def test_linear_combination_skims(self):
target = np.zeros((5, 5, 1))
source = np.random.rand(25).reshape(5, 5, 1)
linear_combination_skims(target, source, target, 0.8, 1)
self.assertEqual((source - target * 5 / 4).max(), 0,
'Linear combination failed')
def execute(self):
for c in self.traffic_classes:
c.graph.set_graph(self.time_field)
logger.info("{} Assignment STATS".format(self.algorithm))
logger.info("Iteration, RelativeGap, stepsize")
for self.iter in range(1, self.max_iter + 1):
if pyqt:
self.equilibration.emit(['rgap', self.rgap])
self.equilibration.emit(['iterations', self.iter])
flows = []
aon_flows = []
for c in self.traffic_classes:
aon = allOrNothing(c.matrix, c.graph, c._aon_results)
if pyqt:
aon.assignment.connect(self.signal_handler)
aon.execute()
c._aon_results.total_flows()
aon_flows.append(c._aon_results.total_link_loads * c.pce)
self.aon_total_flow = np.sum(aon_flows, axis=0)
if self.iter == 1:
for c in self.traffic_classes:
copy_two_dimensions(c.results.link_loads,
c._aon_results.link_loads, self.cores)
c.results.total_flows()
copy_one_dimension(c.results.total_link_loads,
c._aon_results.total_link_loads,
self.cores)
if c.results.num_skims > 0:
copy_three_dimensions(c.results.skims.matrix_view,
c._aon_results.skims.matrix_view,
self.cores)
flows.append(c.results.total_link_loads * c.pce)
else:
self.__calculate_step_direction()
self.calculate_stepsize()
for c in self.traffic_classes:
stp_dir = self.step_direction[c.mode]
cls_res = c.results
linear_combination(cls_res.link_loads, stp_dir.link_loads,
cls_res.link_loads, self.stepsize,
self.cores)
if cls_res.num_skims > 0:
linear_combination_skims(cls_res.skims.matrix_view,
stp_dir.skims.matrix_view,
cls_res.skims.matrix_view,
self.stepsize, self.cores)
cls_res.total_flows()
flows.append(cls_res.total_link_loads * c.pce)
self.fw_total_flow = np.sum(flows, axis=0)
# Check convergence
# This needs ot be done with the current costs, and not the future ones
if self.iter > 1:
if self.check_convergence():
if self.steps_below >= self.steps_below_needed_to_terminate:
break
else:
self.steps_below += 1
self.vdf.apply_vdf(self.congested_time, self.fw_total_flow,
self.capacity, self.free_flow_tt,
*self.vdf_parameters)
for c in self.traffic_classes:
c.graph.cost = self.congested_time
if self.time_field in c.graph.skim_fields:
idx = c.graph.skim_fields.index(self.time_field)
c.graph.skims[:, idx] = self.congested_time[:]
c._aon_results.reset()
logger.info("{},{},{}".format(self.iter, self.rgap, self.stepsize))
if self.rgap > self.rgap_target:
logger.error("Desired RGap of {} was NOT reached".format(
self.rgap_target))
logger.info(
f"{self.algorithm} Assignment finished. {self.iter} iterations and {self.rgap} final gap"
)
if pyqt:
self.equilibration.emit(['rgap', self.rgap])
self.equilibration.emit(['iterations', self.iter])
self.equilibration.emit(['finished_threaded_procedure'])
def __calculate_step_direction(self):
"""Calculates step direction depending on the method"""
sd_flows = []
