InteriorPointSolver

(class from pyomo.contrib.interior_point.interior_point)

class pyomo.contrib.interior_point.interior_point.InteriorPointSolver(linear_solver, max_iter=100, tol=1e-08, linear_solver_log_filename=None, max_reallocation_iterations=5, reallocation_factor=2)[source]

Bases: object

Class for creating interior point solvers with different options

__init__(linear_solver, max_iter=100, tol=1e-08, linear_solver_log_filename=None, max_reallocation_iterations=5, reallocation_factor=2)[source]

Methods

__init__(linear_solver[, max_iter, tol, ...])

check_convergence(barrier[, timer])

factorize(kkt[, timer])

fraction_to_the_boundary()

process_init(x, lb, ub)

process_init_duals_lb(x, lb)

process_init_duals_ub(x, ub)

set_interface(interface)

set_linear_solver(linear_solver)

This method exists to hopefully make it easy to try the same IP algorithm with different linear solvers.

solve(interface[, timer, report_timing])

update_barrier_parameter()

Member Documentation

check_convergence(barrier, timer=None)[source]
Parameters:
Returns:

  • primal_inf (float)

  • dual_inf (float)

  • complimentarity_inf (float)

set_linear_solver(linear_solver)[source]

This method exists to hopefully make it easy to try the same IP algorithm with different linear solvers. Subclasses may have linear-solver specific methods, in which case this should not be called.

Hopefully the linear solver interface can be standardized such that this is not a problem. (Need a generalized method for set_options)

solve(interface, timer=None, report_timing=False)[source]
Parameters: