Skip to content

Commit

Permalink
fixes to latest version
Browse files Browse the repository at this point in the history
  • Loading branch information
@avdudchenko
avdudchenko committed Sep 27, 2024
1 parent ff6c4c8 commit 770f188
Showing 1 changed file with 119 additions and 138 deletions.
257 changes: 119 additions & 138 deletions backend/app/internal/parameter_sweep.py
View file
Original file line number Diff line number Diff line change
@@ -1,11 +1,13 @@
from pathlib import Path
from fastapi import HTTPException
import numpy as np
from parameter_sweep import LinearSample, ParameterSweep, parameter_sweep
from parameter_sweep import (
ParameterSweep,
ParameterSweepReader,
)
from importlib import import_module
import idaes.logger as idaeslog
from pyomo.environ import (
ConcreteModel,
value,
Var,
units as pyunits,
Expand Down Expand Up @@ -62,21 +64,22 @@ def run_analysis(
solve_function = flowsheet.optimize
except:
solve_function = flowsheet.solve

m = flowsheet.fs_exp.m
solve_function = flowsheet.get_action("solve")
# TODO: this does not return an actual solve function from the flowsheet...
# solve_function_action= ui_config.get_action("solve")
# print(solve_function_action) # This does not return a solve, but a add_action object... not sure why
outputs, optimize_kwargs, opt_function = set_up_sensitivity(
solve_function, output_params
)

number_of_subprocess = ui_config.get_number_of_subprocesses()
print(number_of_subprocess)
assert False
try:
build_function = flowsheet_ui.build_flowsheet
build_kwargs = {"build_options": ui_config.fs_exp.build_options}
except:
build_function = flowsheet.build
build_kwargs = {}
sweep_params = {}
# sensitivity analysis
i = 0
for each in parameters:
sweep_params[each["param"]] = {
Expand All @@ -91,41 +94,33 @@ def run_analysis(
for f_param in fixed_parameters:
if f_param["fixed"]:
sweep_params[f_param["param"] + "_fixed_state"] = {
"type": "LinearSample",
"type": "PredeterminedFixedSample",
"param": f_param["param"],
"lower_limit": f_param["value"],
"upper_limit": f_param["value"],
"num_samples": 1,
"array": np.array([f_param["value"]]),
}
else:
sweep_params[f_param["param"] + "_fixed_state"] = {
"type": "LinearSample",
"type": "PredeterminedFixedSample",
"param": f_param["param"],
"lower_limit": 0,
"upper_limit": 0,
"num_samples": 1,
"array": np.array([0]),
"set_mode": "set_fixed_state",
"default_value": f_param["value"],
}
if f_param["lb"] is not None:
sweep_params[f_param["param"] + "_lb"] = {
"type": "LinearSample",
"type": "PredeterminedFixedSample",
"param": f_param["param"],
"lower_limit": f_param["lb"],
"upper_limit": f_param["lb"],
"num_samples": 1,
"array": np.array([f_param["lb"]]),
"set_mode": "set_lb",
}
if f_param["ub"] is not None:
sweep_params[f_param["param"] + "_ub"] = {
"type": "LinearSample",
"type": "PredeterminedFixedSample",
"param": f_param["param"],
"lower_limit": f_param["ub"],
"upper_limit": f_param["ub"],
"num_samples": 1,
"array": np.array([f_param["ub"]]),
"set_mode": "set_ub",
}
print("sweep_params", sweep_params)
# Check if user provided custom kwargs, if not don't use cutm swep param
# else check if user provided custom sweep function, if not use watertap default (will be merged)
if custom_do_param_sweep_kwargs is None:
Expand All @@ -136,17 +131,15 @@ def run_analysis(
)
if custom_do_param_sweep is None:
custom_do_param_sweep_kwargs = None
print(fixed_parameters)
ps = ParameterSweep(
csv_results_file_name=results_path,
optimize_function=solve_function,
# optimize_kwargs=optimize_kwargs,
interpolate_nan_outputs=False,
custom_do_param_sweep=custom_do_param_sweep,
custom_do_param_sweep_kwargs=custom_do_param_sweep_kwargs,
reinitialize_before_sweep=False,
parallel_back_end="MultiProcessing",
number_of_subprocesses=10,
number_of_subprocesses=2,
)
global_results = ps.parameter_sweep(
build_model=build_function,
Expand All @@ -161,133 +154,121 @@ def run_analysis(


