├── LICENSE ├── README.md ├── doc.go ├── example ├── README.md ├── ackley │ └── ackley_test.go ├── queens │ └── queens_test.go └── tsp │ └── tsp_test.go ├── integer ├── cross.go └── int_test.go ├── interface.go ├── perm ├── cross.go ├── doc.go ├── mutation.go ├── perm.go └── perm_test.go ├── pop ├── gen │ └── generational.go └── graph │ └── graph.go ├── real ├── cross.go ├── distributions.go ├── evostrat.go ├── real_test.go └── vector.go ├── sel ├── doc.go ├── elite.go ├── interface.go ├── round_robin.go ├── sel_test.go └── tournament.go ├── stats.go └── stats_test.go /LICENSE: -------------------------------------------------------------------------------- 1 | GNU GENERAL PUBLIC LICENSE 2 | Version 3, 29 June 2007 3 | 4 | Copyright (C) 2007 Free Software Foundation, Inc. 5 | Everyone is permitted to copy and distribute verbatim copies 6 | of this license document, but changing it is not allowed. 7 | 8 | Preamble 9 | 10 | The GNU General Public License is a free, copyleft license for 11 | software and other kinds of works. 12 | 13 | The licenses for most software and other practical works are designed 14 | to take away your freedom to share and change the works. By contrast, 15 | the GNU General Public License is intended to guarantee your freedom to 16 | share and change all versions of a program--to make sure it remains free 17 | software for all its users. We, the Free Software Foundation, use the 18 | GNU General Public License for most of our software; it applies also to 19 | any other work released this way by its authors. You can apply it to 20 | your programs, too. 21 | 22 | When we speak of free software, we are referring to freedom, not 23 | price. Our General Public Licenses are designed to make sure that you 24 | have the freedom to distribute copies of free software (and charge for 25 | them if you wish), that you receive source code or can get it if you 26 | want it, that you can change the software or use pieces of it in new 27 | free programs, and that you know you can do these things. 28 | 29 | To protect your rights, we need to prevent others from denying you 30 | these rights or asking you to surrender the rights. Therefore, you have 31 | certain responsibilities if you distribute copies of the software, or if 32 | you modify it: responsibilities to respect the freedom of others. 33 | 34 | For example, if you distribute copies of such a program, whether 35 | gratis or for a fee, you must pass on to the recipients the same 36 | freedoms that you received. You must make sure that they, too, receive 37 | or can get the source code. And you must show them these terms so they 38 | know their rights. 39 | 40 | Developers that use the GNU GPL protect your rights with two steps: 41 | (1) assert copyright on the software, and (2) offer you this License 42 | giving you legal permission to copy, distribute and/or modify it. 43 | 44 | For the developers' and authors' protection, the GPL clearly explains 45 | that there is no warranty for this free software. For both users' and 46 | authors' sake, the GPL requires that modified versions be marked as 47 | changed, so that their problems will not be attributed erroneously to 48 | authors of previous versions. 49 | 50 | Some devices are designed to deny users access to install or run 51 | modified versions of the software inside them, although the manufacturer 52 | can do so. This is fundamentally incompatible with the aim of 53 | protecting users' freedom to change the software. The systematic 54 | pattern of such abuse occurs in the area of products for individuals to 55 | use, which is precisely where it is most unacceptable. Therefore, we 56 | have designed this version of the GPL to prohibit the practice for those 57 | products. If such problems arise substantially in other domains, we 58 | stand ready to extend this provision to those domains in future versions 59 | of the GPL, as needed to protect the freedom of users. 60 | 61 | Finally, every program is threatened constantly by software patents. 62 | States should not allow patents to restrict development and use of 63 | software on general-purpose computers, but in those that do, we wish to 64 | avoid the special danger that patents applied to a free program could 65 | make it effectively proprietary. To prevent this, the GPL assures that 66 | patents cannot be used to render the program non-free. 67 | 68 | The precise terms and conditions for copying, distribution and 69 | modification follow. 70 | 71 | TERMS AND CONDITIONS 72 | 73 | 0. Definitions. 74 | 75 | "This License" refers to version 3 of the GNU General Public License. 76 | 77 | "Copyright" also means copyright-like laws that apply to other kinds of 78 | works, such as semiconductor masks. 79 | 80 | "The Program" refers to any copyrightable work licensed under this 81 | License. Each licensee is addressed as "you". "Licensees" and 82 | "recipients" may be individuals or organizations. 83 | 84 | To "modify" a work means to copy from or adapt all or part of the work 85 | in a fashion requiring copyright permission, other than the making of an 86 | exact copy. The resulting work is called a "modified version" of the 87 | earlier work or a work "based on" the earlier work. 88 | 89 | A "covered work" means either the unmodified Program or a work based 90 | on the Program. 91 | 92 | To "propagate" a work means to do anything with it that, without 93 | permission, would make you directly or secondarily liable for 94 | infringement under applicable copyright law, except executing it on a 95 | computer or modifying a private copy. Propagation includes copying, 96 | distribution (with or without modification), making available to the 97 | public, and in some countries other activities as well. 98 | 99 | To "convey" a work means any kind of propagation that enables other 100 | parties to make or receive copies. Mere interaction with a user through 101 | a computer network, with no transfer of a copy, is not conveying. 102 | 103 | An interactive user interface displays "Appropriate Legal Notices" 104 | to the extent that it includes a convenient and prominently visible 105 | feature that (1) displays an appropriate copyright notice, and (2) 106 | tells the user that there is no warranty for the work (except to the 107 | extent that warranties are provided), that licensees may convey the 108 | work under this License, and how to view a copy of this License. If 109 | the interface presents a list of user commands or options, such as a 110 | menu, a prominent item in the list meets this criterion. 111 | 112 | 1. Source Code. 113 | 114 | The "source code" for a work means the preferred form of the work 115 | for making modifications to it. "Object code" means any non-source 116 | form of a work. 117 | 118 | A "Standard Interface" means an interface that either is an official 119 | standard defined by a recognized standards body, or, in the case of 120 | interfaces specified for a particular programming language, one that 121 | is widely used among developers working in that language. 122 | 123 | The "System Libraries" of an executable work include anything, other 124 | than the work as a whole, that (a) is included in the normal form of 125 | packaging a Major Component, but which is not part of that Major 126 | Component, and (b) serves only to enable use of the work with that 127 | Major Component, or to implement a Standard Interface for which an 128 | implementation is available to the public in source code form. A 129 | "Major Component", in this context, means a major essential component 130 | (kernel, window system, and so on) of the specific operating system 131 | (if any) on which the executable work runs, or a compiler used to 132 | produce the work, or an object code interpreter used to run it. 133 | 134 | The "Corresponding Source" for a work in object code form means all 135 | the source code needed to generate, install, and (for an executable 136 | work) run the object code and to modify the work, including scripts to 137 | control those activities. However, it does not include the work's 138 | System Libraries, or general-purpose tools or generally available free 139 | programs which are used unmodified in performing those activities but 140 | which are not part of the work. For example, Corresponding Source 141 | includes interface definition files associated with source files for 142 | the work, and the source code for shared libraries and dynamically 143 | linked subprograms that the work is specifically designed to require, 144 | such as by intimate data communication or control flow between those 145 | subprograms and other parts of the work. 146 | 147 | The Corresponding Source need not include anything that users 148 | can regenerate automatically from other parts of the Corresponding 149 | Source. 150 | 151 | The Corresponding Source for a work in source code form is that 152 | same work. 153 | 154 | 2. Basic Permissions. 155 | 156 | All rights granted under this License are granted for the term of 157 | copyright on the Program, and are irrevocable provided the stated 158 | conditions are met. This License explicitly affirms your unlimited 159 | permission to run the unmodified Program. The output from running a 160 | covered work is covered by this License only if the output, given its 161 | content, constitutes a covered work. This License acknowledges your 162 | rights of fair use or other equivalent, as provided by copyright law. 163 | 164 | You may make, run and propagate covered works that you do not 165 | convey, without conditions so long as your license otherwise remains 166 | in force. You may convey covered works to others for the sole purpose 167 | of having them make modifications exclusively for you, or provide you 168 | with facilities for running those works, provided that you comply with 169 | the terms of this License in conveying all material for which you do 170 | not control copyright. Those thus making or running the covered works 171 | for you must do so exclusively on your behalf, under your direction 172 | and control, on terms that prohibit them from making any copies of 173 | your copyrighted material outside their relationship with you. 174 | 175 | Conveying under any other circumstances is permitted solely under 176 | the conditions stated below. Sublicensing is not allowed; section 10 177 | makes it unnecessary. 178 | 179 | 3. Protecting Users' Legal Rights From Anti-Circumvention Law. 180 | 181 | No covered work shall be deemed part of an effective technological 182 | measure under any applicable law fulfilling obligations under article 183 | 11 of the WIPO copyright treaty adopted on 20 December 1996, or 184 | similar laws prohibiting or restricting circumvention of such 185 | measures. 186 | 187 | When you convey a covered work, you waive any legal power to forbid 188 | circumvention of technological measures to the extent such circumvention 189 | is effected by exercising rights under this License with respect to 190 | the covered work, and you disclaim any intention to limit operation or 191 | modification of the work as a means of enforcing, against the work's 192 | users, your or third parties' legal rights to forbid circumvention of 193 | technological measures. 194 | 195 | 4. Conveying Verbatim Copies. 196 | 197 | You may convey verbatim copies of the Program's source code as you 198 | receive it, in any medium, provided that you conspicuously and 199 | appropriately publish on each copy an appropriate copyright notice; 200 | keep intact all notices stating that this License and any 201 | non-permissive terms added in accord with section 7 apply to the code; 202 | keep intact all notices of the absence of any warranty; and give all 203 | recipients a copy of this License along with the Program. 204 | 205 | You may charge any price or no price for each copy that you convey, 206 | and you may offer support or warranty protection for a fee. 207 | 208 | 5. Conveying Modified Source Versions. 209 | 210 | You may convey a work based on the Program, or the modifications to 211 | produce it from the Program, in the form of source code under the 212 | terms of section 4, provided that you also meet all of these conditions: 213 | 214 | a) The work must carry prominent notices stating that you modified 215 | it, and giving a relevant date. 216 | 217 | b) The work must carry prominent notices stating that it is 218 | released under this License and any conditions added under section 219 | 7. This requirement modifies the requirement in section 4 to 220 | "keep intact all notices". 221 | 222 | c) You must license the entire work, as a whole, under this 223 | License to anyone who comes into possession of a copy. This 224 | License will therefore apply, along with any applicable section 7 225 | additional terms, to the whole of the work, and all its parts, 226 | regardless of how they are packaged. This License gives no 227 | permission to license the work in any other way, but it does not 228 | invalidate such permission if you have separately received it. 229 | 230 | d) If the work has interactive user interfaces, each must display 231 | Appropriate Legal Notices; however, if the Program has interactive 232 | interfaces that do not display Appropriate Legal Notices, your 233 | work need not make them do so. 234 | 235 | A compilation of a covered work with other separate and independent 236 | works, which are not by their nature extensions of the covered work, 237 | and which are not combined with it such as to form a larger program, 238 | in or on a volume of a storage or distribution medium, is called an 239 | "aggregate" if the compilation and its resulting copyright are not 240 | used to limit the access or legal rights of the compilation's users 241 | beyond what the individual works permit. Inclusion of a covered work 242 | in an aggregate does not cause this License to apply to the other 243 | parts of the aggregate. 244 | 245 | 6. Conveying Non-Source Forms. 246 | 247 | You may convey a covered work in object code form under the terms 248 | of sections 4 and 5, provided that you also convey the 249 | machine-readable Corresponding Source under the terms of this License, 250 | in one of these ways: 251 | 252 | a) Convey the object code in, or embodied in, a physical product 253 | (including a physical distribution medium), accompanied by the 254 | Corresponding Source fixed on a durable physical medium 255 | customarily used for software interchange. 256 | 257 | b) Convey the object code in, or embodied in, a physical product 258 | (including a physical distribution medium), accompanied by a 259 | written offer, valid for at least three years and valid for as 260 | long as you offer spare parts or customer support for that product 261 | model, to give anyone who possesses the object code either (1) a 262 | copy of the Corresponding Source for all the software in the 263 | product that is covered by this License, on a durable physical 264 | medium customarily used for software interchange, for a price no 265 | more than your reasonable cost of physically performing this 266 | conveying of source, or (2) access to copy the 267 | Corresponding Source from a network server at no charge. 