| 1 |
[TOC] |
| 2 |
|
| 3 |
# The C++ library {#main_page} |
| 4 |
\image html "Gudhi_banner.png" |
| 5 |
<br><br><br><br> |
| 6 |
|
| 7 |
## Data structures for cell complexes {#Complexes} |
| 8 |
### Cubical complexes |
| 9 |
|
| 10 |
<table> |
| 11 |
<tr> |
| 12 |
<td width="35%" rowspan=2> |
| 13 |
\image html "Cubical_complex_representation.png" |
| 14 |
</td> |
| 15 |
<td width="50%"> |
| 16 |
The cubical complex is an example of a structured complex useful in computational mathematics (specially |
| 17 |
rigorous numerics) and image analysis.<br> |
| 18 |
</td> |
| 19 |
<td width="15%"> |
| 20 |
<b>Author:</b> Pawel Dlotko<br> |
| 21 |
<b>Introduced in:</b> GUDHI 1.3.0<br> |
| 22 |
<b>License:</b> MIT<br> |
| 23 |
</td> |
| 24 |
</tr> |
| 25 |
<tr> |
| 26 |
<td colspan=2 height="25"> |
| 27 |
<b>User manual:</b> \ref cubical_complex |
| 28 |
</td> |
| 29 |
</tr> |
| 30 |
</table> |
| 31 |
|
| 32 |
### Simplicial complexes |
| 33 |
|
| 34 |
#### Simplex tree |
| 35 |
<table> |
| 36 |
<tr> |
| 37 |
<td width="35%" rowspan=2> |
| 38 |
\image html "Simplex_tree_representation.png" |
| 39 |
</td> |
| 40 |
<td width="50%"> |
| 41 |
The simplex tree is an efficient and flexible |
| 42 |
data structure for representing general (filtered) simplicial complexes. The data structure |
| 43 |
is described in \cite boissonnatmariasimplextreealgorithmica . |
| 44 |
</td> |
| 45 |
<td width="15%"> |
| 46 |
<b>Author:</b> Clément Maria<br> |
| 47 |
<b>Introduced in:</b> GUDHI 1.0.0<br> |
| 48 |
<b>License:</b> MIT<br> |
| 49 |
</td> |
| 50 |
</tr> |
| 51 |
<tr> |
| 52 |
<td colspan=2 height="25"> |
| 53 |
<b>User manual:</b> \ref simplex_tree |
| 54 |
</td> |
| 55 |
</tr> |
| 56 |
</table> |
| 57 |
|
| 58 |
#### Toplex Map |
| 59 |
|
| 60 |
<table> |
| 61 |
<tr> |
| 62 |
<td width="35%" rowspan=2> |
| 63 |
\image html "map.png" |
| 64 |
</td> |
| 65 |
<td width="50%"> |
| 66 |
The Toplex map data structure is composed firstly of a raw storage of toplices (the maximal simplices) |
| 67 |
and secondly of a map which associate any vertex to a set of pointers toward all toplices |
| 68 |
containing this vertex. |
| 69 |
</td> |
| 70 |
<td width="15%"> |
| 71 |
<b>Author:</b> François Godi<br> |
| 72 |
<b>Introduced in:</b> GUDHI 2.1.0<br> |
| 73 |
<b>License:</b> MIT<br> |
| 74 |
</td> |
| 75 |
</tr> |
| 76 |
<tr> |
| 77 |
<td colspan=2 height="25"> |
| 78 |
<b>User manual:</b> \ref toplex_map |
| 79 |
</td> |
| 80 |
</tr> |
| 81 |
</table> |
| 82 |
|
| 83 |
#### Skeleton blocker |
| 84 |
|
| 85 |
<table> |
| 86 |
<tr> |
| 87 |
<td width="35%" rowspan=2> |
| 88 |
\image html "ds_representation.png" |
| 89 |
</td> |
| 90 |
<td width="50%"> |
| 91 |
The Skeleton-Blocker data-structure proposes a light encoding for simplicial complexes by storing only an *implicit* |
| 92 |
representation of its simplices \cite socg_blockers_2011,\cite blockers2012. Intuitively, it just stores the |
| 93 |
1-skeleton of a simplicial complex with a graph and the set of its "missing faces" that is very small in practice. |
| 94 |
This data-structure handles all simplicial complexes operations such as simplex enumeration or simplex removal but |
| 95 |
operations that are particularly efficient are operations that do not require simplex enumeration such as edge |
| 96 |
iteration, link computation or simplex contraction. |
| 97 |
</td> |
| 98 |
<td width="15%"> |
| 99 |
