simoptlib 1.1.0__tar.gz → 1.1.1__tar.gz
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- {simoptlib-1.1.0/simoptlib.egg-info → simoptlib-1.1.1}/PKG-INFO +2 -2
- {simoptlib-1.1.0 → simoptlib-1.1.1}/README.md +1 -1
- {simoptlib-1.1.0 → simoptlib-1.1.1}/pyproject.toml +1 -1
- {simoptlib-1.1.0 → simoptlib-1.1.1}/simopt/experiment_base.py +14 -14
- {simoptlib-1.1.0 → simoptlib-1.1.1}/simopt/gui/data_farming_window.py +20 -6
- {simoptlib-1.1.0 → simoptlib-1.1.1}/simopt/gui/df_object.py +57 -2
- {simoptlib-1.1.0 → simoptlib-1.1.1}/simopt/gui/new_experiment_window.py +189 -67
- {simoptlib-1.1.0 → simoptlib-1.1.1}/simopt/models/amusementpark.py +2 -0
- {simoptlib-1.1.0 → simoptlib-1.1.1}/simopt/models/chessmm.py +1 -0
- {simoptlib-1.1.0 → simoptlib-1.1.1}/simopt/models/cntnv.py +1 -0
- {simoptlib-1.1.0 → simoptlib-1.1.1}/simopt/models/contam.py +1 -0
- {simoptlib-1.1.0 → simoptlib-1.1.1}/simopt/models/dualsourcing.py +2 -0
- {simoptlib-1.1.0 → simoptlib-1.1.1}/simopt/models/dynamnews.py +1 -0
- {simoptlib-1.1.0 → simoptlib-1.1.1}/simopt/models/example.py +1 -0
- {simoptlib-1.1.0 → simoptlib-1.1.1}/simopt/models/facilitysizing.py +2 -0
- {simoptlib-1.1.0 → simoptlib-1.1.1}/simopt/models/fixedsan.py +1 -0
- {simoptlib-1.1.0 → simoptlib-1.1.1}/simopt/models/hotel.py +1 -0
- {simoptlib-1.1.0 → simoptlib-1.1.1}/simopt/models/ironore.py +2 -0
- {simoptlib-1.1.0 → simoptlib-1.1.1}/simopt/models/mm1queue.py +1 -0
- {simoptlib-1.1.0 → simoptlib-1.1.1}/simopt/models/network.py +1 -0
- {simoptlib-1.1.0 → simoptlib-1.1.1}/simopt/models/paramesti.py +1 -0
- {simoptlib-1.1.0 → simoptlib-1.1.1}/simopt/models/rmitd.py +1 -0
- {simoptlib-1.1.0 → simoptlib-1.1.1}/simopt/models/san.py +2 -0
- {simoptlib-1.1.0 → simoptlib-1.1.1}/simopt/models/sscont.py +3 -0
- {simoptlib-1.1.0 → simoptlib-1.1.1}/simopt/models/tableallocation.py +1 -0
- {simoptlib-1.1.0 → simoptlib-1.1.1}/simopt/solvers/astrodf.py +52 -43
- {simoptlib-1.1.0 → simoptlib-1.1.1/simoptlib.egg-info}/PKG-INFO +2 -2
- {simoptlib-1.1.0 → simoptlib-1.1.1}/LICENSE +0 -0
- {simoptlib-1.1.0 → simoptlib-1.1.1}/setup.cfg +0 -0
- {simoptlib-1.1.0 → simoptlib-1.1.1}/simopt/GUI.py +0 -0
- {simoptlib-1.1.0 → simoptlib-1.1.1}/simopt/__init__.py +0 -0
- {simoptlib-1.1.0 → simoptlib-1.1.1}/simopt/__main__.py +0 -0
- {simoptlib-1.1.0 → simoptlib-1.1.1}/simopt/base.py +0 -0
- {simoptlib-1.1.0 → simoptlib-1.1.1}/simopt/data_farming_base.py +0 -0
- {simoptlib-1.1.0 → simoptlib-1.1.1}/simopt/directory.py +0 -0
- {simoptlib-1.1.0 → simoptlib-1.1.1}/simopt/gui/experiment_window.py +0 -0
- {simoptlib-1.1.0 → simoptlib-1.1.1}/simopt/gui/main_menu.py +0 -0
- {simoptlib-1.1.0 → simoptlib-1.1.1}/simopt/gui/plot_window.py +0 -0
- {simoptlib-1.1.0 → simoptlib-1.1.1}/simopt/gui/toplevel_custom.py +0 -0
- {simoptlib-1.1.0 → simoptlib-1.1.1}/simopt/models/__init__.py +0 -0
- {simoptlib-1.1.0 → simoptlib-1.1.1}/simopt/solvers/__init__.py +0 -0
- {simoptlib-1.1.0 → simoptlib-1.1.1}/simopt/solvers/adam.py +0 -0
- {simoptlib-1.1.0 → simoptlib-1.1.1}/simopt/solvers/aloe.py +0 -0
- {simoptlib-1.1.0 → simoptlib-1.1.1}/simopt/solvers/neldmd.py +0 -0
- {simoptlib-1.1.0 → simoptlib-1.1.1}/simopt/solvers/randomsearch.py +0 -0
- {simoptlib-1.1.0 → simoptlib-1.1.1}/simopt/solvers/spsa.py +0 -0
- {simoptlib-1.1.0 → simoptlib-1.1.1}/simopt/solvers/strong.py +0 -0
- {simoptlib-1.1.0 → simoptlib-1.1.1}/simoptlib.egg-info/SOURCES.txt +0 -0
- {simoptlib-1.1.0 → simoptlib-1.1.1}/simoptlib.egg-info/dependency_links.txt +0 -0
- {simoptlib-1.1.0 → simoptlib-1.1.1}/simoptlib.egg-info/requires.txt +0 -0
- {simoptlib-1.1.0 → simoptlib-1.1.1}/simoptlib.egg-info/top_level.txt +0 -0
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
Metadata-Version: 2.1
|
|
2
2
|
Name: simoptlib
|
|
3
|
-
Version: 1.1.
|
|
3
|
+
Version: 1.1.1
|
|
4
4
|
Summary: A testbed for simulation-optimization experiments.
|
|
5
5
|
Author-email: David Eckman <eckman@tamu.edu>, Shane Henderson <sgh9@cornell.edu>, Sara Shashaani <sshasha2@ncsu.edu>, William Grochocinski <wagrocho@ncsu.edu>
|
|
6
6
|
License: MIT License
|
|
@@ -76,7 +76,7 @@ Full documentation for the source code can be found on our **[readthedocs page](
|
|
|
76
76
|
- Python >= 3.8
|
|
77
77
|
- To check your Python version, open a terminal and run `python --version`. If you see a message along the lines of `Command not found`, then you likely don't have Python installed. If you know you have it installed but are getting a `Command not found` error, then you may need to [add Python to your PATH](https://realpython.com/add-python-to-path/).
|
|
78
78
|
- For new installs, [Miniconda or Anaconda](https://www.anaconda.com/download) is recommended ([read about the differences between Miniconda and Anaconda](https://docs.anaconda.com/distro-or-miniconda/)). If you already have a compatible IDE (such as VS Code), we've found that Miniconda will work fine at 1/10 of the size of Anaconda. It is ***highly recommended*** to check the box during installation to add Python/Miniconda/Anaconda to your system PATH.
|
|
79
|
-
- Ruby >=
|
|
79
|
+
- Ruby >= 2.5 (required for datafarming)
|
|
80
80
|
- Included on MacOS, but Windows users will need to grab it from [here](https://rubyinstaller.org/).
|
|
81
81
|
- `datafarming` gem < 2.0 (required for datafarming)
|
|
82
82
|
- This can be installed via `gem install datafarming -v 1.4` once Ruby is installed/configured.
