jaxopt.ScipyBoundedMinimize

class jaxopt.ScipyBoundedMinimize(method=None, dtype=<class 'numpy.float64'>, jit=True, implicit_diff_solve=None, has_aux=False, fun=None, callback=None, tol=None, options=None, maxiter=500, value_and_grad=False)

scipy.optimize.minimize wrapper.

This wrapper is for minimization subject to box constraints only.

Parameters

• method (Optional[str]) –

• dtype (Optional[Any]) –

• jit (bool) –

• implicit_diff_solve (Optional[Callable]) –

• has_aux (bool) –

• fun (Callable) –

• callback (Callable) –

• tol (Optional[float]) –

• options (Optional[Dict[str, Any]]) –

• maxiter (int) –

• value_and_grad (Union[bool, Callable]) –

fun

a smooth function of the form fun(x, *args, **kwargs).

Type

Callable

method

the method argument for scipy.optimize.minimize.

Type

Optional[str]

tol

the tol argument for scipy.optimize.minimize.

Type

Optional[float]

options

the options argument for scipy.optimize.minimize.

Type

Optional[Dict[str, Any]]

dtype

if not None, cast all NumPy arrays to this dtype. Note that some methods relying on FORTRAN code, such as the L-BFGS-B solver for scipy.optimize.minimize, require casting to float64.

Type

Optional[Any]

jit

whether to JIT-compile JAX-based values and grad evals.

Type

bool

implicit_diff_solve

the linear system solver to use.

Type

Optional[Callable]

has_aux

whether function fun outputs one (False) or more values (True). When True it will be assumed by default that fun(…)[0] is the objective.

Type

bool

__init__(method=None, dtype=<class 'numpy.float64'>, jit=True, implicit_diff_solve=None, has_aux=False, fun=None, callback=None, tol=None, options=None, maxiter=500, value_and_grad=False)

Parameters

• method (Optional[str]) –

• dtype (Optional[Any]) –

• jit (bool) –

• implicit_diff_solve (Optional[Callable]) –

• has_aux (bool) –

• fun (Optional[Callable]) –

• callback (Optional[Callable]) –

• tol (Optional[float]) –

• options (Optional[Dict[str, Any]]) –

• maxiter (int) –

• value_and_grad (Union[bool, Callable]) –

Return type

None

Methods

__init__([method, dtype, jit, ...])
attribute_names()
attribute_values()
l2_optimality_error(params, *args, **kwargs)	Computes the L2 optimality error.
optimality_fun(sol, bounds, *args, **kwargs)	Optimality function mapping compatible with @custom_root.
run(init_params, bounds, *args, **kwargs)	Runs the solver.

Attributes

callback
fun
has_aux
implicit_diff_solve
jit
maxiter
method
options
tol
value_and_grad

dtype

alias of float64

l2_optimality_error(params, *args, **kwargs)

Computes the L2 optimality error.

optimality_fun(sol, bounds, *args, **kwargs)

Optimality function mapping compatible with @custom_root.

run(init_params, bounds, *args, **kwargs)

Runs the solver.

Parameters

• init_params (Any) – pytree containing the initial parameters.

• bounds (Optional[Any]) – an optional tuple (lb, ub) of pytrees with structure identical to init_params, representing box constraints.

• *args – additional positional arguments to be passed to fun.

• **kwargs – additional keyword arguments to be passed to fun.

Return type

OptStep

Returns

(params, info).