BlindLP module¶

class panco.blind.blindLP.BlindLP(network: Network, list_edges=None, filename='blind.lp')¶

Bases: object

Class for the analysis of a network with the linear programming methods. The network is decomposed into a forest (self.forest) with well-numbered servers, regarding the decomposition.

Parameters:

network (Network) – the network to analyze
list_edges (List[Tuple[int, int]]) – the edges to keep in the decomposition in a forest
filename (str) – name of the file to write the linear program

property all_delays: List[float]¶

Returns the delay bounds for all the flows

Returns:: the list of delay bounds

backlog(foi: int) → float¶

Computes the backlog of the flow of interest

Parameters:: foi (int) – the flow of interest
Returns:: the backlog of foi
Return type:: float

delay(foi: int) → float¶

Returns the delay bounds for flows foi

Parameters:: foi – the flow of interest
Returns:: the delay bound of foi

ff_analysis() → Network¶

Analysis performed if the network is feed-forward.

Returns:: the updated version of self.forest (updated burst parameters)
Return type:: Network

property ff_equiv: Network¶

Construct the equivalent network by solving the fix-point equations. If the network has not been decomposed, then returns the original network

Returns:: the equivalent network

lp_constraint_flow(foi: int, file)¶

Writes the linear constraints for flow foi of the forest decomposition

Parameters:

foi – flow of interest
file – file where the constraints are written

Returns:

None

lp_constraints(file)¶

Writes the constraints linear program in file

Parameters:: file – the file where the linear program is written
Returns:: None

property lp_program: ndarray¶

Writes the linear program and solves it to obtain the unknown burst where the flows have been cut

Returns:: the list of bursts of flows in the forest

update_sigma(f: int, sigma: ndarray) → ndarray¶

If the network is feed-forward, computes the burst parameters where the flows are cut. Checks if the burst of the flow has already been computed, otherwise, generate the linear programs needed for the computation.

Parameters:

f (int) – (cut) flow analyzed, its backlog is computed
sigma (np.ndarray) – the current array of backlogs for cur flows

Returns:

the new array of backlogs for cut flows

Return type:

np.ndarray