|
|
@@ -34,7 +34,7 @@ Full documentation for the source code can be found on our **[readthedocs page](
|
|
|
34
34
|
- Python >= 3.8
|
|
35
35
|
- To check your Python version, open a terminal and run `python --version`. If you see a message along the lines of `Command not found`, then you likely don't have Python installed. If you know you have it installed but are getting a `Command not found` error, then you may need to [add Python to your PATH](https://realpython.com/add-python-to-path/).
|
|
36
36
|
- For new installs, [Miniconda or Anaconda](https://www.anaconda.com/download) is recommended ([read about the differences between Miniconda and Anaconda](https://docs.anaconda.com/distro-or-miniconda/)). If you already have a compatible IDE (such as VS Code), we've found that Miniconda will work fine at 1/10 of the size of Anaconda. It is ***highly recommended*** to check the box during installation to add Python/Miniconda/Anaconda to your system PATH.
|
|
37
|
-
- Ruby >=
|
|
37
|
+
- Ruby >= 2.5 (required for datafarming)
|
|
38
38
|
- Included on MacOS, but Windows users will need to grab it from [here](https://rubyinstaller.org/).
|
|
39
39
|
- `datafarming` gem < 2.0 (required for datafarming)
|
|
40
40
|
- This can be installed via `gem install datafarming -v 1.4` once Ruby is installed/configured.
|
|
@@ -7,7 +7,7 @@ packages = ["simopt", "simopt.gui", "simopt.models", "simopt.solvers"]
|
|
|
7
7
|
|
|
8
8
|
[project]
|
|
9
9
|
name = "simoptlib"
|
|
10
|
-
version = "1.1.
|
|
10
|
+
version = "1.1.1"
|
|
11
11
|
authors = [
|
|
12
12
|
{ name="David Eckman", email="eckman@tamu.edu" },
|
|
13
13
|
{ name="Shane Henderson", email="sgh9@cornell.edu" },
|
|
@@ -318,7 +318,7 @@ def mean_of_curves(curves: list[Curve]) -> Curve:
|
|
|
318
318
|
float(np.mean([curve.lookup(float(x_val)) for curve in curves]))
|
|
319
319
|
for x_val in unique_x_vals
|
|
320
320
|
]
|
|
321
|
-
mean_curve = Curve(x_vals=unique_x_vals.tolist(), y_vals=mean_y_vals)
|
|
321
|
+
mean_curve = Curve(x_vals=unique_x_vals.tolist(), y_vals=mean_y_vals) # type: ignore
|
|
322
322
|
return mean_curve
|
|
323
323
|
|
|
324
324
|
|
|
@@ -362,7 +362,8 @@ def quantile_of_curves(curves: list[Curve], beta: float) -> Curve:
|
|
|
362
362
|
for x_val in unique_x_vals
|
|
363
363
|
]
|
|
364
364
|
quantile_curve = Curve(
|
|
365
|
-
x_vals=unique_x_vals.tolist(),
|
|
365
|
+
x_vals=unique_x_vals.tolist(), # type: ignore
|
|
366
|
+
y_vals=quantile_y_vals,
|
|
366
367
|
)
|
|
367
368
|
return quantile_curve
|
|
368
369
|
|
|
@@ -1105,7 +1106,7 @@ class ProblemSolver:
|
|
|
1105
1106
|
self.solver.attach_rngs(rng_list)
|
|
1106
1107
|
|
|
1107
1108
|
# Start a timer
|
|
1108
|
-
|
|
1109
|
+
function_start = time.time()
|
|
1109
1110
|
|
|
1110
1111
|
print("Starting macroreplications in parallel")
|
|
1111
1112
|
with Pool() as process_pool:
|
|
@@ -1118,6 +1119,10 @@ class ProblemSolver:
|
|
|
1118
1119
|
# Update status bar here
|
|
1119
1120
|
result.wait(1)
|
|
1120
1121
|
|
|
1122
|
+
print(
|
|
1123
|
+
f"Finished running {n_macroreps} macroreplications in {round(time.time() - function_start, 3)} seconds."
|
|
1124
|
+
)
|
|
1125
|
+
|
|
1121
1126
|
# Grab all the data out of the result
|
|
1122
1127
|
for mrep in range(n_macroreps):
|
|
1123
1128
|
(
|
|
@@ -1125,12 +1130,6 @@ class ProblemSolver:
|
|
|
1125
1130
|
self.all_intermediate_budgets[mrep],
|
|
1126
1131
|
self.timings[mrep],
|
|
1127
1132
|
) = result.get()[mrep]
|
|
1128
|
-
print(
|
|
1129
|
-
f"Finished running {n_macroreps} macroreplications in {round(time.time() - self.function_start, 3)} seconds."
|
|
1130
|
-
)
|
|
1131
|
-
|
|
1132
|
-
# Delete stuff we don't need to save
|
|
1133
|
-
del self.function_start
|
|
1134
1133
|
|
|
1135
1134
|
self.has_run = True
|
|
1136
1135
|
self.has_postreplicated = False
|
|
@@ -6519,7 +6518,7 @@ def validate_ruby_install() -> bool:
|
|
|
6519
6518
|
|
|
6520
6519
|
# If we're here, Ruby is not installed on the system or on WSL
|
|
6521
6520
|
# Raise an exception
|
|
6522
|
-
error_msg = "Ruby is not installed on the system or is not on the system path. Please install Ruby or add it to the system path."
|
|
6521
|
+
error_msg = "Ruby is not installed on the system or is not on the system path. Please install Ruby or add it to the system path. If you just installed Ruby, you may need to restart your terminal/IDE."
|
|
6523
6522
|
raise Exception(error_msg)
|
|
6524
6523
|
|
|
6525
6524
|
|
|
@@ -6595,10 +6594,11 @@ def validate_gem_install(design_type: str, installed_via_wsl: bool) -> str:
|
|
|
6595
6594
|
)
|
|
6596
6595
|
error_msg += f" This can be done by running: `{command_prefix}gem install datafarming -v 1.4{command_suffix}`"
|
|
6597
6596
|
raise Exception(error_msg)
|
|
6598
|
-
#
|
|
6599
|
-
#
|
|
6600
|
-
#
|
|
6601
|
-
|
|
6597
|
+
# We get here if the gem is installed and the version is correct, but
|
|
6598
|
+
# we still can't run the stack script. This is likely due to the gem
|
|
6599
|
+
# not being in the system path. We'll let the user know that they need
|
|
6600
|
+
# to restart their terminal/IDE.
|
|
6601
|
+
error_msg = "Ruby was able to detect the datafarming gem, but was unable to run the stack script. If you just installed the datafarming gem, it may be necessary to restart your terminal/IDE."
|
|
6602
6602
|
raise Exception(error_msg)
|
|
6603
6603
|
else:
|
|
6604
6604
|
error_msg = "Invalid design type."
|
|
@@ -684,7 +684,10 @@ class DataFarmingWindow(Toplevel):
|
|
|
684
684
|
].get("description")
|
|
685
685
|
self.factor_default = self.model_object.specifications[factor].get(
|
|
686
686
|
"default"
|
|
687
|
+
|
|
687
688
|
)
|
|
689
|
+
self.factor_isDatafarmable = self.model_object.specifications[factor].get(
|
|
690
|
+
"isDatafarmable")
|
|
688
691
|
|
|
689
692
|
# Values to help with formatting
|
|
690
693
|
entry_width = 10
|
|
@@ -707,7 +710,7 @@ class DataFarmingWindow(Toplevel):
|
|
|
707
710
|
padx=10,
|
|
708
711
|
)
|
|
709
712
|
|
|
710
|
-
if self.factor_datatype is float:
|
|
713
|
+
if self.factor_datatype is float and self.factor_isDatafarmable is not False:
|
|
711
714
|
self.factors_frame.grid_rowconfigure(
|
|
712
715
|
self.factor_que_length, weight=1
|
|
713
716
|
)
|
|
@@ -842,7 +845,7 @@ class DataFarmingWindow(Toplevel):
|
|
|
842
845
|
|
|
843
846
|
self.factor_que_length += 1
|
|
844
847
|
|
|
845
|
-
elif self.factor_datatype is int:
|
|
848
|
+
elif self.factor_datatype is int and self.factor_isDatafarmable is not False:
|
|
846
849
|
self.factors_frame.grid_rowconfigure(
|
|
847
850
|
self.factor_que_length, weight=1
|
|
848
851
|
)
|
|
@@ -947,12 +950,17 @@ class DataFarmingWindow(Toplevel):
|
|
|
947
950
|
|
|
948
951
|
self.factor_que_length += 1
|
|
949
952
|
|
|
950
|
-
elif self.factor_datatype is list:
|
|
953
|
+
elif self.factor_datatype is list or self.factor_isDatafarmable is False:
|
|
951
954
|
self.factors_frame.grid_rowconfigure(
|
|
952
955
|
self.factor_que_length, weight=1
|
|
953
956
|
)
|
|
954
957
|
|
|
955
|
-
self.
