-
Notifications
You must be signed in to change notification settings - Fork 7
/
Copy pathchangelog.txt
121 lines (109 loc) · 5.01 KB
/
changelog.txt
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
since last version:
- Switched from setup.py to pyproject.toml
-
v2022.06.0
- Found and fixed further errors in the Branin function: now matches paper
- Added MultiFidelityFunction.x_opt attribute for known optimum of function
- Added warnings in MultiFidelityFunction if not l_bound <= x_opt <= u_bound
- Added callable AdjustableMultiFidelityFunction class to serve as factory
instead of the previous 'bare' functions
- Tests:
* 'test_Nd_functions' tests now parameterized using hypothesis' data strategy
- Docs:
* Added mathematical definitions in docstring of every function's file
v2021.10.0
- Added an 'invert' function to enable switching between minimization and maximization
- Update branin_hf definition: correct constant from 2.25 to 22.5
- Update six_hump_camelback_hf definition:
* include omitted minus sign
* rewrite to speed up calculation
v2021.2.0:
- enabled branch coverage and -Werror flag for tests
- updated naming of functions: no space between function and number
- MultiFidelityFunction stores original name as _name, .name is now a property
v2020.8.0:
- can now install development requirements using `pip install mf2[dev]` notation
- conda references now point to conda-forge channel
- updated array-generation strategy in property-test
- removed obsolete code (himmelblau 3-fidelity)
- Added type hints for adjustable functions
- moved contributions section in Readme to separate CONTRIBUTING.md file
- Tests:
* array-generation in property-test now uses hypothesis' numpy array strategy
* coverage now at 100%: Added tests for remaining uncovered code
* Rewrote regression tests to use pytest-regressions plugin
- JOSS paper:
* Fixed typo
* Rewrote 'statement of need' paragraphs
* Improved closing of the paper: how is package used by authors
- Documentation:
* hardware and software version info now mentioned on performance page
* example-usage separated out into separate file in docs/scripts
* figure of example-usage improved to 600dpi
* page on performance added, script for plots added in docs/scripts, including
original matlab source code for borehole, currin and park91a/b
v2020.4.3:
- Bugfix: corrected bounds for adjustable Trid function
v2020.4.2:
- Removed unused artificialMultiidelity.py
- De-duplicated borehole fidelities
- Added LICENSE file to source distribution
- Documentation:
* updated copyright
* updated develop install instructions
* fixed typo: Burrin -> Currin
* Added 'Adjustable' to titles of adjustable function pages
- JOSS paper:
* Added statement of need paragraph
v2020.4.1:
- Added Manifest.in to include requirements and docs in source distribution
v2020.4.0:
- Added documentation for adjustable functions
- Added documentation to bounds variables and MultiFidelityFunction instances
- Expanded getting-started page of documentation
- Documentation: added generic introduction
- Updated JOSS paper:
* added acknowledgements,
* added Forrester reference
* itemized functions included in package
* extended introduction of multi-fidelity definition
* added scalability comparison script + results figure
* updated surjanovic and bingham entry in bib-file
- Added README badges: conda, JOSS, license. Now formatted in a table
- Dependencies automatically installed upon install
- Requirements for tests moved to `requirements-dev.txt`
- Separated requirements into regular 'requirements' and 'requirements-dev'
- Added requirements-dev install instruction to README
- __init__ now uses pkg_resources to version number, defined in setup.py
- Added simple logo to documentation
- Added conda install instruction
- Clarified available fidelity levels and their use in docs/getting-started
- Added installation page to docs
- documentation now shows difference between in- or excluding `fidelity_names=`
in MultiFidelityFunction definition
- Adjustable functions: bounds now separate variables, docstrings updated with correct parameter values for high-low equality
- Tests cleanup: removed old 'import mf2 as mff', paths fixed using pyprojroot
- Update README: added correct module-script instruction for create_regression_files
- Updated travis script to install the package
- Bugfix: fixed mistake in low-fidelity Forrester definition
- Added example-usage to docs: reproducing fig. 1 from Forrester 2007
- Bugfix: mistake in low-fidelity of Forrester function
- Included changelog
v2019.11.3:
- @row_vectorize decorator on functions replaced by np.atleast_2d call
- Improved docstrings for all functions
- Added JOSS paper
- Added README badges: PyPI, gitter chat
- Added README sections: Installation, Example Usage, Contributing and Contact
- Added CI for Python3.8
- Upated PyPI page specifiers: markdown text + Python versions (3.6-3.8)
v2019.11.2:
- Added GPL 3.0 license
- Renamed package: multifidelityfunctions -> mf2
v2019.11.1:
- Correlation-adjustable functions moved to .adjustable subpackage
- Refactored benchmarks to test into single definition list
- Added docs for all non-adjustable functions
- Added docstrings for MultiFidelityFunction properties
v2019.11.0:
- First release