268 | 269 | c) Convey individual copies of the object code with a copy of the 270 | written offer to provide the Corresponding Source. This 271 | alternative is allowed only occasionally and noncommercially, and 272 | only if you received the object code with such an offer, in accord 273 | with subsection 6b. 274 | 275 | d) Convey the object code by offering access from a designated 276 | place (gratis or for a charge), and offer equivalent access to the 277 | Corresponding Source in the same way through the same place at no 278 | further charge. You need not require recipients to copy the 279 | Corresponding Source along with the object code. If the place to 280 | copy the object code is a network server, the Corresponding Source 281 | may be on a different server (operated by you or a third party) 282 | that supports equivalent copying facilities, provided you maintain 283 | clear directions next to the object code saying where to find the 284 | Corresponding Source. Regardless of what server hosts the 285 | Corresponding Source, you remain obligated to ensure that it is 286 | available for as long as needed to satisfy these requirements. 287 | 288 | e) Convey the object code using peer-to-peer transmission, provided 289 | you inform other peers where the object code and Corresponding 290 | Source of the work are being offered to the general public at no 291 | charge under subsection 6d. 292 | 293 | A separable portion of the object code, whose source code is excluded 294 | from the Corresponding Source as a System Library, need not be 295 | included in conveying the object code work. 296 | 297 | A "User Product" is either (1) a "consumer product", which means any 298 | tangible personal property which is normally used for personal, family, 299 | or household purposes, or (2) anything designed or sold for incorporation 300 | into a dwelling. In determining whether a product is a consumer product, 301 | doubtful cases shall be resolved in favor of coverage. For a particular 302 | product received by a particular user, "normally used" refers to a 303 | typical or common use of that class of product, regardless of the status 304 | of the particular user or of the way in which the particular user 305 | actually uses, or expects or is expected to use, the product. A product 306 | is a consumer product regardless of whether the product has substantial 307 | commercial, industrial or non-consumer uses, unless such uses represent 308 | the only significant mode of use of the product. 309 | 310 | "Installation Information" for a User Product means any methods, 311 | procedures, authorization keys, or other information required to install 312 | and execute modified versions of a covered work in that User Product from 313 | a modified version of its Corresponding Source. The information must 314 | suffice to ensure that the continued functioning of the modified object 315 | code is in no case prevented or interfered with solely because 316 | modification has been made. 317 | 318 | If you convey an object code work under this section in, or with, or 319 | specifically for use in, a User Product, and the conveying occurs as 320 | part of a transaction in which the right of possession and use of the 321 | User Product is transferred to the recipient in perpetuity or for a 322 | fixed term (regardless of how the transaction is characterized), the 323 | Corresponding Source conveyed under this section must be accompanied 324 | by the Installation Information. But this requirement does not apply 325 | if neither you nor any third party retains the ability to install 326 | modified object code on the User Product (for example, the work has 327 | been installed in ROM). 328 | 329 | The requirement to provide Installation Information does not include a 330 | requirement to continue to provide support service, warranty, or updates 331 | for a work that has been modified or installed by the recipient, or for 332 | the User Product in which it has been modified or installed. Access to a 333 | network may be denied when the modification itself materially and 334 | adversely affects the operation of the network or violates the rules and 335 | protocols for communication across the network. 336 | 337 | Corresponding Source conveyed, and Installation Information provided, 338 | in accord with this section must be in a format that is publicly 339 | documented (and with an implementation available to the public in 340 | source code form), and must require no special password or key for 341 | unpacking, reading or copying. 342 | 343 | 7. Additional Terms. 344 | 345 | "Additional permissions" are terms that supplement the terms of this 346 | License by making exceptions from one or more of its conditions. 347 | Additional permissions that are applicable to the entire Program shall 348 | be treated as though they were included in this License, to the extent 349 | that they are valid under applicable law. If additional permissions 350 | apply only to part of the Program, that part may be used separately 351 | under those permissions, but the entire Program remains governed by 352 | this License without regard to the additional permissions. 353 | 354 | When you convey a copy of a covered work, you may at your option 355 | remove any additional permissions from that copy, or from any part of 356 | it. (Additional permissions may be written to require their own 357 | removal in certain cases when you modify the work.) You may place 358 | additional permissions on material, added by you to a covered work, 359 | for which you have or can give appropriate copyright permission. 360 | 361 | Notwithstanding any other provision of this License, for material you 362 | add to a covered work, you may (if authorized by the copyright holders of 363 | that material) supplement the terms of this License with terms: 364 | 365 | a) Disclaiming warranty or limiting liability differently from the 366 | terms of sections 15 and 16 of this License; or 367 | 368 | b) Requiring preservation of specified reasonable legal notices or 369 | author attributions in that material or in the Appropriate Legal 370 | Notices displayed by works containing it; or 371 | 372 | c) Prohibiting misrepresentation of the origin of that material, or 373 | requiring that modified versions of such material be marked in 374 | reasonable ways as different from the original version; or 375 | 376 | d) Limiting the use for publicity purposes of names of licensors or 377 | authors of the material; or 378 | 379 | e) Declining to grant rights under trademark law for use of some 380 | trade names, trademarks, or service marks; or 381 | 382 | f) Requiring indemnification of licensors and authors of that 383 | material by anyone who conveys the material (or modified versions of 384 | it) with contractual assumptions of liability to the recipient, for 385 | any liability that these contractual assumptions directly impose on 386 | those licensors and authors. 387 | 388 | All other non-permissive additional terms are considered "further 389 | restrictions" within the meaning of section 10. If the Program as you 390 | received it, or any part of it, contains a notice stating that it is 391 | governed by this License along with a term that is a further 392 | restriction, you may remove that term. If a license document contains 393 | a further restriction but permits relicensing or conveying under this 394 | License, you may add to a covered work material governed by the terms 395 | of that license document, provided that the further restriction does 396 | not survive such relicensing or conveying. 397 | 398 | If you add terms to a covered work in accord with this section, you 399 | must place, in the relevant source files, a statement of the 400 | additional terms that apply to those files, or a notice indicating 401 | where to find the applicable terms. 402 | 403 | Additional terms, permissive or non-permissive, may be stated in the 404 | form of a separately written license, or stated as exceptions; 405 | the above requirements apply either way. 406 | 407 | 8. Termination. 408 | 409 | You may not propagate or modify a covered work except as expressly 410 | provided under this License. Any attempt otherwise to propagate or 411 | modify it is void, and will automatically terminate your rights under 412 | this License (including any patent licenses granted under the third 413 | paragraph of section 11). 414 | 415 | However, if you cease all violation of this License, then your 416 | license from a particular copyright holder is reinstated (a) 417 | provisionally, unless and until the copyright holder explicitly and 418 | finally terminates your license, and (b) permanently, if the copyright 419 | holder fails to notify you of the violation by some reasonable means 420 | prior to 60 days after the cessation. 421 | 422 | Moreover, your license from a particular copyright holder is 423 | reinstated permanently if the copyright holder notifies you of the 424 | violation by some reasonable means, this is the first time you have 425 | received notice of violation of this License (for any work) from that 426 | copyright holder, and you cure the violation prior to 30 days after 427 | your receipt of the notice. 428 | 429 | Termination of your rights under this section does not terminate the 430 | licenses of parties who have received copies or rights from you under 431 | this License. If your rights have been terminated and not permanently 432 | reinstated, you do not qualify to receive new licenses for the same 433 | material under section 10. 434 | 435 | 9. Acceptance Not Required for Having Copies. 436 | 437 | You are not required to accept this License in order to receive or 438 | run a copy of the Program. Ancillary propagation of a covered work 439 | occurring solely as a consequence of using peer-to-peer transmission 440 | to receive a copy likewise does not require acceptance. However, 441 | nothing other than this License grants you permission to propagate or 442 | modify any covered work. These actions infringe copyright if you do 443 | not accept this License. Therefore, by modifying or propagating a 444 | covered work, you indicate your acceptance of this License to do so. 445 | 446 | 10. Automatic Licensing of Downstream Recipients. 447 | 448 | Each time you convey a covered work, the recipient automatically 449 | receives a license from the original licensors, to run, modify and 450 | propagate that work, subject to this License. You are not responsible 451 | for enforcing compliance by third parties with this License. 452 | 453 | An "entity transaction" is a transaction transferring control of an 454 | organization, or substantially all assets of one, or subdividing an 455 | organization, or merging organizations. If propagation of a covered 456 | work results from an entity transaction, each party to that 457 | transaction who receives a copy of the work also receives whatever 458 | licenses to the work the party's predecessor in interest had or could 459 | give under the previous paragraph, plus a right to possession of the 460 | Corresponding Source of the work from the predecessor in interest, if 461 | the predecessor has it or can get it with reasonable efforts. 462 | 463 | You may not impose any further restrictions on the exercise of the 464 | rights granted or affirmed under this License. For example, you may 465 | not impose a license fee, royalty, or other charge for exercise of 466 | rights granted under this License, and you may not initiate litigation 467 | (including a cross-claim or counterclaim in a lawsuit) alleging that 468 | any patent claim is infringed by making, using, selling, offering for 469 | sale, or importing the Program or any portion of it. 470 | 471 | 11. Patents. 472 | 473 | A "contributor" is a copyright holder who authorizes use under this 474 | License of the Program or a work on which the Program is based. The 475 | work thus licensed is called the contributor's "contributor version". 476 | 477 | A contributor's "essential patent claims" are all patent claims 478 | owned or controlled by the contributor, whether already acquired or 479 | hereafter acquired, that would be infringed by some manner, permitted 480 | by this License, of making, using, or selling its contributor version, 481 | but do not include claims that would be infringed only as a 482 | consequence of further modification of the contributor version. For 483 | purposes of this definition, "control" includes the right to grant 484 | patent sublicenses in a manner consistent with the requirements of 485 | this License. 486 | 487 | Each contributor grants you a non-exclusive, worldwide, royalty-free 488 | patent license under the contributor's essential patent claims, to 489 | make, use, sell, offer for sale, import and otherwise run, modify and 490 | propagate the contents of its contributor version. 491 | 492 | In the following three paragraphs, a "patent license" is any express 493 | agreement or commitment, however denominated, not to enforce a patent 494 | (such as an express permission to practice a patent or covenant not to 495 | sue for patent infringement). To "grant" such a patent license to a 496 | party means to make such an agreement or commitment not to enforce a 497 | patent against the party. 498 | 499 | If you convey a covered work, knowingly relying on a patent license, 500 | and the Corresponding Source of the work is not available for anyone 501 | to copy, free of charge and under the terms of this License, through a 502 | publicly available network server or other readily accessible means, 503 | then you must either (1) cause the Corresponding Source to be so 504 | available, or (2) arrange to deprive yourself of the benefit of the 505 | patent license for this particular work, or (3) arrange, in a manner 506 | consistent with the requirements of this License, to extend the patent 507 | license to downstream recipients. "Knowingly relying" means you have 508 | actual knowledge that, but for the patent license, your conveying the 509 | covered work in a country, or your recipient's use of the covered work 510 | in a country, would infringe one or more identifiable patents in that 511 | country that you have reason to believe are valid. 512 | 513 | If, pursuant to or in connection with a single transaction or 514 | arrangement, you convey, or propagate by procuring conveyance of, a 515 | covered work, and grant a patent license to some of the parties 516 | receiving the covered work authorizing them to use, propagate, modify 517 | or convey a specific copy of the covered work, then the patent license 518 | you grant is automatically extended to all recipients of the covered 519 | work and works based on it. 520 | 521 | A patent license is "discriminatory" if it does not include within 522 | the scope of its coverage, prohibits the exercise of, or is 523 | conditioned on the non-exercise of one or more of the rights that are 524 | specifically granted under this License. You may not convey a covered 525 | work if you are a party to an arrangement with a third party that is 526 | in the business of distributing software, under which you make payment 527 | to the third party based on the extent of your activity of conveying 528 | the work, and under which the third party grants, to any of the 529 | parties who would receive the covered work from you, a discriminatory 530 | patent license (a) in connection with copies of the covered work 531 | conveyed by you (or copies made from those copies), or (b) primarily 532 | for and in connection with specific products or compilations that 533 | contain the covered work, unless you entered into that arrangement, 534 | or that patent license was granted, prior to 28 March 2007. 535 | 536 | Nothing in this License shall be construed as excluding or limiting 537 | any implied license or other defenses to infringement that may 538 | otherwise be available to you under applicable patent law. 539 | 540 | 12. No Surrender of Others' Freedom. 