def test_ama_creator(self):
options = _get_base_options()
ama_options = {
"EF-2stage": True,
}
ama_options.update(options['opt'])
ama_object = ama.from_module(refmodelname,
options=ama_options,
use_command_line=False)
def test_ama_running(self):
options = _get_base_options()
ama_options = {"EF-2stage": True}
ama_options.update(options['opt'])
ama_object = ama.from_module(refmodelname,
ama_options,
use_command_line=False)
ama_object.run()
obj = round_pos_sig(ama_object.EF_Obj, 2)
self.assertEqual(obj, -130000)
def main():
solution_files = {"first_stage_solution":"farmer_first_stage.csv",
}
ama_options = {"EF-2stage": True, # We are solving directly the EF
"write_solution":solution_files}
#The module can be a local file
ama = amalgomator.from_module("afarmer", ama_options)
ama.run()
print("first_stage_solution=", ama.first_stage_solution)
print("inner bound=", ama.best_inner_bound)
print("outer bound=", ama.best_outer_bound)
def xhat_generator_farmer(scenario_names, solvername="gurobi", solver_options=None, crops_multiplier=1):
'''Farmer example applied to sequential sampling. Given scenario names and
options, create the scenarios and compute the xhat that is minimizing the
approximate probleme associatd with these scenarios.
Parameters
----------
scenario_names: int
Names of the scenario we use
solvername: str, optional
Name of the solver used. The default is "gurobi".
solver_options: dict, optional
Solving options. The default is None.
crops_multiplier: int, optional
A parameter of the farmer model. The default is 1.
Returns
-------
xhat: xhat object (dict containing a 'ROOT' key with a np.array)
A generated xhat.
'''
num_scens = len(scenario_names)
ama_options = { "EF-2stage": True,
"EF_solver_name": solvername,
"EF_solver_options": solver_options,
"use_integer": False,
"crops_multiplier": crops_multiplier,
"num_scens": num_scens,
"_mpisppy_probability": 1/num_scens,
}
#We use from_module to build easily an Amalgomator object
ama = amalgomator.from_module("mpisppy.tests.examples.farmer",
ama_options,use_command_line=False)
#Correcting the building by putting the right scenarios.
ama.scenario_names = scenario_names
ama.run()
# get the xhat
xhat = sputils.nonant_cache_from_ef(ama.ef)
return xhat
def xhat_generator_apl1p(scenario_names,
solvername="gurobi",
solver_options=None):
'''
For sequential sampling.
Takes scenario names as input and provide the best solution for the
approximate problem associated with the scenarios.
Parameters
----------
scenario_names: int
Names of the scenario we use
solvername: str, optional
Name of the solver used. The default is "gurobi"
solver_options: dict, optional
Solving options. The default is None.
Returns
-------
xhat: str
A generated xhat, solution to the approximate problem induced by scenario_names.
'''
num_scens = len(scenario_names)
ama_options = {
"EF-2stage": True,
"EF_solver_name": solvername,
"EF_solver_options": solver_options,
"num_scens": num_scens,
"_mpisppy_probability": 1 / num_scens,
}
#We use from_module to build easily an Amalgomator object
ama = amalgomator.from_module("mpisppy.tests.examples.apl1p",
ama_options,
use_command_line=False)
#Correcting the building by putting the right scenarios.