|
|
958
|
+
if self.factor_datatype is list:
|
|
959
|
+
self.str_type = "list"
|
|
960
|
+
elif self.factor_datatype is int:
|
|
961
|
+
self.str_type = "int"
|
|
962
|
+
elif self.factor_datatype is float:
|
|
963
|
+
self.str_type = "float"
|
|
956
964
|
|
|
957
965
|
# Add label for factor type
|
|
958
966
|
self.factortype_label = tk.Label(
|
|
@@ -1211,6 +1219,9 @@ class DataFarmingWindow(Toplevel):
|
|
|
1211
1219
|
factor_datatype = self.model_object.specifications[factor].get(
|
|
1212
1220
|
"datatype"
|
|
1213
1221
|
)
|
|
1222
|
+
factor_isDatafarmable = self.model_object.specifications[factor].get(
|
|
1223
|
+
"isDatafarmable"
|
|
1224
|
+
)
|
|
1214
1225
|
factor_include = check_values[factor_index]
|
|
1215
1226
|
|
|
1216
1227
|
# get user inputs for design factors
|
|
@@ -1218,7 +1229,7 @@ class DataFarmingWindow(Toplevel):
|
|
|
1218
1229
|
if factor_include:
|
|
1219
1230
|
self.factor_names.append(factor)
|
|
1220
1231
|
|
|
1221
|
-
if factor_datatype in (float, int):
|
|
1232
|
+
if factor_datatype in (float, int) and factor_isDatafarmable is not False:
|
|
1222
1233
|
factor_min = str(min_values[maxmin_index])
|
|
1223
1234
|
factor_max = str(max_values[maxmin_index])
|
|
1224
1235
|
maxmin_index += 1
|
|
@@ -1352,9 +1363,12 @@ class DataFarmingWindow(Toplevel):
|
|
|
1352
1363
|
self.factor_default = self.model_object.specifications[factor].get(
|
|
1353
1364
|
"default"
|
|
1354
1365
|
)
|
|
1366
|
+
self.factor_isDatafarmable = self.model_object.specifications[factor].get(
|
|
1367
|
+
"isDatafarmable"
|
|
1368
|
+
)
|
|
1355
1369
|
|
|
1356
1370
|
# Disable / enable experiment option widgets depending on factor type
|
|
1357
|
-
if self.factor_datatype in (int, float):
|
|
1371
|
+
if self.factor_datatype in (int, float) and self.factor_isDatafarmable is not False:
|
|
1358
1372
|
self.current_min_entry = self.min_widgets[factor]
|
|
1359
1373
|
self.current_max_entry = self.max_widgets[factor]
|
|
1360
1374
|
|
|
@@ -491,6 +491,54 @@ class DFInteger(DFFactor):
|
|
|
491
491
|
self.__maximum = tk.IntVar(value=default)
|
|
492
492
|
|
|
493
493
|
|
|
494
|
+
class DFIntegerNonDatafarmable(DFFactor):
|
|
495
|
+
"""Class to store non-datafarmable integer factors for problems and solvers."""
|
|
496
|
+
|
|
497
|
+
@property
|
|
498
|
+
def type(self) -> tk.StringVar:
|
|
499
|
+
"""The type of the factor."""
|
|
500
|
+
return tk.StringVar(value="int")
|
|
501
|
+
|
|
502
|
+
@property
|
|
503
|
+
def default(self) -> tk.IntVar:
|
|
504
|
+
"""The default value of the factor."""
|
|
505
|
+
return self.__default
|
|
506
|
+
|
|
507
|
+
@default.setter
|
|
508
|
+
def default(self, default: tk.Variable) -> None:
|
|
509
|
+
if not isinstance(default, tk.IntVar):
|
|
510
|
+
error_msg = "Default value must be an IntVar."
|
|
511
|
+
raise ValueError(error_msg)
|
|
512
|
+
self.__default = default
|
|
513
|
+
|
|
514
|
+
@property
|
|
515
|
+
def default_eval(self) -> int:
|
|
516
|
+
"""Evaluated default value of the factor."""
|
|
517
|
+
try:
|
|
518
|
+
return int(self.default.get())
|
|
519
|
+
except ValueError:
|
|
520
|
+
raise ValueError(
|
|
521
|
+
f"Default value for {self.name.get()} must be an integer."
|
|
522
|
+
) from None
|
|
523
|
+
|
|
524
|
+
def __init__(self, name: str, description: str, default: int) -> None:
|
|
525
|
+
"""Initialize the
|
|
526
|
+
|
|
527
|
+
Parameters
|
|
528
|
+
----------
|
|
529
|
+
name : str
|
|
530
|
+
The name of the factor
|
|
531
|
+
description : str
|
|
532
|
+
The description of the factor
|
|
533
|
+
default : list
|
|
534
|
+
The default value of the factor
|
|
535
|
+
|
|
536
|
+
"""
|
|
537
|
+
appended_description = "[Non-Datafarmable] " + description
|
|
538
|
+
super().__init__(name, appended_description)
|
|
539
|
+
self.__default = tk.IntVar(value=default)
|
|
540
|
+
|
|
541
|
+
|
|
494
542
|
class DFFloat(DFFactor):
|
|
495
543
|
"""Class to store float factors for problems and solvers."""
|
|
496
544
|
|
|
@@ -748,8 +796,15 @@ def spec_to_df(spec_name: str, spec: dict) -> DFFactor:
|
|
|
748
796
|
list: DFList,
|
|
749
797
|
}
|
|
750
798
|
|
|
751
|
-
#
|
|
752
|
-
if
|
|
799
|
+
# Check to see if we have a non-datafarmable integer
|
|
800
|
+
if (
|
|
801
|
+
f_type is int
|
|
802
|
+
and "isDatafarmable" in spec
|
|
803
|
+
and not spec["isDatafarmable"]
|
|
804
|
+
):
|
|
805
|
+
return DFIntegerNonDatafarmable(spec_name, f_description, f_default)
|
|
806
|
+
# Otherwise, just use the default mapping
|
|
807
|
+
elif f_type in df_factor_map:
|
|
753
808
|
return df_factor_map[f_type](spec_name, f_description, f_default)
|
|
754
809
|
else:
|
|
755
810
|
raise NotImplementedError(
|
|
@@ -1,6 +1,7 @@
|
|
|
1
1
|
import os
|
|
2
2
|
import pickle
|
|
3
3
|
import re
|
|
4
|
+
import threading
|
|
4
5
|
import tkinter as tk
|
|
5
6
|
from abc import ABCMeta
|
|
6
7
|
from tkinter import filedialog, messagebox, ttk
|
|
@@ -1461,11 +1462,13 @@ class NewExperimentWindow(Toplevel):