# 2nd iteration is a fw step. if the previous step replaced the aggregated
# solution so far, we need to start anew.
if ((self.iter == 2) or (self.stepsize == 1.0) or (self.do_fw_step)
or (self.algorithm == "frank-wolfe")
or (self.algorithm == "msa")):
# logger.info("FW step")
self.do_fw_step = False
self.do_conjugate_step = True
self.conjugate_stepsize = 0.0
for c in self.traffic_classes:
aon_res = c._aon_results
stp_dir_res = self.step_direction[c.mode]
copy_two_dimensions(stp_dir_res.link_loads, aon_res.link_loads,
self.cores)
if c.results.num_skims > 0:
copy_three_dimensions(stp_dir_res.skims.matrix_view,
aon_res.skims.matrix_view,
self.cores)
aon_res.total_flows()
sd_flows.append(aon_res.total_link_loads * c.pce)
# 3rd iteration is cfw. also, if we had to reset direction search we need a cfw step before bfw
elif (self.iter == 3) or (self.do_conjugate_step) or (self.algorithm
== "cfw"):
self.do_conjugate_step = False
self.calculate_conjugate_stepsize()
# logger.info("CFW step, conjugate stepsize = {}".format(self.conjugate_stepsize))
for c in self.traffic_classes:
stp_dr = self.step_direction[c.mode]
pre_previous = self.pre_previous_step_direction[c.mode]
copy_two_dimensions(pre_previous.link_loads, stp_dr.link_loads,
self.cores)
if c.results.num_skims > 0:
copy_three_dimensions(pre_previous.skims.matrix_view,
stp_dr.skims.matrix_view, self.cores)
linear_combination(stp_dr.link_loads, stp_dr.link_loads,
c._aon_results.link_loads,
self.conjugate_stepsize, self.cores)
if c.results.num_skims > 0:
linear_combination_skims(stp_dr.skims.matrix_view,
stp_dr.skims.matrix_view,
c._aon_results.skims.matrix_view,
self.conjugate_stepsize,
self.cores)
sd_flows.append(np.sum(stp_dr.link_loads, axis=1) * c.pce)
# biconjugate
else:
self.calculate_biconjugate_direction()
# deep copy because we overwrite step_direction but need it on next iteration
for c in self.traffic_classes:
ppst = self.pre_previous_step_direction[
c.mode] # type: AssignmentResults
prev_stp_dir = self.previous_step_direction[
c.mode] # type: AssignmentResults
stp_dir = self.step_direction[
c.mode] # type: AssignmentResults
copy_two_dimensions(ppst.link_loads, stp_dir.link_loads,
self.cores)
if c.results.num_skims > 0:
copy_three_dimensions(ppst.skims.matrix_view,
stp_dir.skims.matrix_view,
self.cores)
triple_linear_combination(stp_dir.link_loads,
c._aon_results.link_loads,
stp_dir.link_loads,
prev_stp_dir.link_loads, self.betas,
self.cores)
if c.results.num_skims > 0:
triple_linear_combination_skims(
stp_dir.skims.matrix_view,
c._aon_results.skims.matrix_view,
stp_dir.skims.matrix_view,
prev_stp_dir.skims.matrix_view, self.betas, self.cores)
sd_flows.append(np.sum(stp_dir.link_loads, axis=1) * c.pce)
copy_two_dimensions(prev_stp_dir.link_loads, ppst.link_loads,
self.cores)
if c.results.num_skims > 0:
copy_three_dimensions(prev_stp_dir.skims.matrix_view,
ppst.skims.matrix_view, self.cores)
self.step_direction_flow = np.sum(sd_flows, axis=0)
def execute(self):
for c in self.traffic_classes:
c.graph.set_graph(self.time_field)
logger.info(f"{self.algorithm} Assignment STATS")
logger.info("Iteration, RelativeGap, stepsize")
for self.iter in range(1, self.max_iter + 1):
self.iteration_issue = []
if pyqt:
self.equilibration.emit(["rgap", self.rgap])
self.equilibration.emit(["iterations", self.iter])
aon_flows = []
for c in self.traffic_classes:
aon = allOrNothing(c.matrix, c.graph, c._aon_results)
if pyqt:
aon.assignment.connect(self.signal_handler)
aon.execute()
aon_flows.append(c._aon_results.total_link_loads * c.pce)
self.aon_total_flow = np.sum(aon_flows, axis=0)
flows = []
if self.iter == 1:
for c in self.traffic_classes:
copy_two_dimensions(c.results.link_loads,
c._aon_results.link_loads, self.cores)
copy_one_dimension(c.results.total_link_loads,
c._aon_results.total_link_loads,
self.cores)
if c.results.num_skims > 0:
copy_three_dimensions(c.results.skims.matrix_view,
c._aon_results.skims.matrix_view,
self.cores)
flows.append(c.results.total_link_loads * c.pce)
else:
self.__calculate_step_direction()
self.calculate_stepsize()
for c in self.traffic_classes:
stp_dir = self.step_direction[c.mode]
cls_res = c.results
linear_combination(cls_res.link_loads, stp_dir.link_loads,
cls_res.link_loads, self.stepsize,
self.cores)
if cls_res.num_skims > 0:
linear_combination_skims(
cls_res.skims.matrix_view,
stp_dir.skims.matrix_view,
cls_res.skims.matrix_view,
self.stepsize,
self.cores,
)
cls_res.total_flows()
flows.append(cls_res.total_link_loads * c.pce)
self.fw_total_flow = np.sum(flows, axis=0)
# Check convergence
# This needs to be done with the current costs, and not the future ones
converged = False
if self.iter > 1:
converged = self.check_convergence()
self.convergence_report["iteration"].append(self.iter)
self.convergence_report["rgap"].append(self.rgap)
self.convergence_report["warnings"].append("; ".join(
self.iteration_issue))
self.convergence_report["alpha"].append(self.stepsize)
if self.algorithm == "bfw":
self.convergence_report["beta0"].append(self.betas[0])
self.convergence_report["beta1"].append(self.betas[1])
self.convergence_report["beta2"].append(self.betas[2])
logger.info(f"{self.iter},{self.rgap},{self.stepsize}")
if converged:
if self.steps_below >= self.steps_below_needed_to_terminate:
break
else:
self.steps_below += 1
self.vdf.apply_vdf(
self.congested_time,
self.fw_total_flow,
self.capacity,
self.free_flow_tt,
*self.vdf_parameters,
self.cores,
)
for c in self.traffic_classes:
aggregate_link_costs(self.congested_time, c.graph.compact_cost,
c.results.crosswalk)
if self.time_field in c.graph.skim_fields:
idx = c.graph.skim_fields.index(self.time_field)
c.graph.skims[:, idx] = self.congested_time[:]
c._aon_results.reset()
if self.rgap > self.rgap_target:
logger.error(f"Desired RGap of {self.rgap_target} was NOT reached")
logger.info(
f"{self.algorithm} Assignment finished. {self.iter} iterations and {self.rgap} final gap"
)
if pyqt:
self.equilibration.emit(["rgap", self.rgap])
self.equilibration.emit(["iterations", self.iter])
self.equilibration.emit(["finished_threaded_procedure"])
def execute(self):
# We build the fixed cost field
for c in self.traffic_classes:
if c.fixed_cost_field:
# divide fixed cost by volume-dependent prefactor (vot) such that we don't have to do it for
# each occurence in the objective funtion. TODO: Need to think about cost skims here, we do
# not want this there I think
c.fixed_cost[c.graph.graph.__supernet_id__] = (
c.graph.graph[c.fixed_cost_field].values[:] *
c.fc_multiplier / c.vot)
c.fixed_cost[np.isnan(c.fixed_cost)] = 0
# TODO: Review how to eliminate this. It looks unnecessary
# Just need to create some arrays for cost
for c in self.traffic_classes:
c.graph.set_graph(self.time_field)
logger.info(f"{self.algorithm} Assignment STATS")
logger.info("Iteration, RelativeGap, stepsize")
for self.iter in range(1, self.max_iter + 1):
self.iteration_issue = []
if pyqt:
self.equilibration.emit(["rgap", self.rgap])
self.equilibration.emit(["iterations", self.iter])
aon_flows = []
self.__maybe_create_path_file_directories()
for c in self.traffic_classes: # type: TrafficClass
# cost = c.fixed_cost / c.vot + self.congested_time # now only once
cost = c.fixed_cost + self.congested_time
aggregate_link_costs(cost, c.graph.compact_cost,
c.results.crosswalk)
aon = allOrNothing(c.matrix, c.graph, c._aon_results)
if pyqt:
aon.assignment.connect(self.signal_handler)
aon.execute()
c._aon_results.link_loads *= c.pce
c._aon_results.total_flows()
aon_flows.append(c._aon_results.total_link_loads)
self.aon_total_flow = np.sum(aon_flows, axis=0)
flows = []
if self.iter == 1:
for c in self.traffic_classes:
copy_two_dimensions(c.results.link_loads,
c._aon_results.link_loads, self.cores)
c.results.total_flows()
if c.results.num_skims > 0:
copy_three_dimensions(c.results.skims.matrix_view,
c._aon_results.skims.matrix_view,
self.cores)
flows.append(c.results.total_link_loads)
if self.algorithm == "all-or-nothing":
break
else:
self.__calculate_step_direction()
self.calculate_stepsize()
for c in self.traffic_classes:
stp_dir = self.step_direction[c.__id__]
cls_res = c.results
linear_combination(cls_res.link_loads, stp_dir.link_loads,
cls_res.link_loads, self.stepsize,
self.cores)
if cls_res.num_skims > 0:
linear_combination_skims(
cls_res.skims.matrix_view,
stp_dir.skims.matrix_view,
cls_res.skims.matrix_view,
self.stepsize,
self.cores,
)
cls_res.total_flows()
flows.append(cls_res.total_link_loads)
self.fw_total_flow = np.sum(flows, axis=0)
# Check convergence
# This needs to be done with the current costs, and not the future ones
converged = self.check_convergence() if self.iter > 1 else False
self.vdf.apply_vdf(
self.congested_time,
self.fw_total_flow,
self.capacity,
self.free_flow_tt,
*self.vdf_parameters,
self.cores,
)
self.convergence_report["iteration"].append(self.iter)
self.convergence_report["rgap"].append(self.rgap)
self.convergence_report["warnings"].append("; ".join(
self.iteration_issue))
self.convergence_report["alpha"].append(self.stepsize)
if self.algorithm in ["cfw", "bfw"]:
self.convergence_report["beta0"].append(self.betas[0])
self.convergence_report["beta1"].append(self.betas[1])
self.convergence_report["beta2"].append(self.betas[2])
for c in self.traffic_classes:
c._aon_results.reset()
if self.time_field not in c.graph.skim_fields:
continue
idx = c.graph.skim_fields.index(self.time_field)
c.graph.skims[:, idx] = self.congested_time[:]
logger.info(f"{self.iter},{self.rgap},{self.stepsize}")
if converged:
self.steps_below += 1
if self.steps_below >= self.steps_below_needed_to_terminate:
break
else:
self.steps_below = 0
for c in self.traffic_classes:
c.results.link_loads /= c.pce
c.results.total_flows()