ama.scenario_names = scenario_names
ama.run()
# get the xhat
xhat = sputils.nonant_cache_from_ef(ama.ef)
return xhat
def main():
solution_files = {
"first_stage_solution": "uc_first_stage.csv",
#"tree_solution":"uc_ama_full_solution"
#It takes too long to right the full solution
}
ama_options = {
"2stage": True, # 2stage vs. mstage
"cylinders": ['ph', 'xhatshuffle', 'lagranger'],
"extensions": ['fixer'],
"id_fix_list_fct":
id_fix_list_fct, #Needed for fixer, but not passed in baseparsers
"write_solution": solution_files
}
ama = amalgomator.from_module("uc_funcs", ama_options)
ama.run()
if ama.on_hub:
print("first_stage_solution=", ama.first_stage_solution)
print("inner bound=", ama.best_inner_bound)
print("outer bound=", ama.best_outer_bound)
ama_extraargs.add_argument(
"--MMW-num-batches",
help="number of batches used for MMW confidence interval (default 1)",
dest="num_batches",
type=int,
default=1)
ama_extraargs.add_argument(
"--MMW-batch-size",
help="batch size used for MMW confidence interval (default None)",
dest="batch_size",
type=int,
default=None) #None means take batch_size=num_scens
ama_object = ama.from_module(refmodel,
ama_options,
extraargs=ama_extraargs)
ama_object.run()
if global_rank == 0:
print("inner bound=", ama_object.best_inner_bound)
# This the xhat of the left term of LHS of MMW (9)
print("outer bound=", ama_object.best_outer_bound)
########### get the nonants (the xhat)
nonant_cache = sputils.nonant_cache_from_ef(ama_object.ef)
ciutils.write_xhat(nonant_cache, path="xhat.npy")
#Set parameters for run()
options = ama_object.options
if __name__ == "__main__":
bfs = [4, 3, 2]
num_scens = np.prod(bfs) #To check with a full tree
ama_options = {
"EF-mstage": True,
"EF_solver_name": "gurobi_direct",
"num_scens": num_scens,
"_mpisppy_probability": 1 / num_scens,
"BFs": bfs,
"mudev": 0,
"sigmadev": 80
}
refmodel = "mpisppy.tests.examples.aircond_submodels" # WARNING: Change this in SPInstances
#We use from_module to build easily an Amalgomator object
ama = amalgomator.from_module(refmodel,
ama_options,
use_command_line=False)
ama.run()
print(f"inner bound=", ama.best_inner_bound)
print(f"outer bound=", ama.best_outer_bound)
from mpisppy.confidence_intervals.mmw_ci import MMWConfidenceIntervals
options = ama.options
options['solver_options'] = options['EF_solver_options']
xhat = sputils.nonant_cache_from_ef(ama.ef)['ROOT']
num_batches = 10
batch_size = 100
mmw = MMWConfidenceIntervals(refmodel,
options,
def xhat_generator_aircond(scenario_names,
solvername="gurobi",
solver_options=None,
BFs=[3, 2, 3],
mudev=0,
sigmadev=40,
start_seed=0):
'''
For sequential sampling.
Takes scenario names as input and provide the best solution for the
approximate problem associated with the scenarios.
Parameters
----------
scenario_names: int
Names of the scenario we use
solvername: str, optional
Name of the solver used. The default is "gurobi"
solver_options: dict, optional
Solving options. The default is None.
BFs: list, optional
Branching factors of the scenario 3. The default is [3,2,3]
(a 4 stage model with 18 different scenarios)
mudev: float, optional
The average deviation of demand between two stages; The default is 0.
sigma_dev: float, optional
The standard deviation from mudev for the demand difference between
two stages. The default is 40.
start_seed: int, optional
The starting seed, used to create different sample scenario trees.
The default is 0.
Returns
-------
xhat: str
A generated xhat, solution to the approximate problem induced by scenario_names.
NOTE: This tool only works when the file is in mpisppy. In SPInstances,
you must change the from_module line.
'''
num_scens = len(scenario_names)
ama_options = {
"EF-mstage": True,
"EF_solver_name": solvername,
"EF_solver_options": solver_options,
"num_scens": num_scens,
"_mpisppy_probability": 1 / num_scens,
"BFs": BFs,
"mudev": mudev,
"start_seed": start_seed,
"sigmadev": sigmadev
}
#We use from_module to build easily an Amalgomator object
ama = amalgomator.from_module("mpisppy.tests.examples.aircond_submodels",
ama_options,
use_command_line=False)
#Correcting the building by putting the right scenarios.