|
|
|
1461
1462
|
# Open file dialog to select design file
|
|
1462
1463
|
# Pickle files only, but all files can be selected (in case someone
|
|
1463
1464
|
# forgets to change file type)
|
|
1464
|
-
# NOTE:
|
|
1465
|
-
#
|
|
1465
|
+
# NOTE: Trying to accept both .pickle and .pkl files using
|
|
1466
|
+
# "*.pickle;*.pkl" causes Python to crash on MacOS but works fine on
|
|
1467
|
+
# Windows. As long as we only have one pickle file extension, we
|
|
1468
|
+
# should be fine.
|
|
1466
1469
|
# TODO: standardize Pickle file extension (see GitHub issue #71)
|
|
1467
1470
|
experiment_file = filedialog.askopenfilename(
|
|
1468
|
-
filetypes=[("Pickle files", "*.pickle
|
|
1471
|
+
filetypes=[("Pickle files", "*.pickle"), ("All files", "*.*")]
|
|
1469
1472
|
)
|
|
1470
1473
|
# Exit w/o message if no file selected
|
|
1471
1474
|
if experiment_file == "" or not experiment_file:
|
|
@@ -2280,6 +2283,183 @@ class NewExperimentWindow(Toplevel):
|
|
|
2280
2283
|
self.__update_problem_dropdown()
|
|
2281
2284
|
self.__update_solver_dropdown()
|
|
2282
2285
|
|
|
2286
|
+
def __spawn_new_thread(self, function: Callable) -> threading.Thread:
|
|
2287
|
+
thread = threading.Thread(target=function)
|
|
2288
|
+
thread.start()
|
|
2289
|
+
return thread
|
|
2290
|
+
|
|
2291
|
+
def __disable_exp_buttons(self, experiment_name: str) -> None:
|
|
2292
|
+
name_base: Final[str] = "exp." + experiment_name
|
|
2293
|
+
# Get all the buttons
|
|
2294
|
+
buttons = [bttn for bttn in self.tk_buttons if name_base in bttn]
|
|
2295
|
+
# Disable all buttons (except view)
|
|
2296
|
+
for button in buttons:
|
|
2297
|
+
if not button.endswith(".view"):
|
|
2298
|
+
self.tk_buttons[button].configure(state="disabled")
|
|
2299
|
+
|
|
2300
|
+
def __enable_exp_buttons(self, experiment_name: str) -> None:
|
|
2301
|
+
name_base: Final[str] = "exp." + experiment_name
|
|
2302
|
+
# Get all the buttons
|
|
2303
|
+
buttons = [bttn for bttn in self.tk_buttons if name_base in bttn]
|
|
2304
|
+
# Enable all buttons
|
|
2305
|
+
for button in buttons:
|
|
2306
|
+
self.tk_buttons[button].configure(state="normal")
|
|
2307
|
+
|
|
2308
|
+
def __update_action_button(
|
|
2309
|
+
self,
|
|
2310
|
+
experiment_name: str,
|
|
2311
|
+
text: str,
|
|
2312
|
+
command: Union[Callable, None] = None,
|
|
2313
|
+
) -> None:
|
|
2314
|
+
name_base: Final[str] = "exp." + experiment_name
|
|
2315
|
+
action_bttn_name: Final[str] = name_base + ".action"
|
|
2316
|
+
if command is None:
|
|
2317
|
+
self.tk_buttons[action_bttn_name].configure(text=text)
|
|
2318
|
+
else:
|
|
2319
|
+
self.tk_buttons[action_bttn_name].configure(
|
|
2320
|
+
text=text, command=command
|
|
2321
|
+
)
|
|
2322
|
+
|
|
2323
|
+
def __update_experiment_label(
|
|
2324
|
+
self, experiment_name: str, status: str
|
|
2325
|
+
) -> None:
|
|
2326
|
+
name_base: Final[str] = "exp." + experiment_name
|
|
2327
|
+
lbl_name: Final[str] = name_base + ".name"
|
|
2328
|
+
text = f"{experiment_name}\n({status})"
|
|
2329
|
+
self.tk_labels[lbl_name].configure(text=text)
|
|
2330
|
+
|
|
2331
|
+
def __run_experiment_gui(self, experiment_name: str) -> None:
|
|
2332
|
+
# Setup
|
|
2333
|
+
self.__disable_exp_buttons(experiment_name)
|
|
2334
|
+
self.__update_experiment_label(experiment_name, "Running")
|
|
2335
|
+
# Try to run the experiment
|
|
2336
|
+
try:
|
|
2337
|
+
self.run_experiment(experiment_name)
|
|
2338
|
+
# If successful, update the label and button
|
|
2339
|
+
self.__update_experiment_label(experiment_name, "Ran")
|
|
2340
|
+
self.__update_action_button(
|
|
2341
|
+
experiment_name,
|
|
2342
|
+
"Post-Replicate",
|
|
2343
|
+
lambda exp_name=experiment_name: self.__post_process_gui_thread(
|
|
2344
|
+
exp_name
|
|
2345
|
+
),
|
|
2346
|
+
)
|
|
2347
|
+
except Exception as e:
|
|
2348
|
+
messagebox.showerror("Error", str(e))
|
|
2349
|
+
self.__update_experiment_label(experiment_name, "Initialized")
|
|
2350
|
+
# Enable the buttons
|
|
2351
|
+
self.__enable_exp_buttons(experiment_name)
|
|
2352
|
+
|
|
2353
|
+
def __run_experiment_gui_thread(
|
|
2354
|
+
self, experiment_name: str
|
|
2355
|
+
) -> threading.Thread:
|
|
2356
|
+
return self.__spawn_new_thread(
|
|
2357
|
+
lambda: self.__run_experiment_gui(experiment_name)
|
|
2358
|
+
)
|
|
2359
|
+
|
|
2360
|
+
def __post_process_gui(self, experiment_name: str) -> None:
|
|
2361
|
+
# Setup
|
|
2362
|
+
self.__disable_exp_buttons(experiment_name)
|
|
2363
|
+
self.__update_experiment_label(experiment_name, "Post-Processing")
|
|
2364
|
+
# Try to run the post-processing
|
|
2365
|
+
try:
|
|
2366
|
+
self.post_process(experiment_name)
|
|
2367
|
+
# If successful, update the label and button
|
|
2368
|
+
self.__update_experiment_label(experiment_name, "Post-Processed")
|
|
2369
|
+
self.__update_action_button(
|
|
2370
|
+
experiment_name,
|
|
2371
|
+
"Post-Replicated",
|
|
2372
|
+
lambda exp_name=experiment_name: self.__post_normalize_gui_thread(
|
|
2373
|
+
exp_name
|
|
2374
|
+
),
|
|
2375
|
+
)
|
|
2376
|
+
except Exception as e:
|
|
2377
|
+
messagebox.showerror("Error", str(e))
|
|
2378
|
+
self.__update_experiment_label(experiment_name, "Ran")
|
|
2379
|
+
# Enable the buttons
|
|
2380
|
+
self.__enable_exp_buttons(experiment_name)
|
|
2381
|
+
|
|
2382
|
+
def __post_process_gui_thread(
|
|
2383
|
+
self, experiment_name: str
|
|
2384
|
+
) -> threading.Thread:
|
|
2385
|
+
return self.__spawn_new_thread(
|
|
2386
|
+
lambda: self.__post_process_gui(experiment_name)
|
|
2387
|
+
)
|
|
2388
|
+
|
|
2389
|
+
def __post_normalize_gui(self, experiment_name: str) -> None:
|
|
2390
|
+
# Setup
|
|
2391
|
+
self.__disable_exp_buttons(experiment_name)
|
|
2392
|
+
self.__update_experiment_label(experiment_name, "Post-Normalizing")
|
|
2393
|
+
# Try to run the post-normalization
|
|
2394
|
+
try:
|
|
2395
|
+
self.post_normalize(experiment_name)
|
|
2396
|
+
# If successful, update the label and button
|
|
2397
|
+
self.__update_experiment_label(experiment_name, "Post-Normalized")
|
|
2398
|
+
self.__update_action_button(
|
|
2399
|
+
experiment_name,
|
|
2400
|
+
"Log Results",
|
|
2401
|
+
lambda exp_name=experiment_name: self.__log_results_gui_thread(