def run_parameter_sweep(flowsheet, info):
try:
_log.info("trying to sweep")
parameters = []
# for keeping track of user defined paramters in flowsheet that are changed
fixed_parameters = []
fixed_parameters_keys = []
output_params = []
# for keeping track of key to export, will be a list in same order a keys, and conversion factors
# but defines sweep_params and outputs.
export_keys = []
keys = []
conversion_factors = []
results_table = {"headers": []}
for key in flowsheet.fs_exp.exports:
if flowsheet.fs_exp.exports[key].is_sweep:
if (
flowsheet.fs_exp.exports[key].lb is not None
and flowsheet.fs_exp.exports[key].ub is not None
):
results_table["headers"].append(flowsheet.fs_exp.exports[key].name)
conversion_factor = get_conversion_unit(flowsheet, key)
try:
parameters.append(
{
"name": flowsheet.fs_exp.exports[key].name,
"lb": flowsheet.fs_exp.exports[key].obj.lb,
"ub": flowsheet.fs_exp.exports[key].obj.ub,
"num_samples": flowsheet.fs_exp.exports[
key
].num_samples,
"param": flowsheet.fs_exp.exports[key].obj,
}
)
except:
parameters.append(
{
"name": flowsheet.fs_exp.exports[key].name,
"lb": flowsheet.fs_exp.exports[key].obj.lb,
"ub": flowsheet.fs_exp.exports[key].obj.ub,
"num_samples": "5",
"param": flowsheet.fs_exp.exports[key].obj,
}
)
export_keys.append(
["sweep_params", flowsheet.fs_exp.model_objects[key].obj.name]
)
# HTTPException(500, detail=f"Sweep failed: {parameters}")
flowsheet.fs_exp.exports[key].obj.fix()
conversion_factors.append(conversion_factor)
keys.append(key)
elif flowsheet.fs_exp.model_objects[key].is_input:
fixed_parameters.append(
{
"name": flowsheet.fs_exp.model_objects[key].obj.name,
"value": flowsheet.fs_exp.model_objects[key].obj.value,
"num_samples": "1",
"param": flowsheet.fs_exp.model_objects[key].obj.name,
"fixed": flowsheet.fs_exp.model_objects[key].fixed,
"lb": flowsheet.fs_exp.model_objects[key].obj.lb,
"ub": flowsheet.fs_exp.model_objects[key].obj.ub,
}
)
fixed_parameters_keys.append(flowsheet.fs_exp.model_objects[key].name)
for key in flowsheet.fs_exp.exports:
# try:
_log.info("trying to sweep")
parameters = []
# for keeping track of user defined paramters in flowsheet that are changed
fixed_parameters = []
fixed_parameters_keys = []
output_params = []
# for keeping track of key to export, will be a list in same order a keys, and conversion factors
# but defines sweep_params and outputs.
export_keys = []
keys = []
conversion_factors = {}
results_table = {"headers": []}
for key in flowsheet.fs_exp.exports:
if flowsheet.fs_exp.exports[key].is_sweep:
if (
flowsheet.fs_exp.exports[key].is_output
or (
not flowsheet.fs_exp.exports[key].is_output
and flowsheet.fs_exp.exports[key].is_input
and not flowsheet.fs_exp.exports[key].fixed
)
# and not flowsheet.fs_exp.exports[key].is_input
flowsheet.fs_exp.exports[key].lb is not None
and flowsheet.fs_exp.exports[key].ub is not None
):
results_table["headers"].append(flowsheet.fs_exp.exports[key].name)