ama.scenario_names = scenario_names
ama.verbose = False
ama.run()
# get the xhat
xhat = sputils.nonant_cache_from_ef(ama.ef)
return {'ROOT': xhat['ROOT']}
return xhats, seed
if __name__ == "__main__":
BFs = [3, 2, 4, 4]
num_scens = np.prod(BFs)
mname = "mpisppy.tests.examples.aircond_submodels"
ama_options = {
"EF-mstage": True,
"num_scens": num_scens,
"_mpisppy_probability": 1 / num_scens,
"BFs": BFs,
}
#We use from_module to build easily an Amalgomator object
ama = amalgomator.from_module(mname, ama_options, use_command_line=False)
ama.run()
# get the xhat
xhat_one = sputils.nonant_cache_from_ef(ama.ef)['ROOT']
#----------Find a feasible solution for a single scenario-------------
scenario = ama.ef.scen0
seed = sputils.number_of_nodes(BFs)
options = dict() #We take default aircond options
xhats, seed = feasible_solution(mname, scenario, xhat_one, BFs, seed,
options)
print(xhats)
#----------Find feasible solutions for every scenario ------------
return None
else:
if verbose and self.cylinder_rank == 0:
print(" Feasible xhat found")
return self.Eobjective(verbose=verbose)
if __name__ == "__main__":
#==============================
# hardwired by dlw for debugging (this main is like MMW, but strange)
import mpisppy.tests.examples.farmer as refmodel
import mpisppy.utils.amalgomator as ama
# do the right term of MMW (9) using the first scenarios
ama_options = {"EF-2stage": True} # 2stage vs. mstage
ama_object = ama.from_module("mpisppy.tests.examples.farmer", ama_options)
ama_object.run()
print(f"inner bound=", ama_object.best_inner_bound)
# This the right term of LHS of MMW (9)
print(f"outer bound=", ama_object.best_outer_bound)
############### now get an xhat using different scenarios
# (use use the ama args to get problem parameters)
ScenCount = ama_object.options['num_scens']
scenario_creator = refmodel.scenario_creator
scenario_denouement = refmodel.scenario_denouement
crops_multiplier = ama_object.options['crops_multiplier']
solvername = ama_object.options['EF_solver_name']
scenario_creator_kwargs = {
"use_integer": False,
def gap_estimators(xhat_one,
mname,
solving_type="EF-2stage",
scenario_names=None,
sample_options=None,
ArRP=1,
scenario_creator_kwargs={},
scenario_denouement=None,
solvername='gurobi',
solver_options=None,
verbose=True,
objective_gap=False
):
''' Given a xhat, scenario names, a scenario creator and options,
gap_estimators creates a scenario tree and the associatd estimators
G and s from ยง2 of [bm2011].
Returns G and s evaluated at xhat.
If ArRP>1, G and s are pooled, from a number ArRP of estimators,
computed with different scenario trees.
Parameters
----------
xhat_one : dict
A candidate first stage solution
mname: str
Name of the reference model, e.g. 'mpisppy.tests.examples.farmer'.
solving_type: str, optional
The way we solve the approximate problem. Can be "EF-2stage" (default)
or "EF-mstage".
scenario_names: list, optional
List of scenario names used to compute G_n and s_n. Default is None
Must be specified for 2 stage, but can be missing for multistage
sample_options: dict, optional
Only for multistage. Must contain a 'seed' and a 'BFs' attribute,
specifying the starting seed and the branching factors
of the scenario tree
ArRP:int,optional
Number of batches (we create a ArRP model). Default is 1 (one batch).
scenario_creator_kwargs: dict, optional
Additional arguments for scenario_creator. Default is {}
scenario_denouement: function, optional
Function to run after scenario creation. Default is None.
solvername : str, optional
Solver. Default is 'gurobi'
solver_options: dict, optional
Solving options. Default is None
verbose: bool, optional
Should it print the gap estimator ? Default is True
objective_gap: bool, optional
Returns a gap estimate around approximate objective value
BFs: list, optional
Only for multistage. List of branching factors of the sample scenario tree.
Returns
-------
G_k and s_k, gap estimator and associated standard deviation estimator.
'''
if solving_type not in ["EF-2stage","EF-mstage"]:
raise RuntimeError("Only EF solve for the approximate problem is supported yet.")
else:
is_multi = (solving_type=="EF-mstage")
if is_multi:
try:
BFs = sample_options['BFs']
start = sample_options['seed']
except (TypeError,KeyError,RuntimeError):
raise RuntimeError('For multistage problems, sample_options must be a dict with BFs and seed attributes.')
else:
start = sputils.extract_num(scenario_names[0])
if ArRP>1: #Special case : ArRP, G and s are pooled from r>1 estimators.
if is_multi:
raise RuntimeError("Pooled estimators are not supported for multistage problems yet.")
n = len(scenario_names)
if(n%ArRP != 0):
raise RuntimeWarning("You put as an input a number of scenarios"+\
f" which is not a mutliple of {ArRP}.")