541 | 542 | If conditions are imposed on you (whether by court order, agreement or 543 | otherwise) that contradict the conditions of this License, they do not 544 | excuse you from the conditions of this License. If you cannot convey a 545 | covered work so as to satisfy simultaneously your obligations under this 546 | License and any other pertinent obligations, then as a consequence you may 547 | not convey it at all. For example, if you agree to terms that obligate you 548 | to collect a royalty for further conveying from those to whom you convey 549 | the Program, the only way you could satisfy both those terms and this 550 | License would be to refrain entirely from conveying the Program. 551 | 552 | 13. Use with the GNU Affero General Public License. 553 | 554 | Notwithstanding any other provision of this License, you have 555 | permission to link or combine any covered work with a work licensed 556 | under version 3 of the GNU Affero General Public License into a single 557 | combined work, and to convey the resulting work. The terms of this 558 | License will continue to apply to the part which is the covered work, 559 | but the special requirements of the GNU Affero General Public License, 560 | section 13, concerning interaction through a network will apply to the 561 | combination as such. 562 | 563 | 14. Revised Versions of this License. 564 | 565 | The Free Software Foundation may publish revised and/or new versions of 566 | the GNU General Public License from time to time. Such new versions will 567 | be similar in spirit to the present version, but may differ in detail to 568 | address new problems or concerns. 569 | 570 | Each version is given a distinguishing version number. If the 571 | Program specifies that a certain numbered version of the GNU General 572 | Public License "or any later version" applies to it, you have the 573 | option of following the terms and conditions either of that numbered 574 | version or of any later version published by the Free Software 575 | Foundation. If the Program does not specify a version number of the 576 | GNU General Public License, you may choose any version ever published 577 | by the Free Software Foundation. 578 | 579 | If the Program specifies that a proxy can decide which future 580 | versions of the GNU General Public License can be used, that proxy's 581 | public statement of acceptance of a version permanently authorizes you 582 | to choose that version for the Program. 583 | 584 | Later license versions may give you additional or different 585 | permissions. However, no additional obligations are imposed on any 586 | author or copyright holder as a result of your choosing to follow a 587 | later version. 588 | 589 | 15. Disclaimer of Warranty. 590 | 591 | THERE IS NO WARRANTY FOR THE PROGRAM, TO THE EXTENT PERMITTED BY 592 | APPLICABLE LAW. EXCEPT WHEN OTHERWISE STATED IN WRITING THE COPYRIGHT 593 | HOLDERS AND/OR OTHER PARTIES PROVIDE THE PROGRAM "AS IS" WITHOUT WARRANTY 594 | OF ANY KIND, EITHER EXPRESSED OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, 595 | THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 596 | PURPOSE. THE ENTIRE RISK AS TO THE QUALITY AND PERFORMANCE OF THE PROGRAM 597 | IS WITH YOU. SHOULD THE PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF 598 | ALL NECESSARY SERVICING, REPAIR OR CORRECTION. 599 | 600 | 16. Limitation of Liability. 601 | 602 | IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING 603 | WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MODIFIES AND/OR CONVEYS 604 | THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES, INCLUDING ANY 605 | GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING OUT OF THE 606 | USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED TO LOSS OF 607 | DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY YOU OR THIRD 608 | PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER PROGRAMS), 609 | EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE POSSIBILITY OF 610 | SUCH DAMAGES. 611 | 612 | 17. Interpretation of Sections 15 and 16. 613 | 614 | If the disclaimer of warranty and limitation of liability provided 615 | above cannot be given local legal effect according to their terms, 616 | reviewing courts shall apply local law that most closely approximates 617 | an absolute waiver of all civil liability in connection with the 618 | Program, unless a warranty or assumption of liability accompanies a 619 | copy of the Program in return for a fee. 620 | 621 | END OF TERMS AND CONDITIONS 622 | 623 | How to Apply These Terms to Your New Programs 624 | 625 | If you develop a new program, and you want it to be of the greatest 626 | possible use to the public, the best way to achieve this is to make it 627 | free software which everyone can redistribute and change under these terms. 628 | 629 | To do so, attach the following notices to the program. It is safest 630 | to attach them to the start of each source file to most effectively 631 | state the exclusion of warranty; and each file should have at least 632 | the "copyright" line and a pointer to where the full notice is found. 633 | 634 | {one line to give the program's name and a brief idea of what it does.} 635 | Copyright (C) {year} {name of author} 636 | 637 | This program is free software: you can redistribute it and/or modify 638 | it under the terms of the GNU General Public License as published by 639 | the Free Software Foundation, either version 3 of the License, or 640 | (at your option) any later version. 641 | 642 | This program is distributed in the hope that it will be useful, 643 | but WITHOUT ANY WARRANTY; without even the implied warranty of 644 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the 645 | GNU General Public License for more details. 646 | 647 | You should have received a copy of the GNU General Public License 648 | along with this program. If not, see . 649 | 650 | Also add information on how to contact you by electronic and paper mail. 651 | 652 | If the program does terminal interaction, make it output a short 653 | notice like this when it starts in an interactive mode: 654 | 655 | {project} Copyright (C) {year} {fullname} 656 | This program comes with ABSOLUTELY NO WARRANTY; for details type `show w'. 657 | This is free software, and you are welcome to redistribute it 658 | under certain conditions; type `show c' for details. 659 | 660 | The hypothetical commands `show w' and `show c' should show the appropriate 661 | parts of the General Public License. Of course, your program's commands 662 | might be different; for a GUI interface, you would use an "about box". 663 | 664 | You should also get your employer (if you work as a programmer) or school, 665 | if any, to sign a "copyright disclaimer" for the program, if necessary. 666 | For more information on this, and how to apply and follow the GNU GPL, see 667 | . 668 | 669 | The GNU General Public License does not permit incorporating your program 670 | into proprietary programs. If your program is a subroutine library, you 671 | may consider it more useful to permit linking proprietary applications with 672 | the library. If this is what you want to do, use the GNU Lesser General 673 | Public License instead of this License. But first, please read 674 | . 675 | 676 | 677 | GNU LESSER GENERAL PUBLIC LICENSE 678 | Version 3, 29 June 2007 679 | 680 | Copyright (C) 2007 Free Software Foundation, Inc. 681 | Everyone is permitted to copy and distribute verbatim copies 682 | of this license document, but changing it is not allowed. 683 | 684 | 685 | This version of the GNU Lesser General Public License incorporates 686 | the terms and conditions of version 3 of the GNU General Public 687 | License, supplemented by the additional permissions listed below. 688 | 689 | 0. Additional Definitions. 690 | 691 | As used herein, "this License" refers to version 3 of the GNU Lesser 692 | General Public License, and the "GNU GPL" refers to version 3 of the GNU 693 | General Public License. 694 | 695 | "The Library" refers to a covered work governed by this License, 696 | other than an Application or a Combined Work as defined below. 697 | 698 | An "Application" is any work that makes use of an interface provided 699 | by the Library, but which is not otherwise based on the Library. 700 | Defining a subclass of a class defined by the Library is deemed a mode 701 | of using an interface provided by the Library. 702 | 703 | A "Combined Work" is a work produced by combining or linking an 704 | Application with the Library. The particular version of the Library 705 | with which the Combined Work was made is also called the "Linked 706 | Version". 707 | 708 | The "Minimal Corresponding Source" for a Combined Work means the 709 | Corresponding Source for the Combined Work, excluding any source code 710 | for portions of the Combined Work that, considered in isolation, are 711 | based on the Application, and not on the Linked Version. 712 | 713 | The "Corresponding Application Code" for a Combined Work means the 714 | object code and/or source code for the Application, including any data 715 | and utility programs needed for reproducing the Combined Work from the 716 | Application, but excluding the System Libraries of the Combined Work. 717 | 718 | 1. Exception to Section 3 of the GNU GPL. 719 | 720 | You may convey a covered work under sections 3 and 4 of this License 721 | without being bound by section 3 of the GNU GPL. 722 | 723 | 2. Conveying Modified Versions. 724 | 725 | If you modify a copy of the Library, and, in your modifications, a 726 | facility refers to a function or data to be supplied by an Application 727 | that uses the facility (other than as an argument passed when the 728 | facility is invoked), then you may convey a copy of the modified 729 | version: 730 | 731 | a) under this License, provided that you make a good faith effort to 732 | ensure that, in the event an Application does not supply the 733 | function or data, the facility still operates, and performs 734 | whatever part of its purpose remains meaningful, or 735 | 736 | b) under the GNU GPL, with none of the additional permissions of 737 | this License applicable to that copy. 738 | 739 | 3. Object Code Incorporating Material from Library Header Files. 740 | 741 | The object code form of an Application may incorporate material from 742 | a header file that is part of the Library. You may convey such object 743 | code under terms of your choice, provided that, if the incorporated 744 | material is not limited to numerical parameters, data structure 745 | layouts and accessors, or small macros, inline functions and templates 746 | (ten or fewer lines in length), you do both of the following: 747 | 748 | a) Give prominent notice with each copy of the object code that the 749 | Library is used in it and that the Library and its use are 750 | covered by this License. 751 | 752 | b) Accompany the object code with a copy of the GNU GPL and this license 753 | document. 754 | 755 | 4. Combined Works. 756 | 757 | You may convey a Combined Work under terms of your choice that, 758 | taken together, effectively do not restrict modification of the 759 | portions of the Library contained in the Combined Work and reverse 760 | engineering for debugging such modifications, if you also do each of 761 | the following: 762 | 763 | a) Give prominent notice with each copy of the Combined Work that 764 | the Library is used in it and that the Library and its use are 765 | covered by this License. 766 | 767 | b) Accompany the Combined Work with a copy of the GNU GPL and this license 768 | document. 769 | 770 | c) For a Combined Work that displays copyright notices during 771 | execution, include the copyright notice for the Library among 772 | these notices, as well as a reference directing the user to the 773 | copies of the GNU GPL and this license document. 774 | 775 | d) Do one of the following: 776 | 777 | 0) Convey the Minimal Corresponding Source under the terms of this 778 | License, and the Corresponding Application Code in a form 779 | suitable for, and under terms that permit, the user to 780 | recombine or relink the Application with a modified version of 781 | the Linked Version to produce a modified Combined Work, in the 782 | manner specified by section 6 of the GNU GPL for conveying 783 | Corresponding Source. 784 | 785 | 1) Use a suitable shared library mechanism for linking with the 786 | Library. A suitable mechanism is one that (a) uses at run time 787 | a copy of the Library already present on the user's computer 788 | system, and (b) will operate properly with a modified version 789 | of the Library that is interface-compatible with the Linked 790 | Version. 791 | 792 | e) Provide Installation Information, but only if you would otherwise 793 | be required to provide such information under section 6 of the 794 | GNU GPL, and only to the extent that such information is 795 | necessary to install and execute a modified version of the 796 | Combined Work produced by recombining or relinking the 797 | Application with a modified version of the Linked Version. (If 798 | you use option 4d0, the Installation Information must accompany 799 | the Minimal Corresponding Source and Corresponding Application 800 | Code. If you use option 4d1, you must provide the Installation 801 | Information in the manner specified by section 6 of the GNU GPL 802 | for conveying Corresponding Source.) 803 | 804 | 5. Combined Libraries. 805 | 806 | You may place library facilities that are a work based on the 807 | Library side by side in a single library together with other library 808 | facilities that are not Applications and are not covered by this 809 | License, and convey such a combined library under terms of your 810 | choice, if you do both of the following: 811 | 812 | a) Accompany the combined library with a copy of the same work based 813 | on the Library, uncombined with any other library facilities, 814 | conveyed under the terms of this License. 815 | 816 | b) Give prominent notice with the combined library that part of it 817 | is a work based on the Library, and explaining where to find the 818 | accompanying uncombined form of the same work. 819 | 820 | 6. Revised Versions of the GNU Lesser General Public License. 821 | 822 | The Free Software Foundation may publish revised and/or new versions 823 | of the GNU Lesser General Public License from time to time. Such new 824 | versions will be similar in spirit to the present version, but may 825 | differ in detail to address new problems or concerns. 826 | 827 | Each version is given a distinguishing version number. If the 828 | Library as you received it specifies that a certain numbered version 829 | of the GNU Lesser General Public License "or any later version" 830 | applies to it, you have the option of following the terms and 831 | conditions either of that published version or of any later version 832 | published by the Free Software Foundation. If the Library as you 833 | received it does not specify a version number of the GNU Lesser 834 | General Public License, you may choose any version of the GNU Lesser 835 | General Public License ever published by the Free Software Foundation. 836 | 837 | If the Library as you received it specifies that a proxy can decide 838 | whether future versions of the GNU Lesser General Public License shall 839 | apply, that proxy's public statement of acceptance of any version is 840 | permanent authorization for you to choose that version for the 841 | Library. 