parameter sweep

System Model Parameter Sweep

Parametrization of a System Model configuration that produces multiple configurations.

Set Parameters

Note: M values require up to a maximum of 2 distinct lengths

M = {
    'alpha': [1],
    'beta': [2, 5],
    'gamma': [3, 4],
    'omega': [7]
}

The parameters above produce 2 simulations.

• Simulation 1:
  • alpha = 1
  • beta = 2
  • gamma = 3
  • omega = 7
• Simulation 2:
  • alpha = 1
  • beta = 5
  • gamma = 4
  • omega = 7

All parameters can also be set to include a single parameter each, which will result in a single simulation.

Example State Updates

Previous State: y = 0

def state_update(_params, step, sH, s, _input, **kwargs):
    y = 'state'
    x = s['state'] + _params['alpha'] + _params['gamma']
    return y, x

• Updated State:
  • Simulation 1: y = 4 = 0 + 1 + 3
  • Simulation 2: y = 5 = 0 + 1 + 4

Example Policy Updates

# Internal States per Mechanism
def policies(_params, step, sH, s, **kwargs):
    return {'beta': _params['beta'], 'gamma': _params['gamma']}

• Simulation 1: {'beta': 2, 'gamma': 3]}
• Simulation 2: {'beta': 5, 'gamma': 4}

Configure Simulation

from cadCAD.configuration.utils import config_sim

g = {
    'alpha': [1],
    'beta': [2, 5],
    'gamma': [3, 4],
    'omega': [7]
}

sim_config = config_sim(
    {
        "N": 2,
        "T": range(5),
        "M": g,
    }
)

Example

* System Model Configuration