Objectives API¶

Objectives define numerical targets for Model.tune.

Group firing-rate targets¶

objectives = jtfne.rate_targets(
    groups={"first_half": range(24), "second_half": range(24, 48)},
    targets_hz={"first_half": 5.0, "second_half": 10.0},
)

AGSDR optimizer spec¶

optimizer = jtfne.agsdr(
    parameters={"drive_scale_a": (0.35, 2.25), "drive_scale_b": (0.35, 2.25)},
    generations=8,
    population_size=6,
    seed=42,
)

Tune¶

result = model.tune(objectives=objectives, optimizer=optimizer)
print(result.best_score)
print(result.best_parameters)
print(result.summary)

Single objective¶

objective = jtfne.objective(name="rate", metric="spike_rate_hz", target=10.0)
result = model.tune(objective=objective, parameter="drive_gain", bounds=(0.5, 2.0))

Rate + synchrony targets¶

rate_synchrony_targets builds an Objective with a population-rate term and a synchrony (kappa) term. All four arguments are defaulted to the canonical balanced operating point, so jtfne.rate_synchrony_targets() is the standard starting objective for laminar tuning.

Parameter	Default	Meaning
`target_rate_hz`	`10.0`	Target population firing rate (Hz).
`target_kappa_synchrony`	`0.0`	Target synchrony. `0.0` = asynchronous-irregular; required for an unbiased spectrolaminar readout (a global rhythm masks laminar band structure).
`rate_weight`	`1.0`	Weight of the rate term.
`synchrony_weight`	`0.25`	Weight of the synchrony term.

Returns: an Objective usable with Model.evaluate / Model.tune.

obj = jtfne.rate_synchrony_targets()                  # 10 Hz, kappa 0 (canonical)
obj = jtfne.rate_synchrony_targets(target_rate_hz=5.0, synchrony_weight=0.5)
result = model.tune(obj, optimizer="AGSDR", steps=50)

Result object¶

TuneResult exposes:

best_score
best_parameters
history
summary
model
to_dict()