842 | -------------------------------------------------------------------------------- /README.md: -------------------------------------------------------------------------------- 1 | # Evo 2 | 3 | Evo is a framework for implementing evolutionary algorithms in Go. 4 | 5 | ``` 6 | go get github.com/cbarrick/evo 7 | ``` 8 | 9 | 10 | ## Documentation 11 | 12 | https://godoc.org/github.com/cbarrick/evo 13 | 14 | 15 | ## Status & Contributing 16 | 17 | Evo is a general framework for developing genetic algorithms and more. It began life in fall 2015 as I studied evolutionary algorithms as an undergrad. 18 | 19 | Contributions are welcome! I am currently a student and inexperienced maintainer, so please bear with me as I learn. The focus of the project thus far has been on the API design and less on performance. I am particularly interested in hearing about use-cases that are not well covered and success stories of where Evo excels. Testing, code reviews, and performance audits are always welcome and needed. 20 | 21 | 22 | ## Examples 23 | 24 | You can browse example problems in the [example subpackage]. The examples are maintained as a development tool rather than to provide optimal solutions to the problems they tackle. Reading the examples should give you a good idea of how easy it is to write code with Evo. 25 | 26 | [example subpackage]: https://github.com/cbarrick/evo/tree/master/example 27 | 28 | 29 | ## License (LGPL) 30 | 31 | This program is free software: you can redistribute it and/or modify it under the terms of the GNU Lesser General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version. 32 | 33 | This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public License for more details. 34 | 35 | You should have received a copy of the GNU Lesser General Public License along with this program. If not, see . 36 | -------------------------------------------------------------------------------- /doc.go: -------------------------------------------------------------------------------- 1 | // Evo is a framework for implementing evolutionary algorithms in Go. 2 | // 3 | // Evo exposes a clean and flexible API oriented around two interfaces: `Genome` 4 | // and `Population`. Genomes represent both the function being optimized and the 5 | // representation of solutions. Populations represent the architecture under 6 | // which genomes are evolved. Multiple population types are provided by Evo to 7 | // enable the construction of both common and novel architectures. 8 | // 9 | // The body of the evolutionary loop is defined by an evolve function. For each 10 | // genome in a population, the evolve function is called, receiving some subset 11 | // of the population, called the suitors, as arguments. The evolve function then 12 | // applies the user's variation operators (selection, mutation, etc) and returns 13 | // a genome for the next iteration. common operators for a variety of 14 | // representations are provided as subpackages of Evo. 15 | // 16 | // Populations model the evolution patterns of genomes. A few different 17 | // population types are provided by Evo under the package `evo/pop`. Populations 18 | // themselves implement the Genome interface, making them composeable. Migration 19 | // functions are provided to be used in this context, allowing go novel 20 | // architectures like the island model. 21 | package evo 22 | 23 | // TODO: Keep this in sync with the readme 24 | -------------------------------------------------------------------------------- /example/README.md: -------------------------------------------------------------------------------- 1 | # Evo Examples 2 | 3 | This package contains examples of Evo being used on well know problems. The goal of the examples is not to be the best at solving the particular problem but instead to highlight various use-cases of Evo. Each example is implemented as a unit test and can be easily run with the `go test` command. For example, to run the `ackley` example from the source root: 4 | 5 | go test ./example/ackley -v 6 | 7 | ## Descriptions 8 | 9 | - `ackley`: This example minimizes the Ackley function, a standard benchmark function for real-valued optimization. The problem is highly multimodal with a global minimum of 0 at the origin. The example minimizes the function in 30 dimensions with a self-adaptive (40/2,280)-evolution strategy. 10 | 11 | - `queens`: This example solves the 128-queens problem by minimizing the number of conflicts on the board. The example highlights nested populations by implementing an island model where the population is divided among several sub-populations, called islands, and each island is evolved independently and in parallel. Occasionally migrations of individuals occur between the islands to serve as sources of new genes. 12 | 13 | - `tsp`: This example searches for a minimal tour of the capitals of the 48 contiguous American states (dataset ATT48 of [TSPLIB]). The example uses a diffusion model, where is population is arranged in a hypercube and individuals breed only with their neighbors. The example also highlights hybridization with local search by using a 2-opt hillclimber as a mutation. 14 | 15 | [TSPLIB]: http://comopt.ifi.uni-heidelberg.de/software/TSPLIB95/ 16 | -------------------------------------------------------------------------------- /example/ackley/ackley_test.go: -------------------------------------------------------------------------------- 1 | package ackley 2 | 3 | import ( 4 | "fmt" 5 | "math" 6 | "math/rand" 7 | "sync" 8 | "testing" 9 | 10 | "github.com/cbarrick/evo" 11 | "github.com/cbarrick/evo/pop/gen" 12 | "github.com/cbarrick/evo/real" 13 | "github.com/cbarrick/evo/sel" 14 | ) 15 | 16 | // Tuneables 17 | const ( 18 | dim = 30 // Dimension of the problem. 19 | bounds = 30 // Bounds of object parameters. 20 | precision = 1e-16 // Desired precision. 21 | ) 22 | 23 | // Global objects 24 | var ( 25 | // Count of the number of fitness evaluations. 26 | count struct { 27 | sync.Mutex 28 | n int 29 | } 30 | 31 | // Each of the 40 members of the population generates 7 children and adds 32 | // them to this pool. This pool returns to each member a different one of 33 | // most fit to be their replacement in the next generation. 34 | selector = sel.ElitePool(40, 280) 35 | ) 36 | 37 | // The ackley type specifies our genome. We evolve a real-valued vector that 38 | // optimizes the ackley function. Each genome also contains a vector of strategy 39 | // parameters used with a self-adaptive evolution strategy. 40 | type ackley struct { 41 | gene real.Vector // the object vector to optimize 42 | steps real.Vector // strategy parameters for mutation 43 | fit float64 // the ackley function of the gene 44 | once sync.Once // used to compute fitness lazily 45 | } 46 | 47 | // Returns the fitness as a string. 48 | func (ack *ackley) String() string { 49 | return fmt.Sprint(-ack.Fitness()) 50 | } 51 | 52 | // Fitness returns the ackley function of the gene. We are trying to solve a 53 | // minimization problem, so we return the negative of the traditional formula. 54 | // The fitness of a genome is only computed once across all calls to Fitness by 55 | // using a sync.Once. 56 | func (ack *ackley) Fitness() float64 { 57 | const a, b = 20, 0.2 58 | ack.once.Do(func() { 59 | var sum1, sum2 float64 60 | n := float64(dim) 61 | for _, x := range ack.gene { 62 | sum1 += x * x 63 | sum2 += math.Cos(2 * math.Pi * x) 64 | } 65 | 66 | ack.fit -= a 67 | ack.fit *= math.Exp(-b * math.Sqrt(sum1/n)) 68 | ack.fit -= math.Exp(sum2 / n) 69 | ack.fit += a 70 | ack.fit += math.E 71 | ack.fit *= -1 72 | 73 | count.Lock() 74 | count.n++ 75 | count.Unlock() 76 | }) 77 | return ack.fit 78 | } 79 | 80 | // Evolve implements the inner loop of the evolutionary algorithm. 81 | // The population calls the Evolve method of each genome, in parallel. Then, 82 | // each receiver returns a value to replace it in the next generation. A global 83 | // selector object synchronises replacement among the parallel parents. 84 | func Evolve(ack evo.Genome, suitors []evo.Genome) evo.Genome { 85 | for i := 0; i < 7; i++ { 86 | // Creation: 87 | // We create the child genome from recycled memory when we can. 88 | var child ackley 89 | child.gene = make(real.Vector, dim) 90 | child.steps = make(real.Vector, dim) 91 | 92 | // Crossover: 93 | // Select two parents at random. 94 | // Uniform crossover of object parameters. 95 | // Arithmetic crossover of strategy parameters. 96 | mom := suitors[rand.Intn(len(suitors))].(*ackley) 97 | dad := suitors[rand.Intn(len(suitors))].(*ackley) 98 | real.UniformX(child.gene, mom.gene, dad.gene) 99 | real.ArithX(1, child.steps, mom.steps, dad.steps) 100 | 101 | // Mutation: Evolution Strategy 102 | // Lognormal scaling of strategy parameters. 103 | // Gausian perturbation of object parameters. 104 | child.steps.Adapt() 105 | child.steps.LowBound(precision) 106 | child.gene.Step(child.steps) 107 | child.gene.HighBound(bounds) 108 | child.gene.LowBound(-bounds) 109 | 110 | // Replacement: (40,280) 111 | // Each child is added to the global selection pool. 112 | selector.Put(&child) 113 | } 114 | 115 | // Finally, block until all parallel calls have added their children to the 116 | // selection pool and return one of the selected replacements. 117 | return selector.Get() 118 | } 119 | 120 | func TestAckley(t *testing.T) { 121 | fmt.Printf("Minimize the Ackley function with n=%d\n", dim) 122 | 123 | // Setup: 124 | // We initialize a set of 40 random solutions, 125 | // then add them to a generational population. 126 | seed := make([]evo.Genome, 40) 127 | for i := range seed { 128 | seed[i] = &ackley{ 129 | gene: real.Random(dim, 30), 130 | steps: real.Random(dim, 1), 131 | } 132 | } 133 | var pop gen.Population 134 | pop.Evolve(seed, Evolve) 135 | 136 | // Continuously print statistics while the optimization runs. 137 | pop.Poll(0, func() bool { 138 | count.Lock() 139 | n := count.n 140 | count.Unlock() 141 | stats := pop.Stats() 142 | 143 | // "\x1b[2K" is the escape code to clear the line 144 | // The fitness of minimization problems is negative 145 | fmt.Printf("\x1b[2K\rCount: %7d | Max: %8.3g | Mean: %8.3g | Min: %8.3g | RSD: %9.2e", 146 | n, 147 | -stats.Min(), 148 | -stats.Mean(), 149 | -stats.Max(), 150 | -stats.RSD()) 151 | 152 | return false 153 | }) 154 | 155 | // Terminate after 200,000 fitness evaluations. 156 | pop.Poll(0, func() bool { 157 | count.Lock() 158 | n := count.n 159 | count.Unlock() 160 | return n > 200000 161 | }) 162 | 163 | // Terminate if the standard deviation is low. 164 | pop.Poll(0, func() bool { 165 | stats := pop.Stats() 166 | return stats.SD() < precision 167 | }) 168 | 169 | pop.Wait() 170 | selector.Close() 171 | best := seed[0] 172 | bestFit := seed[0].Fitness() 173 | for i := range seed { 174 | fit := seed[i].Fitness() 175 | if fit > bestFit { 176 | best = seed[i] 177 | bestFit = fit 178 | } 179 | } 180 | fmt.Println("\nSolution:", best) 181 | } 182 | -------------------------------------------------------------------------------- /example/queens/queens_test.go: -------------------------------------------------------------------------------- 1 | package queens 2 | 3 | import ( 4 | "fmt" 5 | "math/rand" 6 | "sync" 7 | "testing" 8 | "time" 9 | 10 | "github.com/cbarrick/evo" 11 | "github.com/cbarrick/evo/perm" 12 | "github.com/cbarrick/evo/pop/gen" 13 | "github.com/cbarrick/evo/pop/graph" 14 | "github.com/cbarrick/evo/sel" 15 | ) 16 | 17 | // Tuneables 18 | const ( 19 | dim = 128 // the dimension of the problem 20 | size = dim * 4 // the size of the population 21 | isl = 4 // the number of islands in which to divide the population 22 | 23 | migration = size / isl / 8 // the size of migrations 24 | delay = 1 * time.Second // the delay between migrations 25 | ) 26 | 27 | // Global objects 28 | var ( 29 | // counts the number of fitness evaluations 30 | count struct { 31 | sync.Mutex 32 | n int 33 | } 34 | ) 35 | 36 | // The queens type is our genome. We evolve a permuation of [0,n) 37 | // representing the position of queens on an n x n board 38 | type queens struct { 39 | gene []int // permutation representation of an n-queens solution 40 | fitness float64 // the negative of the number of conflicts in the solution 41 | once sync.Once // used to compute fitness lazily 42 | } 43 | 44 | // String returns the gene contents and number of conflicts. 45 | func (q *queens) String() string { 46 | return fmt.Sprintf("%v@%v", q.gene, -q.Fitness()) 47 | } 48 | 49 | // Fitness returns the negative of the number of conflicts in the solution. 50 | // The fitness of a genome is only computed once across all calls to Fitness by 51 | // using a sync.Once. 52 | func (q *queens) Fitness() float64 { 53 | q.once.Do(func() { 54 | for i := range q.gene { 55 | for j, k := 1, i-1; k >= 0; j, k = j+1, k-1 { 56 | if q.gene[k] == q.gene[i]+j || q.gene[k] == q.gene[i]-j { 57 | q.fitness-- 58 | } 59 | } 60 | for j, k := 1, i+1; k < len(q.gene); j, k = j+1, k+1 { 61 | if q.gene[k] == q.gene[i]+j || q.gene[k] == q.gene[i]-j { 62 | q.fitness-- 63 | } 64 | } 65 | } 66 | q.fitness /= 2 67 | 68 | count.Lock() 69 | count.n++ 70 | count.Unlock() 71 | }) 72 | return q.fitness 73 | } 74 | 75 | // Evolution implements the body of the evolution loop. 76 | func Evolution(q evo.Genome, suitors []evo.Genome) evo.Genome { 77 | // Crossover: 78 | // We're implementing a diffusion model. For each member of the population, 79 | // we receive a small mating pool containing only our neighbors. We choose 80 | // a mate using a random binary tournament and create a child with 81 | // partially mapped crossover. 82 | mom := q.(*queens) 83 | dad := sel.BinaryTournament(suitors...).(*queens) 84 | child := &queens{gene: make([]int, len(mom.gene))} 85 | perm.PMX(child.gene, mom.gene, dad.gene) 86 | 87 | // Mutation: 88 | // Perform n random swaps where n is taken from an exponential distribution. 89 | // mutationCount := math.Ceil(rand.ExpFloat64() - 0.5) 90 | for i := float64(0); i < 5; i++ { 91 | j := rand.Intn(len(child.gene)) 92 | k := rand.Intn(len(child.gene)) 93 | child.gene[j], child.gene[k] = child.gene[k], child.gene[j] 94 | } 95 | 96 | // Replacement: 97 | // Only replace if the child is better or equal. 98 | if q.Fitness() > child.Fitness() { 99 | return q 100 | } 101 | return child 102 | } 103 | 104 | func TestQueens(t *testing.T) { 105 | fmt.Printf("Find a solution to %d-queens\n", dim) 106 | 107 | // Setup: 108 | // We create an initial set of random candidates and divide them into "islands". 109 | // Each island is evolved independently in a generational population. 110 | // The islands are then linked together into a graph population with 111 | seed := make([]evo.Genome, size) 112 | for i := range seed { 113 | seed[i] = &queens{gene: perm.New(dim)} 114 | } 115 | islands := make([]evo.Genome, isl) 116 | islSize := size / isl 117 | for i := range islands { 118 | var island gen.Population 119 | island.Evolve(seed[i*islSize:(i+1)*islSize], Evolution) 120 | islands[i] = &island 121 | } 122 | pop := graph.Ring(isl) 123 | pop.Evolve(islands, gen.Migrate(migration, delay)) 124 | 125 | // Continuously print statistics while the optimization runs. 126 | pop.Poll(0, func() bool { 127 | count.Lock() 128 | n := count.n 129 | count.Unlock() 130 | stats := pop.Stats() 131 | 132 | // "\x1b[2K" is the xterm escape code to clear the line 133 | // Because this is a minimization problem, the fitness is negative. 134 | // Thus we update the statistics accordingly. 135 | fmt.Printf("\x1b[2K\rCount: %7d | Max: %3.0f | Mean: %3.0f | Min: %3.0f | RSD: %9.2e", 136 | n, 137 | -stats.Min(), 138 | -stats.Mean(), 139 | -stats.Max(), 140 | -stats.RSD()) 141 | 142 | return false 143 | }) 144 | 145 | // Terminate when we've found the solution (when max is 0) 146 | pop.Poll(0, func() bool { 147 | stats := pop.Stats() 148 | return stats.Max() == 0 149 | }) 150 | 151 | // Terminate if We've converged to a deviation is less than 0.01 152 | pop.Poll(0, func() bool { 153 | stats := pop.Stats() 154 | return stats.SD() < 1e-2 155 | }) 156 | 157 | // Terminate after 2,000,000 fitness evaluations. 