|
|
2402
|
+
exp_name
|
|
2403
|
+
),
|
|
2404
|
+
)
|
|
2405
|
+
except Exception as e:
|
|
2406
|
+
messagebox.showerror("Error", str(e))
|
|
2407
|
+
# If the post-normalization fails, revert to the post-process button
|
|
2408
|
+
self.__update_experiment_label(experiment_name, "Post-Processed")
|
|
2409
|
+
# Enable the buttons
|
|
2410
|
+
self.__enable_exp_buttons(experiment_name)
|
|
2411
|
+
|
|
2412
|
+
def __post_normalize_gui_thread(
|
|
2413
|
+
self, experiment_name: str
|
|
2414
|
+
) -> threading.Thread:
|
|
2415
|
+
return self.__spawn_new_thread(
|
|
2416
|
+
lambda: self.__post_normalize_gui(experiment_name)
|
|
2417
|
+
)
|
|
2418
|
+
|
|
2419
|
+
def __log_results_gui(self, experiment_name: str) -> None:
|
|
2420
|
+
# Setup
|
|
2421
|
+
self.__disable_exp_buttons(experiment_name)
|
|
2422
|
+
self.__update_experiment_label(experiment_name, "Logging")
|
|
2423
|
+
# Try to log the experiment
|
|
2424
|
+
try:
|
|
2425
|
+
self.post_normalize(experiment_name)
|
|
2426
|
+
# If successful, update the label and button
|
|
2427
|
+
self.__update_experiment_label(experiment_name, "Logged")
|
|
2428
|
+
self.__update_action_button(experiment_name, "All Steps\nComplete")
|
|
2429
|
+
# Update the all button to reflect that all steps are done
|
|
2430
|
+
all_bttn_name: Final[str] = "exp." + experiment_name + ".all"
|
|
2431
|
+
self.tk_buttons[all_bttn_name].configure(text="All Steps\nComplete")
|
|
2432
|
+
except Exception as e:
|
|
2433
|
+
messagebox.showerror("Error", str(e))
|
|
2434
|
+
self.__update_experiment_label(experiment_name, "Post-Normalized")
|
|
2435
|
+
# Enable the buttons that don't relate to running the experiment
|
|
2436
|
+
name_base: Final[str] = "exp." + experiment_name
|
|
2437
|
+
buttons = [bttn for bttn in self.tk_buttons if name_base in bttn]
|
|
2438
|
+
for button in buttons:
|
|
2439
|
+
if not button.endswith(".action") and not button.endswith(".all"):
|
|
2440
|
+
self.tk_buttons[button].configure(state="normal")
|
|
2441
|
+
|
|
2442
|
+
def __log_results_gui_thread(
|
|
2443
|
+
self, experiment_name: str
|
|
2444
|
+
) -> threading.Thread:
|
|
2445
|
+
return self.__spawn_new_thread(
|
|
2446
|
+
lambda: self.__log_results_gui(experiment_name)
|
|
2447
|
+
)
|
|
2448
|
+
|
|
2449
|
+
def __all_action_gui(self, experiment_name: str) -> None:
|
|
2450
|
+
# None of these steps do anything if they've already been done
|
|
2451
|
+
self.__run_experiment_gui(experiment_name)
|
|
2452
|
+
self.__post_process_gui(experiment_name)
|
|
2453
|
+
self.__post_normalize_gui(experiment_name)
|
|
2454
|
+
self.__log_results_gui(experiment_name)
|
|
2455
|
+
|
|
2456
|
+
def __all_actions_gui_thread(
|
|
2457
|
+
self, experiment_name: str
|
|
2458
|
+
) -> threading.Thread:
|
|
2459
|
+
return self.__spawn_new_thread(
|
|
2460
|
+
lambda: self.__all_action_gui(experiment_name)
|
|
2461
|
+
)
|
|
2462
|
+
|
|
2283
2463
|
def add_exp_row(
|
|
2284
2464
|
self, experiment_name: str, is_imported: bool = False
|
|
2285
2465
|
) -> None:
|
|
@@ -2289,25 +2469,6 @@ class NewExperimentWindow(Toplevel):
|
|
|
2289
2469
|
|
|
2290
2470
|
# Format:
|
|
2291
2471
|
# (past_tense, present_tense, future_tense, function)
|
|
2292
|
-
bttn_text_steps: Final[
|
|
2293
|
-
list[tuple[str, str, str, Callable[[str], None]]]
|
|
2294
|
-
] = [
|
|
2295
|
-
("Run", "Running", "Ran", self.run_experiment),
|
|
2296
|
-
(
|
|
2297
|
-
"Post-Replicate",
|
|
2298
|
-
"Post-Replicating",
|
|
2299
|
-
"Post-Replicated",
|
|
2300
|
-
self.post_process,
|
|
2301
|
-
),
|
|
2302
|
-
(
|
|
2303
|
-
"Post-Normalize",
|
|
2304
|
-
"Post-Normalizing",
|
|
2305
|
-
"Post-Normalized",
|
|
2306
|
-
self.post_normalize,
|
|
2307
|
-
),
|
|
2308
|
-
("Log Results", "Logging", "Logged", self.log_results),
|
|
2309
|
-
]
|
|
2310
|
-
|
|
2311
2472
|
name_base: Final[str] = "exp." + experiment_name
|
|
2312
2473
|
lbl_name: Final[str] = name_base + ".name"
|
|
2313
2474
|
action_bttn_name: Final[str] = name_base + ".action"
|
|
@@ -2338,49 +2499,6 @@ class NewExperimentWindow(Toplevel):
|
|
|
2338
2499
|
bttn_text_run_all = "Run All\nRemaining Steps"
|
|
2339
2500
|
bttn_text_run = "Run\nExperiment"
|
|
2340
2501
|
|
|
2341
|
-
def action(name: str, step: int = 0) -> bool:
|
|
2342
|
-
# Lookup all the verbage for the current step
|
|
2343
|
-
past_tense, present_tense, future_tense, function = bttn_text_steps[
|
|
2344
|
-
step
|
|
2345
|
-
]
|
|
2346
|
-
# Get the buttons and update them for the current step
|
|
2347
|
-
action_button = self.tk_buttons[action_bttn_name]
|
|
2348
|
-
all_button = self.tk_buttons[all_bttn_name]
|
|
2349
|
-
action_button.configure(text=present_tense, state="disabled")
|
|
2350
|
-
all_button.configure(state="disabled")
|
|
2351
|
-
self.update()
|
|
2352
|
-
# Run the function
|
|
2353
|
-
try:
|
|
2354
|
-
function(name) # type: ignore
|
|
2355
|
-
# Set the name to include the updated status
|
|
2356
|
-
name_label = self.tk_labels[lbl_name]
|
|
2357
|
-
name_label.configure(text=name + "\n(" + future_tense + ")")
|
|
2358
|
-
# If there are more steps, setup the next step
|
|
2359
|
-
if step < len(bttn_text_steps) - 1:
|
|
2360
|
-
# Figure out what the next step is
|
|
2361
|
-
next_past_tense, _, _, _ = bttn_text_steps[step + 1]
|
|
2362
|
-
action_button.configure(
|
|
2363
|
-
text=next_past_tense,
|
|
2364
|
-
state="normal",
|
|
2365
|
-
command=lambda: action(name, step + 1),
|
|
2366
|
-
)
|
|
2367
|
-
all_button.configure(state="normal")
|
|
2368
|
-
else:
|
|
2369
|
-
action_button.configure(text="Done")
|
|
2370
|
-
return True
|
|
2371
|
-
|
|
2372
|
-
except Exception as e:
|
|
2373
|
-
messagebox.showerror("Error", str(e))
|
|
2374
|
-
# Set the button text back to past tense and reactivate buttons
|
|
2375
|
-
action_button.configure(text=past_tense, state="normal")
|
|
2376
|
-
all_button.configure(state="normal")
|
|
2377
|
-
return False
|
|
2378
|
-
|
|
2379
|
-
def all_actions(name: str) -> None:
|
|
2380
|
-
for step in range(len(bttn_text_steps)):
|
|
2381
|
-
if not action(name, step):
|
|
2382
|
-
break
|
|
2383
|
-
|
|
2384
2502
|
def view(experiment_name: str) -> None:
