gadget_model

The decision cactus model consists of multiple gadgets.

class vnep_approx.gadget_model.AbstractGadget(name, request, in_node, out_nodes)

An abstract gadget used by GadgetContainerRequest.

Gadgets are connected by sharing a common node. Nodes that are designated to be used by multiple gadgets are called interface nodes.

A gadget has exactly one in-node and one or multiple out-nodes.

adapt_model_objective_according_to_local_profits(obj_expr)

extend_load_constraints(constraint_dict, nodes_handled_by_other_gadgets)

extend_mapping_by_own_solution(mapping, load, nodes_handled_by_other_gadgets)

generate_constraints(model)

generate_flow_induction_at_root_constraint(model)

generate_variables(model)

in_node = None

in-node of the gadget

interface_nodes = None

all interface nodes (in- and out-nodes)

name = None

gadget name

out_nodes = None

out-nodes of the gadget

reduce_flow_on_last_returned_mapping(used_flow)

request = None

the request graph of the gadget

set_substrate(substrate)

substrate = None

the substrate

verify_mapping(mapping)

Perform certain sanity checks usually performed by the alib.solutions.Mapping class.

Parameters:mapping (solutions.Mapping) – the request mapping Returns:a list of out-nodes that are mapped

class vnep_approx.gadget_model.CactusGadget(name, request, in_node, out_nodes, parent_name=None)

adapt_model_objective_according_to_local_profits(obj_expr)

extend_load_constraints(constraint_dict, nodes_handled_by_other_gadgets)

extend_mapping_by_own_solution(mapping, load, nodes_handled_by_other_gadgets)

generate_constraints(model)

generate_flow_induction_at_root_constraint(model)

generate_variables(model)

reduce_flow_on_last_returned_mapping(used_flow)

verify_mapping(mapping)

For a cactus gadget, each node and each edge must be mapped.

Parameters:mapping – Returns:

class vnep_approx.gadget_model.DecisionGadget(name, request, in_node, out_nodes)

adapt_model_objective_according_to_local_profits(obj_expr)

extend_load_constraints(constraint_dict, nodes_handled_by_other_gadgets)

extend_mapping_by_own_solution(mapping, load, nodes_handled_by_other_gadgets)

generate_constraints(model)

generate_flow_induction_at_root_constraint(model)

generate_variables(model)

reduce_flow_on_last_returned_mapping(used_flow)

verify_mapping(mapping)

exception vnep_approx.gadget_model.DecisionModelError

class vnep_approx.gadget_model.ExtendedDecisionGraph(parent_gadget)

class vnep_approx.gadget_model.GadgetContainerRequest(name, profit, rounding_threshold=0.001)

A request that consists of several gadgets.

The gadgets form a tree, there is exactly one root gadget and cycles are forbidden. Gadgets are connected by sharing a common node. Nodes that are designated to be used by multiple gadgets are called interface nodes.

adapt_objective(obj_expr)

Adapt the objective.

Parameters:obj_expr (gurobipy.LinExpr) – the objective expression Returns:the modified objective expression Return type:gurobipy.LinExpr

add_gadget(gadget)

Add a gadget to the request.

Parameters:gadget (AbstractGadget) – the new gadget

add_node_flow_var(model, i, u)

Return a node flow variable of a model.

Creates the variable if it does not exist.

Parameters:

• model (gurobipy.Model) – LP model
• i (str) – request node
• u (str) – substrate node

Returns:

the node flow variable

Return type:

gurobipy.Var

check_and_update()

Check if gadgets form a tree and set the root gadget.

Raises:GadgetError – if a check fails

check_gadget_tree()

Check if the gadgets are connected as a tree.

Raises:GadgetError – if a check fails

check_nodes()

Check interface nodes and inner nodes.

Raises:GadgetError – if a check fails

extract_mappings()

Generator that yields mappings of this GadgetContainerRequest.

Returns:yields tuples of (mapping, flow_value, load_dictionary)

gadgets = None

gadgets by name

gadgets_by_in_nodes = None

map in-nodes to a corresponding list of gadgets

generate_constraints(model)

Generate LP constraints for each gadget and induce flow at root gadget.

Parameters:model (gurobipy.Model) – LP model

generate_variables(model)

Generate LP variables for each gadget.

Parameters:model (gurobipy.Model) – LP model

get_gadget_tree_graph()

name = None

request name

profit = None

request profit

root_gadget = None

the root gadget

rounding_threshold = None

rounding threshold

set_substrate(substrate)

update_root()

Update root gadget.

A root gadget is a gadget with in-nodes that are not out-nodes of other gadgets.

Raises:GadgetError – if there is not exactly one root gadget

verify_request_mapping(mapping)

Verify a request mapping.

Parameters:mapping (solutions.Mapping) – the request mapping

exception vnep_approx.gadget_model.GadgetError

class vnep_approx.gadget_model.GadgetModelCreator(scenario, gurobi_settings=None, optimization_callback=None, lp_output_file=None, potential_iis_filename=None, logger=None)

create_constraints_other_than_bounding_loads_by_capacities()

create_objective()

create_variables_other_than_embedding_decision_and_request_load()

post_process_fractional_computation()

post_process_integral_computation()

recover_fractional_solution_from_variables()

recover_integral_solution_from_variables()

exception vnep_approx.gadget_model.MappingCombinationError