158 | pop.Poll(0, func() bool { 159 | count.Lock() 160 | n := count.n 161 | count.Unlock() 162 | return n > 2e6 163 | }) 164 | 165 | pop.Wait() 166 | best := seed[0] 167 | bestFit := seed[0].Fitness() 168 | for i := range seed { 169 | fit := seed[i].Fitness() 170 | if fit > bestFit { 171 | best = seed[i] 172 | bestFit = fit 173 | } 174 | } 175 | fmt.Println("\nSolution:", best) 176 | } 177 | -------------------------------------------------------------------------------- /example/tsp/tsp_test.go: -------------------------------------------------------------------------------- 1 | package tsp 2 | 3 | import ( 4 | "fmt" 5 | "math" 6 | "math/rand" 7 | "sync" 8 | "testing" 9 | 10 | "github.com/cbarrick/evo" 11 | "github.com/cbarrick/evo/perm" 12 | "github.com/cbarrick/evo/pop/graph" 13 | "github.com/cbarrick/evo/sel" 14 | ) 15 | 16 | // Constants 17 | const ( 18 | dim = len(cities) // the dimension of the problem 19 | size = 256 // the size of the population 20 | ) 21 | 22 | // Global objects 23 | var ( 24 | // The evolutionary loop managed by the population 25 | pop evo.Population 26 | 27 | // Count of the number of fitness evaluations. 28 | count struct { 29 | sync.Mutex 30 | n int 31 | } 32 | 33 | // A free-list used to recycle memory. 34 | pool = sync.Pool{ 35 | New: func() interface{} { 36 | return rand.Perm(dim) 37 | }, 38 | } 39 | ) 40 | 41 | // A city is a coordinate pair giving the position of the city. 42 | type city struct { 43 | x, y float64 44 | } 45 | 46 | // Dist returns the pseudo-euclidian distance between two cities. This is the 47 | // distance function used in the literature on this problem instance. 48 | // http://comopt.ifi.uni-heidelberg.de/software/TSPLIB95/DOC.PS 49 | func dist(a, b city) float64 { 50 | xd := a.x - b.x 51 | yd := a.y - b.y 52 | d := math.Sqrt(xd*xd/10 + yd*yd/10) 53 | return math.Ceil(d) 54 | } 55 | 56 | // The tsp type is our genome type. 57 | type tsp struct { 58 | gene []int // permutation representation of a tour 59 | fitness float64 // the negative length of the tour 60 | once sync.Once // used to compute fitness lazily 61 | } 62 | 63 | // String returns the gene contents and length of the tour. 64 | func (t *tsp) String() string { 65 | return fmt.Sprintf("%v@%v", t.gene, -t.Fitness()) 66 | } 67 | 68 | // Close recycles the memory of this genome to be use for new genomes. 69 | func (t *tsp) Close() { 70 | pool.Put(t.gene) 71 | t.gene = nil 72 | } 73 | 74 | // Fitness returns the negative length of the tour represented by a tsp genome. 75 | // The fitness is negative because TSP is a minimization problem, but the Evo 76 | // API is phrased in terms of maximization. As a consequence, fitness statistics 77 | // (i.e. the result of pop.Stats()) are also negative: the shortest know path 78 | // would be -stats.Max(). 79 | func (t *tsp) Fitness() float64 { 80 | t.once.Do(func() { 81 | t.fitness = 0 82 | for i := range t.gene { 83 | a := cities[t.gene[i]] 84 | b := cities[t.gene[(i+1)%dim]] 85 | t.fitness -= dist(a, b) 86 | } 87 | 88 | count.Lock() 89 | count.n++ 90 | count.Unlock() 91 | }) 92 | return t.fitness 93 | } 94 | 95 | // TwoOpt performs a 2-opt local search for improvement of the gene. The first 96 | // edge is selected at random and inversions between all other edges are 97 | // evaluated in random order. Even if an improvement is not found, the gene will 98 | // be rotated by an uniform-random amount. We use this search as a form of 99 | // mutation. 100 | func (t *tsp) TwoOpt() { 101 | t.once = sync.Once{} 102 | perm.Rotate(t.gene, rand.Intn(dim)) 103 | for _, i := range rand.Perm(dim) { 104 | if i < 2 { 105 | continue 106 | } 107 | a := cities[t.gene[0]] 108 | b := cities[t.gene[i-1]] 109 | y := cities[t.gene[i]] 110 | z := cities[t.gene[dim-1]] 111 | before := dist(b, y) + dist(z, a) 112 | after := dist(a, y) + dist(z, b) 113 | if after < before { 114 | perm.Reverse(t.gene[:i]) 115 | return 116 | } 117 | } 118 | } 119 | 120 | // Evolve implements the inner loop of the evolutionary algorithm. 121 | // The population calls the Evolve method of each genome, in parallel. Then, 122 | // each receiver returns a value to replace it in the next generation. 123 | func Evolve(current evo.Genome, matingPool []evo.Genome) evo.Genome { 124 | // Selection: 125 | // Select each parent using a simple random binary tournament 126 | mom := sel.BinaryTournament(matingPool...).(*tsp) 127 | dad := sel.BinaryTournament(matingPool...).(*tsp) 128 | 129 | // Crossover: 130 | // Edge recombination 131 | child := &tsp{gene: pool.Get().([]int)} 132 | perm.EdgeX(child.gene, mom.gene, dad.gene) 133 | 134 | // Mutation: 135 | // There is an n% chance for the gene to have n random swaps 136 | // and an n% chance to undergo n steps of a greedy 2-opt hillclimber 137 | for rand.Float64() < 0.1 { 138 | perm.RandSwap(child.gene) 139 | } 140 | for rand.Float64() < 0.1 { 141 | child.TwoOpt() 142 | } 143 | 144 | // Replacement: 145 | // Only replace if the child is better or equal 146 | if current.Fitness() > child.Fitness() { 147 | return current 148 | } 149 | return child 150 | } 151 | 152 | func TestTSP(t *testing.T) { 153 | fmt.Println("Minimize tour of US capitals - optimal is", best) 154 | 155 | // Setup: 156 | // We create a random initial population 157 | // and evolve it using a generational model. 158 | seed := make([]evo.Genome, size) 159 | for i := range seed { 160 | seed[i] = &tsp{gene: pool.Get().([]int)} 161 | } 162 | pop = graph.Hypercube(size) 163 | pop.Evolve(seed, Evolve) 164 | 165 | // Continuously print statistics while the optimization runs. 166 | pop.Poll(0, func() bool { 167 | count.Lock() 168 | n := count.n 169 | count.Unlock() 170 | stats := pop.Stats() 171 | 172 | // "\x1b[2K" is the escape code to clear the line 173 | // The fitness of minimization problems is negative 174 | fmt.Printf("\x1b[2K\rCount: %7d | Max: %6.0f | Mean: %6.0f | Min: %6.0f | RSD: %7.2e", 175 | n, 176 | -stats.Min(), 177 | -stats.Mean(), 178 | -stats.Max(), 179 | -stats.RSD()) 180 | 181 | return false 182 | }) 183 | 184 | // Stop when we get close. Finding the true minimum could take a while. 185 | pop.Poll(0, func() bool { 186 | stats := pop.Stats() 187 | return -stats.Max() < best*1.1 188 | }) 189 | 190 | // Terminate after 2,000,000 fitness evaluations. 191 | pop.Poll(0, func() bool { 192 | count.Lock() 193 | n := count.n 194 | count.Unlock() 195 | return n > 2e6 196 | }) 197 | 198 | pop.Wait() 199 | best := seed[0] 200 | bestFit := seed[0].Fitness() 201 | for i := range seed { 202 | fit := seed[i].Fitness() 203 | if fit > bestFit { 204 | best = seed[i] 205 | bestFit = fit 206 | } 207 | } 208 | fmt.Println("\nTour:", best) 209 | } 210 | 211 | // Best is the minimum tour of the cities. 212 | const best = 10628 213 | 214 | // Cities is a list of the capitals of the 48 contiguous American states. The 215 | // minimum tour length is 10628. This is dataset ATT48 from TSPLIB, a collection 216 | // of traveling salesman problem datasets maintained by Dr. Gerhard Reinelt: 217 | // "http://comopt.ifi.uni-heidelberg.de/software/TSPLIB95/" 218 | var cities = [48]city{ 219 | {6734, 1453}, 220 | {2233, 10}, 221 | {5530, 1424}, 222 | {401, 841}, 223 | {3082, 1644}, 224 | {7608, 4458}, 225 | {7573, 3716}, 226 | {7265, 1268}, 227 | {6898, 1885}, 228 | {1112, 2049}, 229 | {5468, 2606}, 230 | {5989, 2873}, 231 | {4706, 2674}, 232 | {4612, 2035}, 233 | {6347, 2683}, 234 | {6107, 669}, 235 | {7611, 5184}, 236 | {7462, 3590}, 237 | {7732, 4723}, 238 | {5900, 3561}, 239 | {4483, 3369}, 240 | {6101, 1110}, 241 | {5199, 2182}, 242 | {1633, 2809}, 243 | {4307, 2322}, 244 | {675, 1006}, 245 | {7555, 4819}, 246 | {7541, 3981}, 247 | {3177, 756}, 248 | {7352, 4506}, 249 | {7545, 2801}, 250 | {3245, 3305}, 251 | {6426, 3173}, 252 | {4608, 1198}, 253 | {23, 2216}, 254 | {7248, 3779}, 255 | {7762, 4595}, 256 | {7392, 2244}, 257 | {3484, 2829}, 258 | {6271, 2135}, 259 | {4985, 140}, 260 | {1916, 1569}, 261 | {7280, 4899}, 262 | {7509, 3239}, 263 | {10, 2676}, 264 | {6807, 2993}, 265 | {5185, 3258}, 266 | {3023, 1942}, 267 | } 268 | -------------------------------------------------------------------------------- /integer/cross.go: -------------------------------------------------------------------------------- 1 | package integer 2 | 3 | import "math/rand" 4 | 5 | // UniformX performs a uniform crossover of some parents into a child. 6 | func UniformX(child []int, parents ...[]int) { 7 | n := len(parents) 8 | for i := range child { 9 | child[i] = parents[rand.Intn(n)][i] 10 | } 11 | } 12 | 13 | // PointX performs n-point crossover of two parents into a child. 14 | func PointX(n int, child, mom, dad []int) { 15 | if rand.Intn(2) == 0 { 16 | mom, dad = dad, mom 17 | } 18 | for 0 < n { 19 | i := rand.Intn(len(child)-n) + 1 20 | copy(child, mom[:i]) 21 | child = child[i:] 22 | mom, dad = dad[i:], mom[i:] 23 | n-- 24 | } 25 | copy(child, mom) 26 | } 27 | -------------------------------------------------------------------------------- /integer/int_test.go: -------------------------------------------------------------------------------- 1 | package integer_test 2 | 3 | import ( 4 | "testing" 5 | 6 | "github.com/cbarrick/evo/integer" 7 | ) 8 | 9 | // cross.go 10 | // ------------------------- 11 | 12 | func TestUniformX(t *testing.T) { 13 | mom := make([]int, 8) 14 | dad := make([]int, 8) 15 | for i := range mom { 16 | mom[i] = 1 17 | } 18 | for i := range dad { 19 | dad[i] = 2 20 | } 21 | child := make([]int, 8) 22 | integer.UniformX(child, mom, dad) 23 | for i := range child { 24 | if child[i] != mom[i] && child[i] != dad[i] { 25 | t.Fail() 26 | } 27 | } 28 | } 29 | 30 | func TestPointX(t *testing.T) { 31 | mom := make([]int, 8) 32 | dad := make([]int, 8) 33 | for i := range mom { 34 | mom[i] = 1 35 | } 36 | for i := range dad { 37 | dad[i] = 2 38 | } 39 | child := make([]int, 8) 40 | integer.PointX(7, child, mom, dad) 41 | for i := range child { 42 | if child[i] != mom[i] && child[i] != dad[i] { 43 | t.Fail() 44 | } 45 | } 46 | } 47 | -------------------------------------------------------------------------------- /interface.go: -------------------------------------------------------------------------------- 1 | package evo 2 | 3 | import "time" 4 | 5 | // A ConditionFn describes a termination condition. 6 | type ConditionFn func() bool 7 | 8 | // An EvolveFn describes an iteration of the evolution loop. The evolve function 9 | // is called once for each member of the population, possibly in parrallel, and 10 | // is responsible for producing new Genomes given some subset of the population, 11 | // called the suitors. The replacement Genome replaces the current Genome within 12 | // the population. 13 | type EvolveFn func(current Genome, suitors []Genome) (replacement Genome) 14 | 15 | // A Genome describes the function being optimized and the representation of 16 | // solutions. Genomes are provided by the user, and Evo provides convenience 17 | // packages for common representations. 18 | type Genome interface { 19 | // The Fitness method is the function being maximized. 20 | // For minimization problems, return the negative or the inverse. 21 | Fitness() float64 22 | } 23 | 24 | // A Population models the interaction between Genomes during evolution. In 25 | // practice, this determines the kind of parallelism and number of suitors 26 | // during the optimization. 27 | // 28 | // Populations implement Genome, making them composable. For example, an island 29 | // model can be built by composing generational populations into a graph 30 | // population. 31 | type Population interface { 32 | // Fitness returns the maximum fitness of the population. 33 | Fitness() float64 34 | 35 | // Evolve starts the evolution of the population in a separate goroutine. 36 | // Genomes are evolved in place; it is not safe to access the genome slice 37 | // while the evolution is running. 38 | Evolve([]Genome, EvolveFn) 39 | 40 | // Stop terminates the optimization. 41 | Stop() 42 | 43 | // Poll executes a function at some frequency for the duration of the 44 | // current optimization. If the function returns true, the current 45 | // optimization is halted. Use a frequency of 0 for continuous polling. 46 | Poll(freq time.Duration, cond ConditionFn) 47 | 48 | // Wait blocks until the evolution terminates. 49 | Wait() 50 | 51 | // Stats returns various statistics about the population. 52 | Stats() Stats 53 | } 54 | -------------------------------------------------------------------------------- /perm/cross.go: -------------------------------------------------------------------------------- 1 | package perm 2 | 3 | import ( 4 | "math/rand" 5 | ) 6 | 7 | // OrderX performs order crossover. Order crossover is a good choice when you 8 | // want to inherit the relative order of values. 9 | func OrderX(child, mom, dad []int) { 10 | if rand.Float64() < 0.5 { 11 | mom, dad = dad, mom 12 | } 13 | sub, left, right := RandSlice(mom) 14 | copy(child[left:right], sub) 15 | i, j := right, right 16 | for i < left || right <= i { 17 | if Search(sub, dad[j]) == -1 { 18 | child[i] = dad[j] 19 | i = (i + 1) % len(child) 20 | } 21 | j = (j + 1) % len(child) 22 | } 23 | } 24 | 25 | // PMX performs partially mapped crossover. PMX inherits a random slice of one 26 | // parent. The position of the other values is more random when there is greater 27 | // difference between the parents. 28 | func PMX(child, mom, dad []int) { 29 | if rand.Float64() < 0.5 { 30 | mom, dad = dad, mom 31 | } 32 | _, left, right := RandSlice(mom) 33 | 34 | for i := range child { 35 | child[i] = -1 36 | } 37 | copy(child[left:right], mom[left:right]) 38 | 39 | for i := left; i < right; i++ { 40 | if Search(child, dad[i]) == -1 { 41 | j := i 42 | for left <= j && j < right { 43 | j = Search(dad, mom[j]) 44 | } 45 | child[j] = dad[i] 46 | } 47 | } 48 | 49 | for i := range child { 50 | if child[i] == -1 { 51 | child[i] = dad[i] 52 | } 53 | } 54 | } 55 | 56 | // CycleX performs cycle crossover. Cycle crossover is a good choice when you 57 | // want to inherit the absolute position of values. 58 | func CycleX(child, mom, dad []int) { 59 | if rand.Float64() < 0.5 { 60 | mom, dad = dad, mom 61 | } 62 | var cycles [][]int 63 | taken := make([]bool, len(mom)) 64 | for i := range mom { 65 | if !taken[i] { 66 | var cycle []int 67 | for j := i; !taken[j]; { 68 | taken[j] = true 69 | cycle = append(cycle, j) 70 | j = Search(mom, dad[j]) 71 | } 72 | cycles = append(cycles, cycle) 73 | } 74 | } 75 | 76 | var who bool 77 | for i := range cycles { 78 | var parent []int 79 | if who { 80 | parent = mom 81 | } else { 82 | parent = dad 83 | } 84 | for _, j := range cycles[i] { 85 | child[j] = parent[j] 86 | } 87 | if len(cycles[i]) > 1 { 88 | who = !who 89 | } 90 | } 91 | } 92 | 93 | // EdgeX performs edge recombination. Edge recombination is a good choice when 94 | // you want to inherit adjacency information. 