conversion_factor = get_conversion_unit(flowsheet, key)
try:
conversion_factor = get_conversion_unit(flowsheet, key)
except Exception as e:
conversion_factor = 1
conversion_factors.append(conversion_factor)
output_params.append(
{
"name": flowsheet.fs_exp.exports[key].name,
"param": flowsheet.fs_exp.exports[key].obj,
}
)
parameters.append(
{
"name": flowsheet.fs_exp.exports[key].name,
"lb": flowsheet.fs_exp.exports[key].obj.lb,
"ub": flowsheet.fs_exp.exports[key].obj.ub,
"num_samples": flowsheet.fs_exp.exports[key].num_samples,
"param": flowsheet.fs_exp.exports[key].obj.name,
}
)
except:
parameters.append(
{
"name": flowsheet.fs_exp.exports[key].name,
"lb": flowsheet.fs_exp.exports[key].obj.lb,
"ub": flowsheet.fs_exp.exports[key].obj.ub,
"num_samples": "5",
"param": flowsheet.fs_exp.exports[key].obj.name,
}
)
export_keys.append(
["outputs", flowsheet.fs_exp.model_objects[key].obj.name]
["sweep_params", flowsheet.fs_exp.exports[key].obj.name]
)
flowsheet.fs_exp.exports[key].obj.fix()
conversion_factors[key] = conversion_factor
keys.append(key)
output_path = Path.home() / ".nawi" / "sweep_outputs" / f"{info.name}_sweep.csv"
results_arr, output_dict = run_analysis(
ui_config=flowsheet,
flowsheet=info.module,
# # flowsheet=info.module[0:-3], # replace _ui instead?
# flowsheet=info.module.replace("_ui", ""),
parameters=parameters,
output_params=output_params,
results_path=output_path,
fixed_parameters=fixed_parameters,
# custom_do_param_sweep_kwargs=flowsheet.custom_do_param_sweep_kwargs,
)
# print(results)
except Exception as err:
_log.error(f"err: {err}")
raise HTTPException(500, detail=f"Sweep failed: {err}")
elif flowsheet.fs_exp.exports[key].is_input:
fixed_parameters.append(
{
"name": flowsheet.fs_exp.exports[key].obj.name,
"value": flowsheet.fs_exp.exports[key].obj.value,
"num_samples": "1",
"param": flowsheet.fs_exp.exports[key].obj.name,
"fixed": flowsheet.fs_exp.exports[key].fixed,
"lb": flowsheet.fs_exp.exports[key].obj.lb,
"ub": flowsheet.fs_exp.exports[key].obj.ub,
}
)
fixed_parameters_keys.append(flowsheet.fs_exp.exports[key].name)
for key in flowsheet.fs_exp.exports:
if (
flowsheet.fs_exp.exports[key].is_output
or (
not flowsheet.fs_exp.exports[key].is_output
and flowsheet.fs_exp.exports[key].is_input
and not flowsheet.fs_exp.exports[key].fixed
)
# and not flowsheet.fs_exp.exports[key].is_input
):
results_table["headers"].append(flowsheet.fs_exp.exports[key].name)

try:
conversion_factor = get_conversion_unit(flowsheet, key)
except Exception as e:
conversion_factor = 1
conversion_factors[key] = conversion_factor
# we will use obj names through out
output_params.append(
{
"name": flowsheet.fs_exp.exports[key].obj.name,
"param": flowsheet.fs_exp.exports[key].obj.name,
}
)
export_keys.append(["outputs", flowsheet.fs_exp.exports[key].obj.name])
keys.append(key)
output_path = Path.home() / ".nawi" / "sweep_outputs" / f"{info.name}_sweep.csv"
results_arr, output_dict = run_analysis(
ui_config=flowsheet,
flowsheet=info.module,
parameters=parameters,
output_params=output_params,
results_path=output_path,
fixed_parameters=fixed_parameters,
)
# except Exception as err:
# _log.error(f"err: {err}")
# raise HTTPException(500, detail=f"Sweep failed: {err}")
# we can't rely on result_arr order, as it will be async and out of order,
# so instead here, we will rebuild expected order from out output_dict
num_samples = output_dict[export_keys[0][0]][export_keys[0][1]]["value"].size
result_arr = []
# for ns in range(num_samples):
# result_arr.append([])
# for i, (result_type, key) in enumerate(export_keys):
# value = output_dict[export_keys[i][0]][export_keys[i][1]]["value"][ns]
for ns in range(num_samples):
result_arr.append([])
for i, (result_type, key) in enumerate(export_keys):
value = output_dict[export_keys[i][0]][export_keys[i][1]]["value"][ns]

# if np.isnan(value) or output_dict["solve_successful"][ns] == False:
# value = None
results_table["values"] = results[0].tolist()
for value in results_table["values"]:
for i in range(len(value)):
if np.isnan(value[i]):
value[i] = None
if np.isnan(value) or output_dict["solve_successful"][ns] == False:
value = None
else:
conversion_factor = conversion_factors[i]
conversion_factor = conversion_factors[key]
value = value * conversion_factor
result_arr[-1].append(value)
results_table["values"] = result_arr
results_table["keys"] = keys
results_table["num_parameters"] = len(parameters)
results_table["num_outputs"] = len(output_params)
print(results_table)
return results_table

0 comments on commit 770f188

Please sign in to comment.