spotPython/RELEASE_NOTES.txt at main · sequential-parameter-optimization/spotPython · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
spotpython 0.21.2:

- pytests added
- spot.py:
    - removed method print_results_old()
- update.py and cifar10datamodule.py removed

spotpython 0.20.2:

 - lightdatamodule handles data_val

spotpython 0.20.0:

- spot can handle penalties for NA evaluations


spotpython 0.19.1:

- objectivefunctions.py:
    - vectorized versions

spotpython 0.19.0:

- objectivefunctions.py:
    - class analytical capitalized to analytical

- kriging.py:
    exp_imp() function updated. Using (y_min - y0) / s0 to ensure that the value is a scalar.

spotpython 0.18.15:

- "numpy>=2.1.0" changed to "numpy>=2.0.0 updated in the dependencies to enable tensorflow compatibility

spotpython 0.18.14:

- "river>=0.22.0" and   "numpy>=2.1.0" updated in the dependencies


- objectivefunctions.py:
    doc updated


spotpackage 0.18.13:

- listgenerator.py:
    - New class class ListGenerator:

spotpython 0.18.11:

- testmodel, predictmodel, and cvmodel functions updated, so that they can handle DataModules specified by the user in fun_control.


spotpython 0.18.8:

- lightdatamodule.py:
    - New function: setup restructured
- lightcrossvalidationdatamodule.py:
    - accepts verbosity as an argument


spotpython 0.18.7:

- Separate train and test data sets can be passed to Lightning DataModules

spotpython 0.18.6:

- split.py:
    New function: compute_lengths_from_fractions()
- lightdatamodule.py:
    train, val, test set computation updated

spotpython 0.18.4:
- architecture.py:
    documentation updated

spotpython 0.18.3:

- init.py:
    show_config=True set as Default


spotpython 0.18.2:

- spot.py:
    Class Spot has new attribute "verbosity. Can be set via fun_control["verbosity"].


spotpython 0.18.1:

- nn_linear_regression.py:
train steo uses log as follows:
    self.log("train_loss", loss, on_step=False, on_epoch=True, prog_bar=False)


spotpython 0.18.0

spor.set_self_attribute() removed


spotpython 0.17.2:


- trainmodel.py:
    Tensorboard logger uses the argument name="".

    doumentation updated

- init.py:
    fun_control_init() function accepts additional arguments for Lightning.


spotpython-0.17.1

- improved get_device function

spotpython-0.17.0

- designs.py and spacefilling.py:
Naming Conventions: The class names have been capitalized to follow the PEP 8 convention for class naming:
1. `designs` -> `Designs`
2. `spacefilling` -> `SpaceFilling`

- The dimension of the design, k (int), must be specified by the user. There is not default dimension "2" anymore.


spotpython-0.16.19:

- kriging.py update completed

spotpython-0.16.15:

- initialize_matrices:
1. Simplified Penalty Calculation: Cached the variance calculation to a variable for clean reuse; reused it to set `pen_val` with efficient branching.
2. Consistent Attribute Initialization: Set `dtype` for arrays like `self.one` for float precision; ensured all attributes are either properly initialized or set to `None`.


spotpython-0.16.14:

- spotpython.utils.aggregate.py: aggregate_mean_var() updated:
    1. Imports: Ensured that necessary imports (`numpy` and `pandas`) are included.
    2. Input Validation: Added checks to ensure that `X` and `y` are `numpy` arrays, `X` is 2D, `y` is 1D, and they have matching shapes, helping prevent common input errors.
    3. Use of Pandas Group Functionality: Improved grouping and aggregation by explicitly handling multi-index resulting from `agg`.
    4. Return Type Consistency: Ensured that the description in the documentation matches the actual operation and expected shapes for return arrays `y_mean` and `y_var`.

- initialize_variables:
    1. Input Validation: Added checks to ensure correct shapes for `nat_X` and `nat_y`, which prevent unexpected errors during execution.
    2. Proper Initialization Instead of Assigning in Tuples: Decomposed the tuple assignment directly for clarity.

- set_variable_types:
    1. Corrected Initialization of `var_type`: If the length of `var_type` is less than `k`, the variable types are now all set to `'num'` consistently, rather than attempting to multiply the list.
    2. NumPy Vectorization: Moved from `map` and list comprehension to NumPy vectorized operations, which are more efficient and readable when handling large data.
    3. Simplified Ordered Mask Logic: Used NumPy's `isin` function, which directly checks membership in a list of types relevant for `ordered_mask`.


spotpython-0.16.12:

- kriging.py: method extract_from_bounds() updated.

spotpython-0.16.11:

- documentation: update documentation of the Kriging class

spotpython-0.16.10:

- documentation: update documentation of the Kriging class

spotpython-0.16.9:

- xai.py: add new function viz_net to visualize the network architecture (linear nets)
- dimensions.py: add new function extract_linear_dims that extracts the input and output dimensions of the Linear layers in a PyTorch model.

spotpython-0.16.8:

- xai.py: automatically handle the orientation of the colorbar in the plot_nn_values_scatter function