95 | func EdgeX(child, mom, dad []int) { 96 | dim := len(mom) 97 | child = child[0:0] 98 | 99 | if rand.Float64() < 0.5 { 100 | mom, dad = dad, mom 101 | } 102 | 103 | // build the table 104 | // doubles are marked by negating the entry 105 | table := make([][]int, dim) 106 | for i := range table { 107 | table[i] = make([]int, 0, 4) 108 | } 109 | for i := range table { 110 | var j int 111 | 112 | var mnext, mprev int 113 | j = Search(mom, i) 114 | if j == 0 { 115 | mnext = 1 116 | mprev = dim - 1 117 | } else if j == dim-1 { 118 | mnext = 0 119 | mprev = dim - 2 120 | } else { 121 | mnext = j + 1 122 | mprev = j - 1 123 | } 124 | table[i] = append(table[i], mom[mnext], mom[mprev]) 125 | 126 | var dnext, dprev int 127 | j = Search(dad, i) 128 | if j == 0 { 129 | dnext = 1 130 | dprev = dim - 1 131 | } else if j == dim-1 { 132 | dnext = 0 133 | dprev = dim - 2 134 | } else { 135 | dnext = j + 1 136 | dprev = j - 1 137 | } 138 | if table[i][0] == dad[dnext] { 139 | table[i][0] = -table[i][0] 140 | } else if table[i][1] == dad[dnext] { 141 | table[i][1] = -table[i][1] 142 | } else { 143 | table[i] = append(table[i], dad[dnext]) 144 | } 145 | if table[i][0] == dad[dprev] { 146 | table[i][0] = -table[i][0] 147 | } else if table[i][1] == dad[dprev] { 148 | table[i][1] = -table[i][1] 149 | } else { 150 | table[i] = append(table[i], dad[dprev]) 151 | } 152 | } 153 | 154 | // clear removes all occurences of x in the table 155 | clear := func(x int) { 156 | for i := range table { 157 | newrow := table[i][0:0] 158 | pos := Search(table[i], x) 159 | neg := Search(table[i], -x) 160 | for j := range table[i] { 161 | if j != pos && j != neg { 162 | newrow = append(newrow, table[i][j]) 163 | } 164 | } 165 | table[i] = newrow 166 | } 167 | } 168 | 169 | // main loop 170 | var reversed bool 171 | current := rand.Intn(dim) 172 | child = append(child, current) 173 | clear(current) 174 | for len(child) < dim { 175 | next := -1 176 | shortest := 5 177 | row := table[current] 178 | if len(row) == 0 { 179 | if !reversed { 180 | Reverse(child) 181 | reversed = true 182 | current = child[len(child)-1] 183 | continue 184 | } else { 185 | for next == -1 || Search(child, next) != -1 { 186 | next = rand.Intn(len(table)) 187 | } 188 | } 189 | } else { 190 | for i := range row { 191 | if row[i] < 0 { 192 | next = -row[i] 193 | break 194 | } else if len(table[row[i]]) < shortest { 195 | shortest = len(table[row[i]]) 196 | next = row[i] 197 | } else if len(table[row[i]]) == shortest { 198 | if rand.Float32() < 0.5 { 199 | next = row[i] 200 | } 201 | } 202 | } 203 | } 204 | reversed = false 205 | child = append(child, next) 206 | clear(next) 207 | current = next 208 | } 209 | } 210 | -------------------------------------------------------------------------------- /perm/doc.go: -------------------------------------------------------------------------------- 1 | // Package perm provides common operators and helpers for integer permutations. 2 | // 3 | // The crossover operators each take 3 integer slices: the "mother" and "father" 4 | // slices provide the genetic material to be filled into the "child" slice. 5 | // This requires the child slice be allocated by the caller. 6 | package perm 7 | -------------------------------------------------------------------------------- /perm/mutation.go: -------------------------------------------------------------------------------- 1 | package perm 2 | 3 | import ( 4 | "math/rand" 5 | ) 6 | 7 | // RandInvert reverses a random slice of the argument. 8 | func RandInvert(gene []int) { 9 | slice, _, _ := RandSlice(gene) 10 | Reverse(slice) 11 | } 12 | 13 | // RandSwap swaps two random elements of the argument. 14 | func RandSwap(gene []int) { 15 | size := len(gene) 16 | i := rand.Intn(size) 17 | j := i 18 | for j == i { 19 | j = rand.Intn(size) 20 | } 21 | gene[i], gene[j] = gene[j], gene[i] 22 | } 23 | -------------------------------------------------------------------------------- /perm/perm.go: -------------------------------------------------------------------------------- 1 | package perm 2 | 3 | import ( 4 | "math/rand" 5 | ) 6 | 7 | // New returns a pseudo-random permutation of the integers [0,n). This function 8 | // is an alias for math/rand.Perm. 9 | func New(n int) []int { 10 | return rand.Perm(n) 11 | } 12 | 13 | // RandSlice returns a random slice of the argument along with the boundaries. 14 | // That is to say: 15 | // sub == slice[left:right] 16 | func RandSlice(slice []int) (sub []int, left, right int) { 17 | left = rand.Intn(len(slice)) 18 | right = left 19 | for right == left { 20 | right = rand.Intn(len(slice)) 21 | } 22 | if right < left { 23 | left, right = right, left 24 | } 25 | return slice[left:right], left, right 26 | } 27 | 28 | // Search searches an int slice for a particular value and returns the index. 29 | // If the value is not found, Search returns -1. 30 | func Search(slice []int, val int) (idx int) { 31 | for idx = range slice { 32 | if slice[idx] == val { 33 | return idx 34 | } 35 | } 36 | return -1 37 | } 38 | 39 | // Reverse reverses an int slice. 40 | func Reverse(slice []int) { 41 | i := 0 42 | j := len(slice) - 1 43 | for i < j { 44 | slice[i], slice[j] = slice[j], slice[i] 45 | i++ 46 | j-- 47 | } 48 | } 49 | 50 | // Rotate rotates a slice by n positions 51 | func Rotate(slice []int, n int) { 52 | size := len(slice) 53 | for n < 0 { 54 | n += size 55 | } 56 | for n >= size { 57 | n -= size 58 | } 59 | Reverse(slice) 60 | Reverse(slice[:n]) 61 | Reverse(slice[n:]) 62 | } 63 | 64 | // Validate panics if the argument is not a permutation. 65 | // This can be useful when testing custom operators. 66 | func Validate(slice []int) { 67 | n := len(slice) 68 | for i := 0; i < n; i++ { 69 | if Search(slice, i) == -1 { 70 | panic("invalid permutation") 71 | } 72 | } 73 | } 74 | -------------------------------------------------------------------------------- /perm/perm_test.go: -------------------------------------------------------------------------------- 1 | package perm_test 2 | 3 | import ( 4 | "math/rand" 5 | "testing" 6 | 7 | "github.com/cbarrick/evo/perm" 8 | ) 9 | 10 | // validate fails the test if perm is not a permutation 11 | func validate(t *testing.T, perm []int) { 12 | n := len(perm) 13 | for i := 0; i < n; i++ { 14 | found := false 15 | for j := range perm { 16 | if perm[j] == i { 17 | found = true 18 | break 19 | } 20 | } 21 | if !found { 22 | t.Fail() 23 | } 24 | } 25 | } 26 | 27 | // cross.go 28 | // ------------------------- 29 | 30 | func TestOrderX(t *testing.T) { 31 | mom := rand.Perm(8) 32 | dad := rand.Perm(8) 33 | child := make([]int, 8) 34 | perm.OrderX(child, mom, dad) 35 | validate(t, child) 36 | } 37 | 38 | func TestPMX(t *testing.T) { 39 | mom := rand.Perm(8) 40 | dad := rand.Perm(8) 41 | child := make([]int, 8) 42 | perm.PMX(child, mom, dad) 43 | validate(t, child) 44 | } 45 | 46 | func TestCycleX(t *testing.T) { 47 | mom := rand.Perm(8) 48 | dad := rand.Perm(8) 49 | child := make([]int, 8) 50 | perm.CycleX(child, mom, dad) 51 | validate(t, child) 52 | } 53 | 54 | func TestEdgeX(t *testing.T) { 55 | mom := rand.Perm(8) 56 | dad := rand.Perm(8) 57 | child := make([]int, 8) 58 | perm.EdgeX(child, mom, dad) 59 | validate(t, child) 60 | } 61 | 62 | // mutation.go 63 | // ------------------------- 64 | 65 | func TestRandInvert(t *testing.T) { 66 | a := rand.Perm(8) 67 | b := make([]int, 8) 68 | copy(b, a) 69 | perm.RandInvert(b) 70 | flipped := false 71 | i, j := 0, 7 72 | for { 73 | if j <= i { 74 | if !flipped { 75 | t.Fail() 76 | } 77 | return 78 | } else if a[i] == b[i] { 79 | i++ 80 | } else if a[j] == b[j] { 81 | j-- 82 | } else { 83 | if flipped { 84 | t.Fail() 85 | return 86 | } 87 | perm.Reverse(b[i : j+1]) 88 | flipped = true 89 | } 90 | } 91 | } 92 | 93 | func TestRandSwap(t *testing.T) { 94 | a := rand.Perm(8) 95 | b := make([]int, 8) 96 | copy(b, a) 97 | perm.RandSwap(b) 98 | swapped := false 99 | i, j := 0, 7 100 | for { 101 | if j <= i { 102 | if !swapped { 103 | t.Fail() 104 | } 105 | return 106 | } else if a[i] == b[i] { 107 | i++ 108 | } else if a[j] == b[j] { 109 | j-- 110 | } else { 111 | if swapped { 112 | t.Fail() 113 | return 114 | } 115 | b[i], b[j] = b[j], b[i] 116 | swapped = true 117 | } 118 | } 119 | } 120 | 121 | // util.go 122 | // ------------------------- 123 | 124 | func TestRandSlice(t *testing.T) { 125 | slice := make([]int, 8) 126 | sub, left, right := perm.RandSlice(slice) 127 | sub[0] = 1 128 | sub[len(sub)-1] = 1 129 | if slice[left] != 1 || slice[right-1] != 1 { 130 | t.Fail() 131 | } 132 | } 133 | 134 | func TestSearch(t *testing.T) { 135 | slice := []int{0, 1, 2, 3, 4, 5, 6, 7} 136 | if perm.Search(slice, 7) != 7 { 137 | t.Fail() 138 | } 139 | if perm.Search(slice, 8) != -1 { 140 | t.Fail() 141 | } 142 | } 143 | 144 | func TestReverse(t *testing.T) { 145 | slice := rand.Perm(8) 146 | rev := make([]int, 8) 147 | copy(rev, slice) 148 | perm.Reverse(rev) 149 | for i, j := 0, 7; i < j; i, j = i+1, j-1 { 150 | if slice[i] != rev[j] { 151 | t.Fail() 152 | } 153 | } 154 | } 155 | 156 | func TestValidate(t *testing.T) { 157 | defer func() { 158 | if recover() == nil { 159 | t.Fail() 160 | } 161 | perm.Validate([]int{0, 1, 2, 3}) 162 | }() 163 | perm.Validate([]int{0, 0, 1, 2}) 164 | } 165 | -------------------------------------------------------------------------------- /pop/gen/generational.go: -------------------------------------------------------------------------------- 1 | // Package gen provides a traditional generational population. 2 | // 3 | // Generational populations evolve their genomes in generations. Each genome 4 | // is given the entire population as suitors to evolve the next generation. Once 5 | // the new generation is constructed, the old generation is replaced. Each 6 | // genome is evolved in parallel, similar to the textbook master-slave 7 | // parallelism. 8 | package gen 9 | 10 | import ( 11 | "math/rand" 12 | "sync" 13 | "time" 14 | 15 | "github.com/cbarrick/evo" 16 | ) 17 | 18 | type Population struct { 19 | members []evo.Genome // the individuals, not safe to touch while running 20 | getc chan chan int // used to access members while running 21 | setc chan chan int // used to mutate members while running 22 | valuec chan evo.Genome // sends/receives genomes for get/set 23 | statsc chan chan evo.Stats // used to get stats while running 24 | stopc chan chan struct{} // used to stop the goroutine 25 | } 26 | 27 | // Evolve initiates the optimization in a separate goroutine. 28 | func (pop *Population) Evolve(members []evo.Genome, body evo.EvolveFn) { 29 | pop.members = members 30 | pop.statsc = make(chan chan evo.Stats) 31 | pop.setc = make(chan chan int) 32 | pop.getc = make(chan chan int) 33 | pop.valuec = make(chan evo.Genome) 34 | pop.stopc = make(chan chan struct{}, 1) 35 | go run(*pop, body) 36 | } 37 | 38 | // Stop terminates the evolution loop. 39 | func (pop *Population) Stop() { 40 | ch := make(chan struct{}) 41 | pop.stopc <- ch 42 | <-ch 43 | close(pop.statsc) 44 | close(pop.setc) 45 | close(pop.getc) 46 | close(pop.valuec) 47 | } 48 | 49 | // Poll executes a function at some frequency for the duration of the 50 | // current optimization. If the function returns true, the current optimization 51 | // is halted. 52 | func (pop *Population) Poll(freq time.Duration, cond evo.ConditionFn) { 53 | done := pop.stopc 54 | go func() { 55 | for { 56 | select { 57 | case <-time.After(freq): 58 | if cond() { 59 | pop.Stop() 60 | return 61 | } 62 | case ch := <-done: 63 | done <- ch 64 | return 65 | } 66 | } 67 | }() 68 | } 69 | 70 | // Wait blocks until the evolution terminates. 71 | func (pop *Population) Wait() { 72 | pop.stopc <- <-pop.stopc 73 | } 74 | 75 | // Stats returns statistics on the fitness of genomes in the population. 76 | func (pop *Population) Stats() (s evo.Stats) { 77 | statsc := <-pop.statsc 78 | if statsc == nil { 79 | for i := range pop.members { 80 | s = s.Put(pop.members[i].Fitness()) 81 | } 82 | return s 83 | } 84 | return <-statsc 85 | } 86 | 87 | // Fitness returns the maximum fitness within the population. 88 | func (pop *Population) Fitness() float64 { 89 | return pop.Stats().Max() 90 | } 91 | 92 | // get returns the ith member of the population. 93 | func (pop *Population) get(i int) (val evo.Genome) { 94 | getter := <-pop.getc 95 | if getter == nil { 96 | val = pop.members[i] 97 | } else { 98 | getter <- i 99 | val = <-pop.valuec 100 | } 101 | return val 102 | } 103 | 104 | // set sets the ith member of the population. 105 | func (pop *Population) set(i int, val evo.Genome) { 106 | setter := <-pop.setc 107 | if setter == nil { 108 | pop.members[i] = val 109 | } else { 110 | setter <- i 111 | pop.valuec <- val 112 | } 113 | } 114 | 115 | // Migrate returns an EvolveFn for using generational populations as genomes. 116 | // The returned migration function exchanges n individuals between the target 117 | // population and one neighboring population. Migration can be slowed down by 118 | // a delay period that occurs before the migration is performed. 119 | // 120 | // The returned migration function can be used to implement an island population 121 | // model where the individuals are divided between some number of generational 122 | // populations, which are themselves linked together into a graph population. 123 | // The generational populations evolve using the user's EvolveFn while the graph 124 | // population evolves using a migration function. 125 | // 126 | // var evolution evo.EvolveFn // the body of the evolution 127 | // var seed []evo.Genome // the initial solutions 128 | // var islands []evo.Genome // the islands 129 | // n := len(seed) / len(islands) // number of solutions per island 130 | // 131 | // for i := range islands { 132 | // var island gen.Population 133 | // island.Evolve(seed[i*n:(i+1)*n], evolution) 134 | // islands[i] = &island 135 | // } 136 | // pop := graph.Ring(len(islands)) 137 | // pop.Evolve(islands, gen.Migrate(5, 1*time.Second)) 138 | func Migrate(n int, delay time.Duration) evo.EvolveFn { 139 | return func(current evo.Genome, suitors []evo.Genome) evo.Genome { 140 | <-time.After(delay) 141 | var a, b *Population 142 | a = current.(*Population) 143 | for b = a; b == a; { 144 | b = suitors[rand.Intn(len(suitors))].(*Population) 145 | } 146 | for i := 0; i < n; i++ { 147 | ai := rand.Intn(len(a.members)) 148 | bi := rand.Intn(len(b.members)) 149 | av := a.get(ai) 150 | bv := b.get(bi) 151 | a.set(ai, bv) 152 | b.set(bi, av) 153 | } 154 | return current 155 | } 156 | } 157 | 158 | // run implements the main goroutine. 159 | func run(pop Population, body evo.EvolveFn) { 160 | var ( 161 | // drives the main loop 162 | loop = make(chan struct{}, 1) 163 | 164 | // receives the results of evolutions 165 | nextgen = make(chan evo.Genome, len(pop.members)) 166 | 167 | // synchronizes pending evolutions 168 | pending sync.WaitGroup 169 | 170 | // used to access/mutate pop.members 171 | getter = make(chan int) 172 | setter = make(chan int) 173 | statsc = make(chan evo.Stats) 174 | ) 175 | 176 | for i := range pop.members { 177 | nextgen <- pop.members[i] 178 | } 179 | loop <- struct{}{} 180 | 181 | for { 182 | select { 183 | case <-loop: 184 | for i := range pop.members { 185 | pop.members[i] = <-nextgen 186 | } 187 | pending.Add(len(pop.members)) 188 | for i := range pop.members { 189 | val := pop.members[i] 190 | go func() { 191 | nextgen <- body(val, pop.members) 192 | pending.Done() 193 | }() 194 | } 195 | go func() { 196 | pending.Wait() 197 | loop <- struct{}{} 198 | }() 199 | 200 | case pop.getc <- getter: 201 | i := <-getter 202 | pop.valuec <- pop.members[i] 203 | 204 | case pop.setc <- setter: 205 | i := <-setter 206 | pop.members[i] = <-pop.valuec 207 | 208 | case pop.statsc <- statsc: 209 | var s evo.Stats 210 | for i := range pop.members { 211 | s = s.Put(pop.members[i].Fitness()) 212 | } 213 | statsc <- s 214 | 215 | case ch := <-pop.stopc: 216 | pending.Wait() 217 | for i := range pop.members { 218 | if subpop, ok := pop.members[i].(evo.Population); ok { 219 | subpop.Stop() 220 | } 221 | } 222 | ch <- struct{}{} 223 | pop.stopc <- ch 224 | return 225 | } 226 | } 227 | } 228 | -------------------------------------------------------------------------------- /pop/graph/graph.go: -------------------------------------------------------------------------------- 1 | // Package graph provides a spatial population for diffusion and island models. 