|
|
2385
2503
|
# Check if there's an experiment in progress
|
|
2386
2504
|
if (
|
|
@@ -2472,7 +2590,9 @@ class NewExperimentWindow(Toplevel):
|
|
|
2472
2590
|
self.tk_buttons[action_bttn_name] = ttk.Button(
|
|
2473
2591
|
master=list_frame,
|
|
2474
2592
|
text=bttn_text_run,
|
|
2475
|
-
command=lambda name=experiment_name:
|
|
2593
|
+
command=lambda name=experiment_name: self.__run_experiment_gui_thread(
|
|
2594
|
+
name
|
|
2595
|
+
),
|
|
2476
2596
|
)
|
|
2477
2597
|
self.tk_buttons[action_bttn_name].grid(
|
|
2478
2598
|
row=row_idx, column=1, padx=5, pady=5, sticky="nsew"
|
|
@@ -2481,7 +2601,9 @@ class NewExperimentWindow(Toplevel):
|
|
|
2481
2601
|
self.tk_buttons[all_bttn_name] = ttk.Button(
|
|
2482
2602
|
master=list_frame,
|
|
2483
2603
|
text=bttn_text_run_all,
|
|
2484
|
-
command=lambda name=experiment_name:
|
|
2604
|
+
command=lambda name=experiment_name: self.__all_actions_gui_thread(
|
|
2605
|
+
name
|
|
2606
|
+
),
|
|
2485
2607
|
)
|
|
2486
2608
|
self.tk_buttons[all_bttn_name].grid(
|
|
2487
2609
|
row=row_idx, column=2, padx=5, pady=5, sticky="nsew"
|
|
@@ -79,6 +79,7 @@ class AmusementPark(Model):
|
|
|
79
79
|
"description": "The number of attractions in the park.",
|
|
80
80
|
"datatype": int,
|
|
81
81
|
"default": NUM_ATTRACTIONS,
|
|
82
|
+
"isDatafarmable": False
|
|
82
83
|
},
|
|
83
84
|
"time_open": {
|
|
84
85
|
"description": "The number of minutes per day the park is open.",
|
|
@@ -556,6 +557,7 @@ class AmusementParkMinDepart(Problem):
|
|
|
556
557
|
"description": "Max # of replications for a solver to take.",
|
|
557
558
|
"datatype": int,
|
|
558
559
|
"default": 100,
|
|
560
|
+
"isDatafarmable": False
|
|
559
561
|
},
|
|
560
562
|
}
|
|
561
563
|
|
|
@@ -92,6 +92,7 @@ class DualSourcing(Model):
|
|
|
92
92
|
"description": "number of days to simulate",
|
|
93
93
|
"datatype": int,
|
|
94
94
|
"default": 1000,
|
|
95
|
+
"isDatafarmable": False
|
|
95
96
|
},
|
|
96
97
|
"initial_inv": {
|
|
97
98
|
"description": "initial inventory",
|
|
@@ -467,6 +468,7 @@ class DualSourcingMinCost(Problem):
|
|
|
467
468
|
"description": "max # of replications for a solver to take",
|
|
468
469
|
"datatype": int,
|
|
469
470
|
"default": 1000,
|
|
471
|
+
"isDatafarmable": False
|
|
470
472
|
},
|
|
471
473
|
}
|
|
472
474
|
|
|
@@ -87,6 +87,7 @@ class FacilitySize(Model):
|
|
|
87
87
|
"description": "number of facilities",
|
|
88
88
|
"datatype": int,
|
|
89
89
|
"default": NUM_FACILITIES,
|
|
90
|
+
"isDatafarmable": False
|
|
90
91
|
},
|
|
91
92
|
}
|
|
92
93
|
|
|
@@ -317,6 +318,7 @@ class FacilitySizingTotalCost(Problem):
|
|
|
317
318
|
"description": "Max # of replications for a solver to take.",
|
|
318
319
|
"datatype": int,
|
|
319
320
|
"default": 10000,
|
|
321
|
+
"isDatafarmable": False
|
|
320
322
|
},
|
|
321
323
|
"installation_costs": {
|
|
322
324
|
"description": "Cost to install a unit of capacity at each facility.",
|
|
@@ -132,6 +132,7 @@ class IronOre(Model):
|
|
|
132
132
|
"description": "number of days to simulate",
|
|
133
133
|
"datatype": int,
|
|
134
134
|
"default": 365,
|
|
135
|
+
"isDatafarmable": False
|
|
135
136
|
},
|
|
136
137
|
}
|
|
137
138
|
|
|
@@ -449,6 +450,7 @@ class IronOreMaxRev(Problem):
|
|
|
449
450
|
"description": "max # of replications for a solver to take",
|
|
450
451
|
"datatype": int,
|
|
451
452
|
"default": 1000,
|
|
453
|
+
"isDatafarmable": False
|
|
452
454
|
},
|
|
453
455
|
}
|
|
454
456
|
|
|
@@ -68,6 +68,7 @@ class SAN(Model):
|
|
|
68
68
|
"description": "number of nodes",
|
|
69
69
|
"datatype": int,
|
|
70
70
|
"default": 9,
|
|
71
|
+
"isDatafarmable": False
|
|
71
72
|
},
|
|
72
73
|
"arcs": {
|
|
73
74
|
"description": "list of arcs",
|
|
@@ -373,6 +374,7 @@ class SANLongestPath(Problem):
|
|
|
373
374
|
"description": "max # of replications for a solver to take",
|
|
374
375
|
"datatype": int,
|
|
375
376
|
"default": 10000,
|
|
377
|
+
"isDatafarmable": False
|
|
376
378
|
},
|
|
377
379
|
"arc_costs": {
|
|
378
380
|
"description": "Cost associated to each arc.",
|
|
@@ -131,11 +131,13 @@ class SSCont(Model):
|
|
|
131
131
|
"description": "number of periods to simulate",
|
|
132
132
|
"datatype": int,
|
|
133
133
|
"default": 100,
|
|
134
|
+
"isDatafarmable": False
|
|
134
135
|
},
|
|
135
136
|
"warmup": {
|
|
136
137
|
"description": "number of periods as warmup before collecting statistics",
|
|
137
138
|
"datatype": int,
|
|
138
139
|
"default": 20,
|
|
140
|
+
"isDatafarmable": False
|
|
139
141
|
},
|
|
140
142
|
}
|
|
141
143
|
|
|
@@ -498,6 +500,7 @@ class SSContMinCost(Problem):
|
|
|
498
500
|
"description": "max # of replications for a solver to take",
|
|
499
501
|
"datatype": int,
|
|
500
502
|
"default": 1000,
|
|
503
|
+
"isDatafarmable": False
|
|
501
504
|
},
|
|
502
505
|
}
|
|
503
506
|
|
|
@@ -132,6 +132,11 @@ class ASTRODF(Solver):
|
|
|
132
132
|
"datatype": float,
|
|
133
133
|
"default": 0.1,
|
|
134
134
|
},
|
|
135
|
+
"use_gradients": {
|
|
136
|
+
"description": "if direct gradient observations are available, use them",
|
|
137
|
+
"datatype": bool,
|
|
138
|
+
"default": True,
|
|
139
|
+
},
|
|
135
140
|
}
|
|
136
141
|
|
|
137
142
|
@property
|
|
@@ -469,8 +474,9 @@ class ASTRODF(Solver):
|
|
|
469
474
|
interpolation_solns.append(visited_pts_list[f_index])
|
|
470
475
|
# for new points, run the simulation with pilot run
|
|
471
476
|
else:
|
|
477
|
+
decision_vars = tuple(var_y[i][0])
|
|
472
478
|
new_solution = self.create_new_solution(
|
|
473
|
-
|
|
479
|
+
decision_vars, problem
|
|
474
480
|
)
|
|
475
481
|
visited_pts_list.append(new_solution)
|
|
476
482
|
problem.simulate(new_solution, pilot_run)
|
|
@@ -479,9 +485,6 @@ class ASTRODF(Solver):
|
|
|
479
485
|
|
|
480
486
|
# adaptive sampling
|
|
481
487
|
while True:
|
|
482
|
-
problem.simulate(new_solution, 1)
|
|
483
|
-
expended_budget += 1
|
|
484
|
-
sample_size += 1
|
|
485
488
|
sig2 = new_solution.objectives_var[0]
|
|
486
489
|
stopping = self.get_stopping_time(
|
|
487
490
|
pilot_run,
|
|
@@ -496,6 +499,10 @@ class ASTRODF(Solver):
|
|
|
496
499
|
or expended_budget >= budget
|
|
497
500
|
):
|
|
498
501
|
break
|
|