n = n- n%ArRP
G =[]
s = []
for k in range(ArRP):
scennames = scenario_names[k*(n//ArRP):(k+1)*(n//ArRP)]
tmp = gap_estimators(xhat_one, mname,
solvername=solvername,
scenario_names=scennames, ArRP=1,
scenario_creator_kwargs=scenario_creator_kwargs,
scenario_denouement=scenario_denouement,
solver_options=solver_options,
solving_type=solving_type
)
G.append(tmp['G'])
s.append(tmp['s'])
#Pooling
G = np.mean(G)
s = np.linalg.norm(s)/np.sqrt(n//ArRP)
return {"G": G, "s": s, "seed": start}
#A1RP
#We start by computing the optimal solution to the approximate problem induced by our scenarios
if is_multi:
#Sample a scenario tree: this is a subtree, but starting from stage 1
samp_tree = sample_tree.SampleSubtree(mname,
xhats =[],
root_scen=None,
starting_stage=1,
BFs=BFs,
seed=start,
options=scenario_creator_kwargs,
solvername=solvername,
solver_options=solver_options)
samp_tree.run()
start += sputils.number_of_nodes(BFs)
ama_object = samp_tree.ama
else:
#We use amalgomator to do it
ama_options = dict(scenario_creator_kwargs)
ama_options['start'] = start
ama_options['num_scens'] = len(scenario_names)
ama_options['EF_solver_name'] = solvername
ama_options['EF_solver_options'] = solver_options
ama_options[solving_type] = True
ama_object = ama.from_module(mname, ama_options,use_command_line=False)
ama_object.scenario_names = scenario_names
ama_object.verbose = False
ama_object.run()
start += len(scenario_names)
#Optimal solution of the approximate problem
zstar = ama_object.best_outer_bound
#Associated policies
xstars = sputils.nonant_cache_from_ef(ama_object.ef)
#Then, we evaluate the fonction value induced by the scenario at xstar.
if is_multi:
# Find feasible policies (i.e. xhats) for every non-leaf nodes
if len(samp_tree.ef._ef_scenario_names)>1:
local_scenarios = {sname:getattr(samp_tree.ef,sname) for sname in samp_tree.ef._ef_scenario_names}
else:
local_scenarios = {samp_tree.ef._ef_scenario_names[0]:samp_tree.ef}
xhats,start = sample_tree.walking_tree_xhats(mname,
local_scenarios,
xhat_one['ROOT'],
BFs,
start,
scenario_creator_kwargs,
solvername=solvername,
solver_options=solver_options)
#Compute then the average function value with this policy
scenario_creator_kwargs = samp_tree.ama.kwargs
all_nodenames = sputils.create_nodenames_from_BFs(BFs)
else:
#In a 2 stage problem, the only non-leaf is the ROOT node
xhats = xhat_one
all_nodenames = None
xhat_eval_options = {"iter0_solver_options": None,
"iterk_solver_options": None,
"display_timing": False,
"solvername": solvername,
"verbose": False,
"solver_options":solver_options}
ev = xhat_eval.Xhat_Eval(xhat_eval_options,
scenario_names,
ama_object.scenario_creator,
scenario_denouement,
scenario_creator_kwargs=scenario_creator_kwargs,
all_nodenames = all_nodenames)
#Evaluating xhat and xstar and getting the value of the objective function
#for every (local) scenario
ev.evaluate(xhats)
objs_at_xhat = ev.objs_dict
ev.evaluate(xstars)
objs_at_xstar = ev.objs_dict
eval_scen_at_xhat = []
eval_scen_at_xstar = []
scen_probs = []
for k,s in ev.local_scenarios.items():
eval_scen_at_xhat.append(objs_at_xhat[k])
eval_scen_at_xstar.append(objs_at_xstar[k])
scen_probs.append(s._mpisppy_probability)
scen_gaps = np.array(eval_scen_at_xhat)-np.array(eval_scen_at_xstar)
local_gap = np.dot(scen_gaps,scen_probs)
local_ssq = np.dot(scen_gaps**2,scen_probs)
local_prob_sqnorm = np.linalg.norm(scen_probs)**2
local_obj_at_xhat = np.dot(eval_scen_at_xhat,scen_probs)
local_estim = np.array([local_gap,local_ssq,local_prob_sqnorm,local_obj_at_xhat])
global_estim = np.zeros(4)
ev.mpicomm.Allreduce(local_estim, global_estim, op=mpi.SUM)
G,ssq, prob_sqnorm,obj_at_xhat = global_estim
if global_rank==0 and verbose:
print(f"G = {G}")
sample_var = (ssq - G**2)/(1-prob_sqnorm) #Unbiased sample variance
s = np.sqrt(sample_var)
use_relative_error = (np.abs(zstar)>1)
G = correcting_numeric(G,objfct=obj_at_xhat,
relative_error=use_relative_error)
if objective_gap:
if is_multi:
return {"G":G,"s":s,"zhats": [obj_at_xhat], "seed":start}
else:
return {"G":G,"s":s,"zhats": eval_scen_at_xhat, "seed":start}
else:
return {"G":G,"s":s,"seed":start}