2 | // 3 | // Graph populations map genomes to nodes in a graph. Each node is evolved in 4 | // parallel, and only sees neighboring nodes as suitors. When used as a 5 | // meta-population, this technique is known as the island model. When used as a 6 | // regular population, it is known as the diffusion model. 7 | package graph 8 | 9 | import ( 10 | "time" 11 | 12 | "github.com/cbarrick/evo" 13 | ) 14 | 15 | type Graph []node 16 | 17 | type node struct { 18 | val *evo.Genome 19 | peers []*node 20 | getc chan chan evo.Genome 21 | setc chan chan evo.Genome 22 | closec chan chan struct{} 23 | done chan struct{} 24 | } 25 | 26 | // Grid creates a new graph population arranged as a 2D grid. 27 | func Grid(size int) Graph { 28 | width := size << 1 29 | layout := make([][]int, size) 30 | for i := range layout { 31 | layout[i] = make([]int, 4) 32 | layout[i][0] = (i + 1 + size) % size 33 | layout[i][1] = (i - 1 + size) % size 34 | layout[i][2] = (i + width + size) % size 35 | layout[i][3] = (i - width + size) % size 36 | } 37 | return Custom(layout) 38 | } 39 | 40 | // Hypercube creates a new graph population arranged as a hypercube. 41 | func Hypercube(size int) Graph { 42 | var dim uint 43 | for size > (1 << dim) { 44 | dim++ 45 | } 46 | layout := make([][]int, size) 47 | for i := 0; i < size; i++ { 48 | layout[i] = make([]int, dim) 49 | for j := range layout[i] { 50 | layout[i][j] = (i ^ (1 << uint(j))) % size 51 | } 52 | } 53 | return Custom(layout) 54 | } 55 | 56 | // Ring creates a new graph population arranged as a ring. 57 | func Ring(size int) Graph { 58 | layout := make([][]int, size) 59 | for i := 0; i < size; i++ { 60 | layout[i] = make([]int, 2) 61 | layout[i][0] = (i - 1 + size) % size 62 | layout[i][0] = (i + 1) % size 63 | } 64 | return Custom(layout) 65 | } 66 | 67 | // Custom creates a new graph population with a custom layout. 68 | // The layout is specified as an adjacency list. 69 | func Custom(layout [][]int) Graph { 70 | g := make([]node, len(layout)) 71 | for i := range g { 72 | peers := make([]*node, len(layout[i])) 73 | for j := range layout[i] { 74 | peers[j] = &g[j] 75 | } 76 | g[i].peers = peers 77 | } 78 | return g 79 | } 80 | 81 | // Stats returns statistics on the fitness of genomes in the population. 82 | func (g Graph) Stats() (s evo.Stats) { 83 | for i := range g { 84 | s = s.Put(g[i].get().Fitness()) 85 | } 86 | return s 87 | } 88 | 89 | // Fitness returns the maximum fitness within the population. 90 | func (g Graph) Fitness() float64 { 91 | return g.Stats().Max() 92 | } 93 | 94 | // Evolve starts the optimization in a separate goroutine. 95 | func (g Graph) Evolve(members []evo.Genome, body evo.EvolveFn) { 96 | for i := range g { 97 | g[i].val = &members[i] 98 | g[i].getc = make(chan chan evo.Genome) 99 | g[i].setc = make(chan chan evo.Genome) 100 | g[i].closec = make(chan chan struct{}, 1) 101 | } 102 | for i := range g { 103 | go g[i].run(body) 104 | } 105 | } 106 | 107 | // Stop terminates the optimization. 108 | func (g Graph) Stop() { 109 | ch := make(chan struct{}) 110 | for i := range g { 111 | g[i].closec <- ch 112 | <-ch 113 | close(g[i].getc) 114 | close(g[i].setc) 115 | } 116 | } 117 | 118 | // Poll executes a function at some frequency for the duration of the 119 | // current optimization. If the function returns true, the current optimization 120 | // is halted. 121 | func (g Graph) Poll(freq time.Duration, cond evo.ConditionFn) { 122 | done := g[0].closec 123 | go func() { 124 | for { 125 | select { 126 | case <-time.After(freq): 127 | if cond() { 128 | g.Stop() 129 | return 130 | } 131 | case ch := <-done: 132 | done <- ch 133 | return 134 | } 135 | } 136 | }() 137 | } 138 | 139 | // Wait blocks until the evolution terminates. 140 | func (g Graph) Wait() { 141 | for i := range g { 142 | g[i].wait() 143 | } 144 | } 145 | 146 | func (n node) wait() { 147 | n.closec <- <-n.closec 148 | } 149 | 150 | // get returns the genome underlying the node. 151 | func (n node) get() evo.Genome { 152 | getter := <-n.getc 153 | if getter == nil { 154 | return *n.val 155 | } 156 | return <-getter 157 | } 158 | 159 | // The main goroutine. 160 | func (n node) run(body evo.EvolveFn) { 161 | var ( 162 | // drives the main loop 163 | loop = make(chan struct{}, 1) 164 | 165 | // used to access/mutate the value 166 | getter = make(chan evo.Genome) 167 | setter = make(chan evo.Genome) 168 | ) 169 | 170 | loop <- struct{}{} 171 | 172 | for { 173 | select { 174 | case <-loop: 175 | go func() { 176 | suiters := make([]evo.Genome, len(n.peers)) 177 | for i := range n.peers { 178 | suiters[i] = n.peers[i].get() 179 | } 180 | setter <- body(*n.val, suiters) 181 | loop <- struct{}{} 182 | }() 183 | 184 | case n.getc <- getter: 185 | getter <- *n.val 186 | 187 | case *n.val = <-setter: 188 | 189 | case ch := <-n.closec: 190 | if subpop, ok := (*n.val).(evo.Population); ok { 191 | subpop.Stop() 192 | } 193 | ch <- struct{}{} 194 | n.closec <- ch 195 | return 196 | } 197 | } 198 | } 199 | -------------------------------------------------------------------------------- /real/cross.go: -------------------------------------------------------------------------------- 1 | package real 2 | 3 | import ( 4 | "math/rand" 5 | ) 6 | 7 | // UniformX performs a uniform crossover of some parents into a child. 8 | func UniformX(child Vector, parents ...Vector) { 9 | n := len(parents) 10 | for i := range child { 11 | child[i] = parents[rand.Intn(n)][i] 12 | } 13 | } 14 | 15 | // ArithX performs arithmetic crossover. When the scale is 1, a child is chosen 16 | // uniformly at random from the line segment between the parents. The scale 17 | // affects the length of the segment about the midpoint. Thus when the scale is 18 | // 0, the child is always the midpoint. 19 | func ArithX(scale float64, child, mom, dad Vector) { 20 | // special case when scale == 0, we can find the midpoint in constant space 21 | if scale == 0 { 22 | copy(child, mom) 23 | child.Subtract(dad) 24 | child.Scale(0.5) 25 | child.Add(dad) 26 | return 27 | } 28 | 29 | copy(child, mom) 30 | child.Subtract(dad) 31 | mid := child.Copy() 32 | mid.Scale(0.5) 33 | child.Scale(scale*rand.Float64() - scale/2) 34 | child.Add(dad) 35 | child.Add(mid) 36 | } 37 | -------------------------------------------------------------------------------- /real/distributions.go: -------------------------------------------------------------------------------- 1 | package real 2 | 3 | import ( 4 | "math" 5 | "math/rand" 6 | ) 7 | 8 | func Normal(stdv float64) float64 { 9 | return stdv * rand.NormFloat64() 10 | } 11 | 12 | func Lognormal(rate float64) float64 { 13 | return math.Exp(Normal(rate)) 14 | } 15 | -------------------------------------------------------------------------------- /real/evostrat.go: -------------------------------------------------------------------------------- 1 | package real 2 | 3 | import ( 4 | "math" 5 | ) 6 | 7 | // Adapt performs a lognormal scaling of the vector using a global learning 8 | // rate of 1/sqrt(n) and a local learning rate of 1/sqrt(2*sqrt(n)). This is 9 | // commonly used in evolution strategies to learn the strategy parameters. 10 | func (v Vector) Adapt() { 11 | n := float64(len(v)) 12 | globalrate := 1 / math.Sqrt(n) 13 | localrate := 1 / math.Sqrt(2*math.Sqrt(n)) 14 | global := Lognormal(globalrate) 15 | for i := range v { 16 | v[i] *= Lognormal(localrate) * global 17 | } 18 | } 19 | 20 | // Step performs a gausian purterbation of the vector using position-wise 21 | // step-sizes. This is commonly used in evolution strategies to mutate the 22 | // object parameters, using the strategy parameters as the step-sizes. 23 | func (v Vector) Step(steps Vector) { 24 | for i := range v { 25 | v[i] += Normal(steps[i]) 26 | } 27 | } 28 | -------------------------------------------------------------------------------- /real/real_test.go: -------------------------------------------------------------------------------- 1 | package real_test 2 | 3 | import ( 4 | "math" 5 | "testing" 6 | 7 | "github.com/cbarrick/evo" 8 | "github.com/cbarrick/evo/real" 9 | ) 10 | 11 | // cross.go 12 | // ------------------------- 13 | 14 | func TestUniformX(t *testing.T) { 15 | mom := real.Random(8, 1) 16 | dad := real.Random(8, 1) 17 | child := make([]float64, 8) 18 | real.UniformX(child, mom, dad) 19 | for i := range child { 20 | if child[i] != mom[i] && child[i] != dad[i] { 21 | t.Fail() 22 | } 23 | } 24 | } 25 | 26 | func TestArithX(t *testing.T) { 27 | mom := []float64{0, 0} 28 | dad := []float64{1, -1} 29 | child := []float64{0, 0} 30 | real.ArithX(1, child, mom, dad) 31 | a := 0 < child[0] && child[0] < 1 32 | b := -1 < child[1] && child[1] < 0 33 | c := child[0] == -child[1] 34 | if !a || !b || !c { 35 | t.Fail() 36 | } 37 | } 38 | 39 | // distributions.go 40 | // ------------------------- 41 | 42 | func TestNormal(t *testing.T) { 43 | var s evo.Stats 44 | for i := 0; i < 65536; i++ { 45 | x := real.Normal(1e-3) 46 | s = s.Put(x) 47 | } 48 | mean := s.Mean() 49 | if mean < -1e-3 || 1e-3 < mean || math.IsNaN(mean) { 50 | t.Fail() 51 | } 52 | } 53 | 54 | func TestLognormal(t *testing.T) { 55 | var s evo.Stats 56 | for i := 0; i < 65536; i++ { 57 | x := math.Log(real.Lognormal(1e-3)) 58 | s = s.Put(x) 59 | } 60 | mean := s.Mean() 61 | if mean < -1e-3 || 1e-3 < mean || math.IsNaN(mean) { 62 | t.Fail() 63 | } 64 | } 65 | 66 | // evostrat.go 67 | // ------------------------- 68 | 69 | func TestAdapt(t *testing.T) { 70 | x := real.Random(8, 1) 71 | y := x.Copy() 72 | y.Adapt() 73 | for i := range x { 74 | if x[i] == y[i] { 75 | t.Fail() 76 | } 77 | } 78 | } 79 | 80 | func TestStep(t *testing.T) { 81 | x := make(real.Vector, 8) 82 | x.Step(real.Vector{1, 1, 1, 1, 1, 1, 1, 1}) 83 | for i := range x { 84 | if x[i] < -3 || 3 < x[i] { 85 | t.Fail() 86 | } 87 | } 88 | } 89 | 90 | // vector.go 91 | // ------------------------- 92 | 93 | func TestRandom(t *testing.T) { 94 | x := real.Random(8, 1) 95 | if len(x) != 8 { 96 | t.Fail() 97 | return 98 | } 99 | for i := range x { 100 | if x[i] < 0 || 1 < x[i] { 101 | t.Fail() 102 | } 103 | } 104 | } 105 | 106 | func TestCopy(t *testing.T) { 107 | x := real.Random(8, 1) 108 | y := x.Copy() 109 | for i := range x { 110 | if x[i] != y[i] { 111 | t.Fail() 112 | } 113 | } 114 | x[0] = 0 115 | if x[0] == y[0] { 116 | t.Fail() 117 | } 118 | } 119 | 120 | func TestAdd(t *testing.T) { 121 | x := real.Random(8, 1) 122 | y := real.Random(8, 1) 123 | z := x.Copy() 124 | z.Add(y) 125 | for i := range z { 126 | if z[i] != x[i]+y[i] { 127 | t.Fail() 128 | } 129 | } 130 | } 131 | 132 | func TestSubtract(t *testing.T) { 133 | x := real.Random(8, 1) 134 | y := real.Random(8, 1) 135 | z := x.Copy() 136 | z.Subtract(y) 137 | for i := range z { 138 | if z[i] != x[i]-y[i] { 139 | t.Fail() 140 | } 141 | } 142 | } 143 | 144 | func TestScale(t *testing.T) { 145 | x := real.Random(8, 1) 146 | y := x.Copy() 147 | y.Scale(3) 148 | for i := range y { 149 | if y[i] != x[i]*3 { 150 | t.Fail() 151 | } 152 | } 153 | } 154 | 155 | func TestHighBound(t *testing.T) { 156 | x := real.Vector{1, 3} 157 | x.HighBound(2) 158 | if x[0] != 1 || x[1] != 2 { 159 | t.Fail() 160 | } 161 | } 162 | 163 | func TestLowBound(t *testing.T) { 164 | x := real.Vector{1, 3} 165 | x.LowBound(2) 166 | if x[0] != 2 || x[1] != 3 { 167 | t.Fail() 168 | } 169 | } 170 | 171 | func TestBound(t *testing.T) { 172 | x := real.Vector{1, 4} 173 | x.Bound(real.Vector{2, 2}, real.Vector{3, 3}) 174 | if x[0] != 2 || x[1] != 3 { 175 | t.Fail() 176 | } 177 | } 178 | -------------------------------------------------------------------------------- /real/vector.go: -------------------------------------------------------------------------------- 1 | package real 2 | 3 | import ( 4 | "math/rand" 5 | ) 6 | 7 | type Vector []float64 8 | 9 | // Random generates a random vector of length n. Values are taken uniformly 10 | // between [0,scale). 11 | func Random(n int, scale float64) (v Vector) { 12 | v = make(Vector, n) 13 | for i := range v { 14 | v[i] = rand.Float64() * scale 15 | } 16 | return v 17 | } 18 | 19 | func (v Vector) Copy() Vector { 20 | w := make(Vector, len(v)) 21 | copy(w, v) 22 | return w 23 | } 24 | 25 | func (v Vector) Add(w Vector) Vector { 26 | for i := range v { 27 | v[i] += w[i] 28 | } 29 | return v 30 | } 31 | 32 | func (v Vector) Subtract(w Vector) Vector { 33 | for i := range v { 34 | v[i] -= w[i] 35 | } 36 | return v 37 | } 38 | 39 | func (v Vector) Scale(s float64) Vector { 40 | for i := range v { 41 | v[i] *= s 42 | } 43 | return v 44 | } 45 | 46 | func (v Vector) LowBound(min float64) Vector { 47 | for i := range v { 48 | if v[i] < min { 49 | v[i] = min 50 | } 51 | } 52 | return v 53 | } 54 | 55 | func (v Vector) HighBound(max float64) Vector { 56 | for i := range v { 57 | if v[i] > max { 58 | v[i] = max 59 | } 60 | } 61 | return v 62 | } 63 | 64 | func (v Vector) Bound(lower, upper Vector) Vector { 65 | for i := range v { 66 | if v[i] > upper[i] { 67 | v[i] = upper[i] 68 | } 69 | if v[i] < lower[i] { 70 | v[i] = lower[i] 71 | } 72 | } 73 | return v 74 | } 75 | -------------------------------------------------------------------------------- /sel/doc.go: -------------------------------------------------------------------------------- 1 | // Package sel provides helpers for different selection techniques. 2 | // 3 | // Selection helpers come in two varieties, function selectors and pool 4 | // selectors. Function selectors are simply functions that take some number of 5 | // genomes as "competitors" and return one or more "winners" from the input. 6 | // 7 | // Pool selectors allow many goroutines to contribute competitors and for the 8 | // winners to be retrieved individually. Once all the winners are retrieved, 9 | // the pool is reset for another round of competition. 10 | package sel 11 | -------------------------------------------------------------------------------- /sel/elite.go: -------------------------------------------------------------------------------- 1 | package sel 2 | 3 | import ( 4 | "sort" 5 | 6 | "github.com/cbarrick/evo" 7 | ) 8 | 9 | // An elcomp competes in an elite tournament. 10 | type elcomp struct { 11 | evo.Genome 12 | fit float64 13 | } 14 | 15 | // Elcomps implements the sort interface in descending order by fitness. 16 | type elcomps []elcomp 17 | 18 | func (h elcomps) Len() int { return len(h) } 19 | func (h elcomps) Swap(i, j int) { h[i], h[j] = h[j], h[i] } 20 | func (h elcomps) Less(i, j int) bool { return h[i].fit > h[j].fit } 21 | 22 | // Sort (re)evaluates the fitness of each competitor and sorts them. 23 | // Fitness evaluation is probably expensive, so we do it in parallel. 24 | func (pool elcomps) sort() { 25 | done := make(chan struct{}) 26 | for i := range pool { 27 | go func(i int) { 28 | pool[i].fit = pool[i].Genome.Fitness() 29 | done <- struct{}{} 30 | }(i) 31 | } 32 | for _ = range pool { 33 | <-done 34 | } 35 | sort.Sort(pool) 36 | } 37 | 38 | // Elite returns the µ genomes with the best fitness. 39 | func Elite(µ int, genomes ...evo.Genome) (winners []evo.Genome) { 40 | winners = make([]evo.Genome, µ) 41 | pool := make(elcomps, len(genomes)) 42 | for i := range genomes { 43 | pool[i] = elcomp{genomes[i], 0} 44 | } 45 | pool.sort() 46 | for i := range winners { 47 | winners[i] = pool[i].Genome 48 | } 49 | return winners 50 | } 51 | 52 | // ElitePool creates an elite pool selector. Once λ competitors have been put 53 | // into the pool, they are sorted by fitness. The best µ competitors must then 54 | // be retrieved from the pool. Once the winners are retrieved, the pool starts 55 | // accepting competitors for another tournamnent. 