502
|
+
problem.simulate(new_solution, 1)
|
|
503
|
+
expended_budget += 1
|
|
504
|
+
sample_size += 1
|
|
505
|
+
|
|
499
506
|
fval.append(
|
|
500
507
|
-1 * problem.minmax[0] * new_solution.objectives_mean
|
|
501
508
|
)
|
|
@@ -659,12 +666,14 @@ class ASTRODF(Solver):
|
|
|
659
666
|
lambda_min: int = self.factors["lambda_min"]
|
|
660
667
|
lambda_max = budget_limit - expended_budget
|
|
661
668
|
# lambda_max = budget_limit / (15 * sqrt(problem.dim))
|
|
662
|
-
|
|
669
|
+
enable_gradient = (
|
|
670
|
+
problem.gradient_available and self.factors["use_gradients"]
|
|
671
|
+
)
|
|
663
672
|
# uncomment the next line to avoid Hessian updating
|
|
664
673
|
# h_k = np.identity(problem.dim)
|
|
665
674
|
# determine power of delta in adaptive sampling rule
|
|
666
675
|
if self.factors["crn_across_solns"]:
|
|
667
|
-
if
|
|
676
|
+
if enable_gradient:
|
|
668
677
|
delta_power = 0
|
|
669
678
|
else:
|
|
670
679
|
delta_power = 2
|
|
@@ -679,9 +688,7 @@ class ASTRODF(Solver):
|
|
|
679
688
|
- 1
|
|
680
689
|
)
|
|
681
690
|
if k == 1:
|
|
682
|
-
incumbent_solution = self.create_new_solution(
|
|
683
|
-
tuple(incumbent_x), problem
|
|
684
|
-
)
|
|
691
|
+
incumbent_solution = self.create_new_solution(incumbent_x, problem)
|
|
685
692
|
if len(visited_pts_list) == 0:
|
|
686
693
|
visited_pts_list.append(incumbent_solution)
|
|
687
694
|
|
|
@@ -692,7 +699,7 @@ class ASTRODF(Solver):
|
|
|
692
699
|
|
|
693
700
|
# adaptive sampling
|
|
694
701
|
while True:
|
|
695
|
-
if
|
|
702
|
+
if enable_gradient:
|
|
696
703
|
rhs_for_kappa = norm(
|
|
697
704
|
incumbent_solution.objectives_gradients_mean[0]
|
|
698
705
|
)
|
|
@@ -726,14 +733,15 @@ class ASTRODF(Solver):
|
|
|
726
733
|
)
|
|
727
734
|
# print("kappa "+str(kappa))
|
|
728
735
|
break
|
|
729
|
-
|
|
730
|
-
|
|
731
|
-
|
|
732
|
-
sample_size += 1
|
|
736
|
+
problem.simulate(incumbent_solution, 1)
|
|
737
|
+
expended_budget += 1
|
|
738
|
+
sample_size += 1
|
|
733
739
|
|
|
734
740
|
recommended_solns.append(incumbent_solution)
|
|
735
741
|
intermediate_budgets.append(expended_budget)
|
|
736
|
-
elif self.factors[
|
|
742
|
+
elif self.factors[
|
|
743
|
+
"crn_across_solns"
|
|
744
|
+
]: # since incument was only evaluated with the sample size of previous incumbent, here we compute its adaptive sample size
|
|
737
745
|
sample_size = incumbent_solution.n_reps
|
|
738
746
|
# adaptive sampling
|
|
739
747
|
while True:
|
|
@@ -755,10 +763,9 @@ class ASTRODF(Solver):
|
|
|
755
763
|
problem.simulate(incumbent_solution, 1)
|
|
756
764
|
expended_budget += 1
|
|
757
765
|
sample_size += 1
|
|
758
|
-
|
|
759
766
|
|
|
760
767
|
# use Taylor expansion if gradient available
|
|
761
|
-
if
|
|
768
|
+
if enable_gradient:
|
|
762
769
|
fval = (
|
|
763
770
|
np.ones(2 * problem.dim + 1)
|
|
764
771
|
* -1
|
|
@@ -801,7 +808,7 @@ class ASTRODF(Solver):
|
|
|
801
808
|
# TODO: why do we need this? Check model reduction calculation too.
|
|
802
809
|
# print("np.dot(np.multiply(grad, Hessian), grad) "+str(np.dot(np.multiply(grad, hessian), grad)))
|
|
803
810
|
# print("np.dot(np.dot(grad, hessian), grad) "+str(np.dot(np.dot(grad, hessian), grad)))
|
|
804
|
-
if
|
|
811
|
+
if enable_gradient:
|
|
805
812
|
# print("hessian " + str(hessian))
|
|
806
813
|
check_positive_definite = np.dot(np.dot(grad, hessian), grad)
|
|
807
814
|
else:
|
|
@@ -817,12 +824,8 @@ class ASTRODF(Solver):
|
|
|
817
824
|
# print("tau "+str(tau))
|
|
818
825
|
grad = np.reshape(grad, (1, problem.dim))[0]
|
|
819
826
|
grad_norm = norm(grad)
|
|
827
|
+
# Make sure we don't divide by 0
|
|
820
828
|
if grad_norm == 0:
|
|
821
|
-
# warning_msg = "Warning: Division by 0 in ASTRO-DF solver (norm(grad) == 0)."
|
|
822
|
-
# print(warning_msg, file=sys.stderr)
|
|
823
|
-
# # TODO: figure out how to handle this
|
|
824
|
-
# # Normally, DB0 errors would result in +-inf, but we have to
|
|
825
|
-
# # add the result to incumbent_x, so we can't have inf values
|
|
826
829
|
candidate_x = incumbent_x
|
|
827
830
|
else:
|
|
828
831
|
product = tau * delta_k * grad
|
|
@@ -854,16 +857,18 @@ class ASTRODF(Solver):
|
|
|
854
857
|
|
|
855
858
|
# print("problem.lower_bounds "+str(problem.lower_bounds))
|
|
856
859
|
# handle the box constraints
|
|
860
|
+
new_candidate_list = []
|
|
857
861
|
for i in range(problem.dim):
|
|
858
862
|
if candidate_x[i] <= problem.lower_bounds[i]:
|
|
859
|
-
|
|
863
|
+
new_candidate_list.append(problem.lower_bounds[i] + 0.01)
|
|
860
864
|
elif candidate_x[i] >= problem.upper_bounds[i]:
|
|
861
|
-
|
|
865
|
+
new_candidate_list.append(problem.upper_bounds[i] - 0.01)
|
|
866
|
+
else:
|
|
867
|
+
new_candidate_list.append(candidate_x[i])
|
|
868
|
+
candidate_x = tuple(new_candidate_list)
|
|
862
869
|
|
|
863
870
|
# store the solution (and function estimate at it) to the subproblem as a candidate for the next iterate
|
|
864
|
-
candidate_solution = self.create_new_solution(
|
|
865
|
-
tuple(candidate_x), problem
|
|
866
|
-
)
|
|
871
|
+
candidate_solution = self.create_new_solution(candidate_x, problem)
|
|
867
872
|
visited_pts_list.append(candidate_solution)
|
|
868
873
|
|
|
869
874
|
# if we use crn, then the candidate solution has the same sample size as the incumbent solution
|
|
@@ -876,7 +881,7 @@ class ASTRODF(Solver):
|
|
|
876
881
|
expended_budget += pilot_run
|
|
877
882
|
sample_size = pilot_run
|
|
878
883
|
while True:
|
|
879
|
-
# if
|
|
884
|
+
# if enable_gradient:
|
|
880
885
|
# # print("incumbent_solution.objectives_gradients_var[0] "+str(candidate_solution.objectives_gradients_var[0]))
|
|
881
886
|
# while norm(candidate_solution.objectives_gradients_var[0]) == 0 and candidate_solution.n_reps < max(pilot_run, lambda_max/100):
|
|
882
887
|
# problem.simulate(candidate_solution, 1)
|
|
@@ -900,7 +905,7 @@ class ASTRODF(Solver):
|
|
|
900
905
|
problem.simulate(candidate_solution, 1)
|
|
901
906
|
expended_budget += 1
|
|
902
907
|
sample_size += 1
|
|
903
|
-
|
|
908
|
+
|
|
904
909
|
# TODO: make sure the solution whose estimated objevtive is abrupted bc of budget is not added to the list of recommended solutions, unless the error is negligible ...
