

ogzaf.warm_start.save_seed(ss_vars, path='/home/runner/work/OG-ZAF/OG-ZAF/ogzaf/data/ss_warm_start.pkl')

Store the parts of a solved steady state needed to warm-start another.

Parameters:

• ss_vars (dict) – a solved SS_vars.pkl

• path (str) – where to write the seed

Returns:

the path written

Return type:

str