56 | func ElitePool(µ, λ int) Pool { 57 | var p Pool 58 | p.in = make(chan evo.Genome) 59 | p.out = make(chan evo.Genome, µ) 60 | p.close = make(chan chan struct{}) 61 | 62 | go func() { 63 | // the competitors, memory shared accross iterations 64 | pool := make(elcomps, 0, λ) 65 | 66 | for { 67 | // wait to receive all competitors 68 | for len(pool) < λ { 69 | select { 70 | case ch := <-p.close: 71 | ch <- struct{}{} 72 | return 73 | 74 | case val := <-p.in: 75 | // we only add the competitor to the pool 76 | // we do _not_ compute the fitness yet 77 | pool = append(pool, elcomp{val, 0}) 78 | } 79 | } 80 | 81 | // we sort the pool by fitness 82 | // this evaluates the fitness of each member for the first time 83 | pool.sort() 84 | 85 | // send out the most fit µ genomes 86 | pool = pool[:µ] 87 | for i := range pool { 88 | select { 89 | case ch := <-p.close: 90 | ch <- struct{}{} 91 | return 92 | 93 | case p.out <- pool[i].Genome: 94 | } 95 | } 96 | pool = pool[:0] 97 | } 98 | }() 99 | 100 | return p 101 | } 102 | -------------------------------------------------------------------------------- /sel/interface.go: -------------------------------------------------------------------------------- 1 | package sel 2 | 3 | import ( 4 | "github.com/cbarrick/evo" 5 | ) 6 | 7 | // Pool selectors allow many goroutines to contribute competitors to a selection 8 | // process. Pool selectors reset after each competition and must be closed when 9 | // they are no longer needed. 10 | type Pool struct { 11 | in chan evo.Genome 12 | out chan evo.Genome 13 | close chan chan struct{} 14 | } 15 | 16 | // Put adds a competitor to the pool. 17 | // Put blocks until all winners of the previous competition have been retrieved. 18 | func (p Pool) Put(val evo.Genome) { 19 | p.in <- val 20 | } 21 | 22 | // Get retrieves a winner from the most current competition. 23 | // Get blocks until all competitors have been added. 24 | func (p Pool) Get() (val evo.Genome) { 25 | val = <-p.out 26 | return val 27 | } 28 | 29 | // Close stops the pool selector. 30 | func (p Pool) Close() { 31 | ch := make(chan struct{}) 32 | p.close <- ch 33 | <-ch 34 | } 35 | -------------------------------------------------------------------------------- /sel/round_robin.go: -------------------------------------------------------------------------------- 1 | package sel 2 | 3 | import ( 4 | "math" 5 | "math/rand" 6 | "sort" 7 | 8 | "github.com/cbarrick/evo" 9 | ) 10 | 11 | // A dummy is used as the opponent in the bye of an odd tournament 12 | type dummy struct{} 13 | 14 | func (d dummy) Evolve(_ ...evo.Genome) evo.Genome { return dummy{} } 15 | func (d dummy) Fitness() float64 { return math.Inf(-1) } 16 | func (d dummy) Close() {} 17 | 18 | // An rrcomp competes in a round-robin tournament. 19 | type rrcomp struct { 20 | evo.Genome 21 | wins int 22 | } 23 | 24 | // rrcomps implements the sort interface in descending order by wins 25 | type rrcomps []rrcomp 26 | 27 | func (h rrcomps) Len() int { return len(h) } 28 | func (h rrcomps) Swap(i, j int) { h[i], h[j] = h[j], h[i] } 29 | func (h rrcomps) Less(i, j int) bool { return h[i].wins > h[j].wins } 30 | 31 | // tourney performs a concurrent round-robin tournament. 32 | // pool becomes sorted by score. 33 | func (pool rrcomps) tourney(rounds int) { 34 | var ( 35 | size = len(pool) // the size of the tournament 36 | half = size / 2 // half that 37 | tcount = rounds * half // number of tournaments 38 | sched = rand.Perm(len(pool)) // the tournamnent schedule 39 | winners = make(chan int) // communicates the winners 40 | ) 41 | 42 | if size%2 != 0 { 43 | panic("odd size round-robin") 44 | } 45 | 46 | // run returns the best competitor among indices i and j 47 | // sends the index of the winner over winners 48 | run := func(i, j int) { 49 | if pool[i].Fitness() < pool[j].Fitness() { 50 | winners <- j 51 | } else { 52 | winners <- i 53 | } 54 | } 55 | 56 | // for each round, start the tournament according to the schedule 57 | // then rotate the schedule, keeping one element in place, and repeat 58 | for round := 0; round < rounds; round++ { 59 | for i := 0; i < half; i++ { 60 | go run(sched[i], sched[size-1-i]) 61 | } 62 | carry := sched[0] 63 | for i := range sched[:size-1] { 64 | sched[i] = sched[i+1] 65 | } 66 | sched[size-2] = carry 67 | } 68 | 69 | // wait for all competitions to end and keep score 70 | for i := 0; i < tcount; i++ { 71 | j := <-winners 72 | pool[j].wins++ 73 | } 74 | 75 | // finally, sort by score 76 | sort.Sort(pool) 77 | } 78 | 79 | // RoundRobin returns the µ best genomes after some rounds of a tournament. 80 | func RoundRobin(µ, rounds int, genomes ...evo.Genome) (winners []evo.Genome) { 81 | pool := make(rrcomps, 0, len(genomes)+1) 82 | for i := range genomes { 83 | pool = append(pool, rrcomp{genomes[i], 0}) 84 | } 85 | if len(pool)%2 != 0 { 86 | pool = append(pool, rrcomp{dummy{}, -1}) 87 | } 88 | pool.tourney(rounds) 89 | winners = make([]evo.Genome, µ) 90 | for i := range winners { 91 | winners[i] = pool[i].Genome 92 | } 93 | return winners 94 | } 95 | 96 | // RoundRobinPool creates a round-robin pool selector. Once λ competitors 97 | // have been put into the pool, the tournamnet is performed. The best µ 98 | // competitors must then be retrieved from the pool. Once the winners are 99 | // retrieved, the pool starts accepting competitors for another tournamnent. 100 | func RoundRobinPool(µ, λ, rounds int) Pool { 101 | var p Pool 102 | p.in = make(chan evo.Genome) 103 | p.out = make(chan evo.Genome) 104 | p.close = make(chan chan struct{}) 105 | 106 | go func() { 107 | // the competitors, memory shared accross iterations 108 | pool := make(rrcomps, 0, λ+(λ%2)) 109 | 110 | for { 111 | // wait to receive all competitors 112 | for len(pool) < λ { 113 | select { 114 | case ch := <-p.close: 115 | ch <- struct{}{} 116 | return 117 | 118 | case val := <-p.in: 119 | pool = append(pool, rrcomp{val, 0}) 120 | } 121 | } 122 | 123 | // do the tournament 124 | if λ%2 != 0 { 125 | pool = append(pool, rrcomp{dummy{}, -1}) 126 | } 127 | pool.tourney(rounds) 128 | 129 | // send out the µ genomes that won the most 130 | pool = pool[:µ] 131 | for i := range pool { 132 | select { 133 | case ch := <-p.close: 134 | ch <- struct{}{} 135 | return 136 | 137 | case p.out <- pool[i].Genome: 138 | } 139 | } 140 | pool = pool[:0] 141 | } 142 | }() 143 | 144 | return p 145 | } 146 | -------------------------------------------------------------------------------- /sel/sel_test.go: -------------------------------------------------------------------------------- 1 | package sel_test 2 | 3 | import ( 4 | "testing" 5 | 6 | "github.com/cbarrick/evo" 7 | "github.com/cbarrick/evo/sel" 8 | ) 9 | 10 | type dummy float64 11 | 12 | func (d dummy) Evolve(_ ...evo.Genome) evo.Genome { return d } 13 | func (d dummy) Fitness() float64 { return float64(d) } 14 | func (d dummy) Close() {} 15 | 16 | func dummies() []evo.Genome { 17 | return []evo.Genome{ 18 | dummy(1), 19 | dummy(2), 20 | dummy(3), 21 | dummy(4), 22 | dummy(5), 23 | dummy(6), 24 | dummy(7), 25 | dummy(8), 26 | dummy(9), 27 | dummy(0), 28 | } 29 | } 30 | 31 | func search(ds []evo.Genome, d float64) bool { 32 | for i := range ds { 33 | if ds[i].(dummy) == dummy(d) { 34 | return true 35 | } 36 | } 37 | return false 38 | } 39 | 40 | // elite.go 41 | // ------------------------- 42 | 43 | func TestElite(t *testing.T) { 44 | pop := dummies() 45 | elite := sel.Elite(5, pop...) 46 | ok := search(elite, 9) && 47 | search(elite, 8) && 48 | search(elite, 7) && 49 | search(elite, 6) && 50 | search(elite, 5) 51 | if !ok { 52 | t.Fail() 53 | } 54 | } 55 | 56 | func TestElitePool(t *testing.T) { 57 | pop := dummies() 58 | pool := sel.ElitePool(5, 10) 59 | for i := range pop { 60 | pool.Put(pop[i]) 61 | } 62 | for i := dummy(9); 4 < i; i-- { 63 | if pool.Get().(dummy) != i { 64 | t.Fail() 65 | return 66 | } 67 | } 68 | } 69 | 70 | // round_robin.go 71 | // ------------------------- 72 | 73 | func TestRoundRobin(t *testing.T) { 74 | pop := dummies() 75 | elite := sel.RoundRobin(5, 10, pop...) 76 | ok := search(elite, 9) && 77 | search(elite, 8) && 78 | search(elite, 7) && 79 | search(elite, 6) && 80 | search(elite, 5) 81 | if !ok { 82 | t.Fail() 83 | } 84 | } 85 | 86 | func TestRoundRobinPool(t *testing.T) { 87 | pop := dummies() 88 | pool := sel.RoundRobinPool(5, 10, 10) 89 | for i := range pop { 90 | pool.Put(pop[i]) 91 | } 92 | for i := dummy(9); 4 < i; i-- { 93 | if pool.Get().(dummy) != i { 94 | t.Fail() 95 | return 96 | } 97 | } 98 | } 99 | 100 | // tournament.go 101 | // ------------------------- 102 | 103 | func TestTournament(t *testing.T) { 104 | pop := dummies() 105 | winner := sel.Tournament(pop...) 106 | if winner != dummy(9) { 107 | t.Fail() 108 | } 109 | } 110 | 111 | func TestBinaryTournament(t *testing.T) { 112 | var stats evo.Stats 113 | pop := dummies() 114 | for i := 0; i < 1e6; i++ { 115 | winner := sel.BinaryTournament(pop...).(dummy) 116 | stats = stats.Put(float64(winner)) 117 | } 118 | if stats.Mean() < 5.5 || 6.5 < stats.Mean() { 119 | t.Fail() 120 | } 121 | } 122 | -------------------------------------------------------------------------------- /sel/tournament.go: -------------------------------------------------------------------------------- 1 | package sel 2 | 3 | import ( 4 | "math" 5 | "math/rand" 6 | 7 | "github.com/cbarrick/evo" 8 | ) 9 | 10 | // Tournament returns the most fit suitor. 11 | func Tournament(suitors ...evo.Genome) (best evo.Genome) { 12 | var fit, bestfit float64 13 | bestfit = math.Inf(-1) 14 | for i := range suitors { 15 | fit = suitors[i].Fitness() 16 | if fit > bestfit { 17 | bestfit = fit 18 | best = suitors[i] 19 | } 20 | } 21 | return best 22 | } 23 | 24 | // BinaryTournament randomly chooses two suitors and returns the most fit. 25 | func BinaryTournament(suitors ...evo.Genome) evo.Genome { 26 | var x, y, size int 27 | size = len(suitors) 28 | if size > 2 { 29 | x = rand.Intn(size) 30 | y = x 31 | for y == x { 32 | y = rand.Intn(size) 33 | } 34 | } else { 35 | x, y = 0, 1 36 | } 37 | if suitors[x].Fitness() < suitors[y].Fitness() { 38 | return suitors[y] 39 | } 40 | return suitors[x] 41 | } 42 | -------------------------------------------------------------------------------- /stats.go: -------------------------------------------------------------------------------- 1 | package evo 2 | 3 | import ( 4 | "fmt" 5 | "math" 6 | ) 7 | 8 | // A Stats object is a statistics collector. A common source of Stats objects is 9 | // the return value of Population.Stats() which gives statistics about the 10 | // fitness of genomes in the population. 11 | type Stats struct { 12 | max, min float64 13 | mean float64 14 | sumsq float64 // sum of squares of deviation from the mean 15 | count float64 16 | } 17 | 18 | // Put inserts a new value into the data. 19 | func (s Stats) Put(x float64) Stats { 20 | if s.count == 0 { 21 | s.max = math.Inf(-1) 22 | s.min = math.Inf(+1) 23 | } 24 | 25 | delta := x - s.mean 26 | newcount := s.count + 1 27 | 28 | // max & min 29 | s.max = math.Max(s.max, x) 30 | s.min = math.Min(s.min, x) 31 | 32 | // mean 33 | s.mean += delta / newcount 34 | 35 | // sum of squares 36 | s.sumsq += delta * delta * (s.count / newcount) 37 | 38 | // count 39 | s.count = newcount 40 | 41 | return s 42 | } 43 | 44 | // Merge merges the data of two Stats objects. 45 | func (s Stats) Merge(t Stats) Stats { 46 | if s.count == 0 { 47 | s.max = math.Inf(-1) 48 | s.min = math.Inf(+1) 49 | } 50 | 51 | delta := t.mean - s.mean 52 | newcount := t.count + s.count 53 | 54 | // max & min 55 | s.max = math.Max(s.max, t.max) 56 | s.min = math.Min(s.min, t.min) 57 | 58 | // mean 59 | s.mean += delta * (t.count / newcount) 60 | 61 | // sum of squares 62 | s.sumsq += t.sumsq 63 | s.sumsq += delta * delta * (t.count * s.count / newcount) 64 | 65 | // count 66 | s.count = newcount 67 | 68 | return s 69 | } 70 | 71 | // Max returns the maximum data point. 72 | func (s Stats) Max() float64 { 73 | return s.max 74 | } 75 | 76 | // Min returns the minimum data point. 77 | func (s Stats) Min() float64 { 78 | return s.min 79 | } 80 | 81 | // Range returns the difference in the maximum and minimum data points. 82 | func (s Stats) Range() float64 { 83 | return s.max - s.min 84 | } 85 | 86 | // Mean returns the average of the data. 87 | func (s Stats) Mean() float64 { 88 | return s.mean 89 | } 90 | 91 | // Var returns the population variance of the data. 92 | func (s Stats) Var() float64 { 93 | return s.sumsq / s.count 94 | } 95 | 96 | // SD returns the population standard deviation of the data. 97 | func (s Stats) SD() float64 { 98 | return math.Sqrt(s.sumsq / s.count) 99 | } 100 | 101 | // RSD returns the population relative standard deviation of the data, also 102 | // known as the coefficient of variation. 103 | func (s Stats) RSD() float64 { 104 | return s.SD() / s.Mean() 105 | } 106 | 107 | // Count returns the size of the data. 108 | func (s Stats) Count() int { 109 | return int(s.count) 110 | } 111 | 112 | // String returns a string listing a summary of the statistics. 113 | func (s Stats) String() string { 114 | return fmt.Sprintf("Max: %f | Min: %f | SD: %f", 115 | s.Max(), 116 | s.Min(), 117 | s.SD()) 118 | } 119 | -------------------------------------------------------------------------------- /stats_test.go: -------------------------------------------------------------------------------- 1 | package evo_test 2 | 3 | import ( 4 | "testing" 5 | 6 | "github.com/cbarrick/evo" 7 | ) 8 | 9 | func TestMerge(t *testing.T) { 10 | var a, b evo.Stats 11 | for i := float64(0); i < 5; i++ { 12 | a = a.Put(i) 13 | } 14 | for i := float64(5); i < 10; i++ { 15 | b = b.Put(i) 16 | } 17 | stats := a.Merge(b) 18 | if stats.Mean() != 4.5 { 19 | t.Fail() 20 | } 21 | if stats.Var() != 8.25 { 22 | t.Fail() 23 | } 24 | } 25 | 26 | func TestMax(t *testing.T) { 27 | stats := data() 28 | if stats.Max() != 855 { 29 | t.Fail() 30 | } 31 | } 32 | 33 | func TestMin(t *testing.T) { 34 | stats := data() 35 | if stats.Min() != 760 { 36 | t.Fail() 37 | } 38 | } 39 | 40 | func TestRange(t *testing.T) { 41 | stats := data() 42 | if stats.Range() != 95 { 43 | t.Fail() 44 | } 45 | } 46 | 47 | func TestMean(t *testing.T) { 48 | stats := data() 49 | if stats.Mean() < 810.1388888 || 810.1388890 < stats.Mean() { 50 | t.Fail() 51 | } 52 | } 53 | 54 | func TestVar(t *testing.T) { 55 | stats := data() 56 | if stats.Var() < 829.841820 || 829.841822 < stats.Var() { 57 | t.Fail() 58 | } 59 | } 60 | 61 | func TestSD(t *testing.T) { 62 | stats := data() 63 | if stats.SD() < 28.80697520 || 28.80697522 < stats.SD() { 64 | t.Fail() 65 | } 66 | } 67 | 68 | func TestRSD(t *testing.T) { 69 | stats := data() 70 | if stats.RSD() < 0.03555806986 || 0.03555806988 < stats.RSD() { 71 | t.Fail() 72 | } 73 | } 74 | 75 | func TestCount(t *testing.T) { 76 | stats := data() 77 | if stats.Count() != 36 { 78 | t.Fail() 79 | } 80 | } 81 | 82 | func data() (s evo.Stats) { 83 | s = s.Put(810) 84 | s = s.Put(820) 85 | s = s.Put(820) 86 | s = s.Put(840) 87 | s = s.Put(840) 88 | s = s.Put(845) 89 | s = s.Put(785) 90 | s = s.Put(790) 91 | s = s.Put(785) 92 | s = s.Put(835) 93 | s = s.Put(835) 94 | s = s.Put(835) 95 | s = s.Put(845) 96 | s = s.Put(855) 97 | s = s.Put(850) 98 | s = s.Put(760) 99 | s = s.Put(760) 100 | s = s.Put(770) 101 | s = s.Put(820) 102 | s = s.Put(820) 103 | s = s.Put(820) 104 | s = s.Put(820) 105 | s = s.Put(820) 106 | s = s.Put(825) 107 | s = s.Put(775) 108 | s = s.Put(775) 109 | s = s.Put(775) 110 | s = s.Put(825) 111 | s = s.Put(825) 112 | s = s.Put(825) 113 | s = s.Put(815) 114 | s = s.Put(825) 115 | s = s.Put(825) 116 | s = s.Put(770) 117 | s = s.Put(760) 118 | s = s.Put(765) 119 | return s 120 | } 121 | --------------------------------------------------------------------------------