|
|
905
910
|
# if (expended_budget >= budget_limit) and (sample_size < stopping):
|
|
906
911
|
# final_ob = fval[0]
|
|
@@ -911,7 +916,7 @@ class ASTRODF(Solver):
|
|
|
911
916
|
# also if the candidate solution's variance is high that could be caused by stopping early due to exhausting budget
|
|
912
917
|
# print("cv "+str(candidate_solution.objectives_var/(candidate_solution.n_reps * candidate_solution.objectives_mean ** 2)))
|
|
913
918
|
# print("fval[0] - min(fval) "+str(fval[0] - min(fval)))
|
|
914
|
-
if not
|
|
919
|
+
if not enable_gradient and (
|
|
915
920
|
(
|
|
916
921
|
(min(fval) < fval_tilde)
|
|
917
922
|
and (
|
|
@@ -931,19 +936,22 @@ class ASTRODF(Solver):
|
|
|
931
936
|
)
|
|
932
937
|
):
|
|
933
938
|
fval_tilde = min(fval)
|
|
934
|
-
candidate_x = y_var[fval.index(min(fval))][0]
|
|
935
|
-
candidate_solution = interpolation_solns[fval.index(min(fval))]
|
|
939
|
+
candidate_x = y_var[fval.index(min(fval))][0] # type: ignore
|
|
940
|
+
candidate_solution = interpolation_solns[fval.index(min(fval))] # type: ignore
|
|
936
941
|
|
|
937
942
|
# compute the success ratio rho
|
|
938
|
-
|
|
939
|
-
|
|
943
|
+
candidate_x_arr = np.array(candidate_x)
|
|
944
|
+
incumbent_x_arr = np.array(incumbent_x)
|
|
945
|
+
s = np.subtract(candidate_x_arr, incumbent_x_arr)
|
|
946
|
+
if enable_gradient:
|
|
940
947
|
model_reduction = -np.dot(s, grad) - 0.5 * np.dot(
|
|
941
948
|
np.dot(s, hessian), s
|
|
942
949
|
)
|
|
943
950
|
else:
|
|
944
951
|
model_reduction = self.evaluate_model(
|
|
945
|
-
np.zeros(problem.dim),
|
|
946
|
-
|
|
952
|
+
np.zeros(problem.dim),
|
|
953
|
+
q, # type: ignore
|
|
954
|
+
) - self.evaluate_model(s, q) # type: ignore
|
|
947
955
|
if model_reduction <= 0:
|
|
948
956
|
rho = 0
|
|
949
957
|
else:
|
|
@@ -959,7 +967,7 @@ class ASTRODF(Solver):
|
|
|
959
967
|
# very successful: expand
|
|
960
968
|
if rho >= eta_2:
|
|
961
969
|
delta_k = min(gamma_1 * delta_k, delta_max)
|
|
962
|
-
if
|
|
970
|
+
if enable_gradient:
|
|
963
971
|
candidate_grad = (
|
|
964
972
|
-1
|
|
965
973
|
* problem.minmax[0]
|
|
@@ -979,10 +987,11 @@ class ASTRODF(Solver):
|
|
|
979
987
|
else:
|
|
980
988
|
r_k = 1.0 / (y_k @ s)
|
|
981
989
|
h_s_k = h_k @ s
|
|
982
|
-
h_k =
|
|
983
|
-
|
|
990
|
+
h_k = (
|
|
991
|
+
h_k
|
|
992
|
+
+ np.outer(y_k, y_k) * r_k
|
|
984
993
|
- np.outer(h_s_k, h_s_k) / (s @ h_s_k)
|
|
985
|
-
)
|
|
994
|
+
) # type: ignore
|
|
986
995
|
# unsuccessful: shrink and reject
|
|
987
996
|
else:
|
|
988
997
|
delta_k = min(gamma_2 * delta_k, delta_max)
|
|
@@ -1059,7 +1068,7 @@ class ASTRODF(Solver):
|
|
|
1059
1068
|
incumbent_solution = self.create_new_solution(
|
|
1060
1069
|
tuple(incumbent_x), problem
|
|
1061
1070
|
)
|
|
1062
|
-
h_k = np.identity(problem.dim)
|
|
1071
|
+
h_k = np.identity(problem.dim).tolist()
|
|
1063
1072
|
|
|
1064
1073
|
while expended_budget < budget:
|
|
1065
1074
|
k += 1
|
|
@@ -1087,7 +1096,7 @@ class ASTRODF(Solver):
|
|
|
1087
1096
|
intermediate_budgets,
|
|
1088
1097
|
kappa,
|
|
1089
1098
|
incumbent_solution,
|
|
1090
|
-
h_k,
|
|
1099
|
+
h_k, # type: ignore
|
|
1091
1100
|
)
|
|
1092
1101
|
|
|
1093
1102
|
return recommended_solns, intermediate_budgets
|
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
Metadata-Version: 2.1
|
|
2
2
|
Name: simoptlib
|
|
3
|
-
Version: 1.1.
|
|
3
|
+
Version: 1.1.1
|
|
4
4
|
Summary: A testbed for simulation-optimization experiments.
|
|
5
5
|
Author-email: David Eckman <eckman@tamu.edu>, Shane Henderson <sgh9@cornell.edu>, Sara Shashaani <sshasha2@ncsu.edu>, William Grochocinski <wagrocho@ncsu.edu>
|
|
6
6
|
License: MIT License
|
|
@@ -76,7 +76,7 @@ Full documentation for the source code can be found on our **[readthedocs page](
|
|
|
76
76
|
- Python >= 3.8
|
|
77
77
|
- To check your Python version, open a terminal and run `python --version`. If you see a message along the lines of `Command not found`, then you likely don't have Python installed. If you know you have it installed but are getting a `Command not found` error, then you may need to [add Python to your PATH](https://realpython.com/add-python-to-path/).
|
|
78
78
|
- For new installs, [Miniconda or Anaconda](https://www.anaconda.com/download) is recommended ([read about the differences between Miniconda and Anaconda](https://docs.anaconda.com/distro-or-miniconda/)). If you already have a compatible IDE (such as VS Code), we've found that Miniconda will work fine at 1/10 of the size of Anaconda. It is ***highly recommended*** to check the box during installation to add Python/Miniconda/Anaconda to your system PATH.
|
|
79
|
-
- Ruby >=
|
|
79
|
+
- Ruby >= 2.5 (required for datafarming)
|
|
80
80
|
- Included on MacOS, but Windows users will need to grab it from [here](https://rubyinstaller.org/).
|
|
81
81
|
- `datafarming` gem < 2.0 (required for datafarming)
|
|
82
82
|
- This can be installed via `gem install datafarming -v 1.4` once Ruby is installed/configured.
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|