# TODO (Jan 18/4/21): Do we want to blob store path files (by iteration, class, origin, destination) in sqlite?
# or do we just use one big hdf5 file?
if (self.rgap > self.rgap_target) and (self.algorithm !=
"all-or-nothing"):
logger.error(f"Desired RGap of {self.rgap_target} was NOT reached")
logger.info(
f"{self.algorithm} Assignment finished. {self.iter} iterations and {self.rgap} final gap"
)
if pyqt:
self.equilibration.emit(["rgap", self.rgap])
self.equilibration.emit(["iterations", self.iter])
self.equilibration.emit(["finished_threaded_procedure"])
def execute(self):
for c in self.traffic_classes:
c.graph.set_graph(self.time_field)
logger.info("{} Assignment STATS".format(self.algorithm))
logger.info("Iteration, RelativeGap, stepsize")
for self.iter in range(1, self.max_iter + 1):
flows = []
aon_flows = []
for c in self.traffic_classes:
aon = allOrNothing(c.matrix, c.graph, c._aon_results)
aon.execute()
c._aon_results.total_flows()
aon_flows.append(c._aon_results.total_link_loads * c.pce)
self.aon_total_flow = np.sum(aon_flows, axis=0)
if self.iter == 1:
for c in self.traffic_classes:
copy_two_dimensions(c.results.link_loads, c._aon_results.link_loads, self.cores)
c.results.total_flows()
copy_one_dimension(c.results.total_link_loads, c._aon_results.total_link_loads, self.cores)
if c.results.num_skims > 0:
copy_three_dimensions(c.results.skims.matrix_view, c._aon_results.skims.matrix_view, self.cores)
flows.append(c.results.total_link_loads * c.pce)
else:
self.__calculate_step_direction()
self.calculate_stepsize()
for c in self.traffic_classes:
stp_dir = self.step_direction[c.mode]
cls_res = c.results
linear_combination(cls_res.link_loads, stp_dir.link_loads, cls_res.link_loads, self.stepsize,
self.cores)
# TODO: We need to compute the step direction for skims as well.
# It is probably a matter of transforming the step_direction values from numpy arrays to
# full AssignmentResults() ones, and cleaning the stuff we don't need
if cls_res.num_skims > 0:
linear_combination_skims(cls_res.skims.matrix_view,
stp_dir.skims.matrix_view,
cls_res.skims.matrix_view,
self.stepsize,
self.cores)
cls_res.total_flows()
flows.append(cls_res.total_link_loads * c.pce)
self.fw_total_flow = np.sum(flows, axis=0)
# Check convergence
# This needs ot be done with the current costs, and not the future ones
if self.iter > 1:
if self.check_convergence():
if self.steps_below >= self.steps_below_needed_to_terminate:
break
else:
self.steps_below += 1
self.vdf.apply_vdf(
self.congested_time, self.fw_total_flow, self.capacity, self.free_flow_tt, *self.vdf_parameters
)
for c in self.traffic_classes:
c.graph.cost = self.congested_time
c._aon_results.reset()
logger.info("{},{},{}".format(self.iter, self.rgap, self.stepsize))
if self.rgap > self.rgap_target:
logger.error("Desired RGap of {} was NOT reached".format(self.rgap_target))
logger.info(
"{} Assignment finished. {} iterations and {} final gap".format(self.algorithm, self.iter, self.rgap)
)
