pymrm.solve

Nonlinear-solver utilities used by pymrm.

Public API¶

Symbol	Type	Summary
`clip_approach`	function	Project values onto bounds, optionally with a relaxed approach rule.
`newton`	function	Solve `function(x) = 0` with Newton iterations.

Project values onto bounds, optionally with a relaxed approach rule.

values (numpy.ndarray) Values to modify in place.
dummy (Any) Placeholder argument kept for API compatibility.
lower_bounds, upper_bounds (float or numpy.ndarray, optional) Lower and upper bounds. Scalars and broadcastable arrays are supported.
factor (float, optional) Relaxation factor for out-of-bound entries. 0 applies strict clipping. Non-zero values apply a linear approach update toward the violated bound.

Solve function(x) = 0 with Newton iterations.

function (callable) Callable with signature function(x, *args) -> (residual, jacobian).
initial_guess (numpy.ndarray) Starting point of the iterations.
args (tuple, optional) Extra positional arguments passed to function.
tol (float, optional) Stopping tolerance on the infinity norm of the Newton update.
maxfev (int, optional) Maximum number of Newton iterations.
solver ({‘spsolve’, ‘cg’, ‘bicgstab’} or callable, optional) Linear solver used for each Newton step. If None, the routine picks 'spsolve' for smaller systems and 'bicgstab' for larger systems. A callable solver must accept (jac_matrix, rhs, **kwargs) and return the solution vector.
lin_solver_kwargs (dict, optional) Keyword arguments forwarded to the selected linear solver.
callback (callable, optional) Optional hook called as callback(x, residual) after each iteration.

scipy.optimize.OptimizeResult Result object with fields x, success, nit, fun, and message.