<b>Author:</b> David Salinas<br> |
| 100 |
<b>Introduced in:</b> GUDHI 1.1.0<br> |
| 101 |
<b>License:</b> MIT<br> |
| 102 |
</td> |
| 103 |
</tr> |
| 104 |
<tr> |
| 105 |
<td colspan=2 height="25"> |
| 106 |
<b>User manual:</b> \ref skbl |
| 107 |
</td> |
| 108 |
</tr> |
| 109 |
</table> |
| 110 |
|
| 111 |
#### Basic operation: contraction |
| 112 |
|
| 113 |
<table> |
| 114 |
<tr> |
| 115 |
<td width="35%" rowspan=2> |
| 116 |
\image html "sphere_contraction_representation.png" |
| 117 |
</td> |
| 118 |
<td width="50%"> |
| 119 |
The purpose of this package is to offer a user-friendly interface for edge contraction simplification of huge |
| 120 |
simplicial complexes. It uses the \ref skbl data-structure whose size remains small during simplification of most |
| 121 |
used geometrical complexes of topological data analysis such as the Rips or the Delaunay complexes. In practice, |
| 122 |
the size of this data-structure is even much lower than the total number of simplices. |
| 123 |
</td> |
| 124 |
<td width="15%"> |
| 125 |
<b>Author:</b> David Salinas<br> |
| 126 |
<b>Introduced in:</b> GUDHI 1.1.0<br> |
| 127 |
<b>License:</b> MIT [(LGPL v3)](../../licensing/)<br> |
| 128 |
<b>Requires:</b> \ref cgal |
| 129 |
</td> |
| 130 |
</tr> |
| 131 |
<tr> |
| 132 |
<td colspan=2 height="25"> |
| 133 |
<b>User manual:</b> \ref contr |
| 134 |
</td> |
| 135 |
</tr> |
| 136 |
</table> |
| 137 |
|
| 138 |
## Filtrations |
| 139 |
### Alpha complex and Delaunay |
| 140 |
|
| 141 |
<table> |
| 142 |
<tr> |
| 143 |
<td width="35%" rowspan=2> |
| 144 |
\image html "alpha_complex_representation.png" |
| 145 |
</td> |
| 146 |
<td width="50%"> |
| 147 |
The alpha-complex is a filtration of the finite cells of a Delaunay Triangulation which is topologically |
| 148 |
equivalent to the Čech. |
| 149 |
The filtration value of each simplex is computed as the squared radius of the smallest Delaunay ball |
| 150 |
of this simplex.<br> |
| 151 |
Also provided are the (unfiltered) Delaunay triangulation, and the *incremental* Delaunay triangulation. |
| 152 |
</td> |
| 153 |
<td width="15%"> |
| 154 |
<b>Authors:</b> Vincent Rouvreau, Michael Kerber, Marc Glisse<br> |
| 155 |
<b>Introduced in:</b> GUDHI 1.3.0<br> |
| 156 |
<b>License:</b> MIT [(GPL v3)](../../licensing/), GPL v3 (only for incremental Delaunay)<br> |
| 157 |
<b>Requires:</b> \ref eigen and \ref cgal |
| 158 |
</td> |
| 159 |
</tr> |
| 160 |
<tr> |
| 161 |
<td colspan=2 height="25"> |
| 162 |
<b>User manual:</b> \ref alpha_complex |
| 163 |
</td> |
| 164 |
</tr> |
| 165 |
</table> |
| 166 |
|
| 167 |
### Čech complex |
| 168 |
|
| 169 |
<table> |
| 170 |
<tr> |
| 171 |
<td width="35%" rowspan=2> |
| 172 |
\image html "cech_complex_representation.png" |
| 173 |
</td> |
| 174 |
<td width="50%"> |
| 175 |
The Čech complex is a filtered simplicial complex defined as the nerve of a union of balls. |
| 176 |
The set of all simplices is filtered by the radius of their minimal enclosing ball. |
| 177 |
</td> |
| 178 |
<td width="15%"> |
| 179 |
<b>Author:</b> Vincent Rouvreau, Hind Montassif<br> |
| 180 |
<b>Introduced in:</b> GUDHI 2.2.0<br> |
| 181 |
<b>License:</b> MIT [(LGPL v3)](../../licensing/)<br> |
| 182 |
<b>Requires:</b> \ref cgal |
| 183 |
</td> |
| 184 |
</tr> |
| 185 |
<tr> |
| 186 |
<td colspan=2 height="25"> |
| 187 |
<b>User manual:</b> \ref cech_complex |
| 188 |
</td> |
| 189 |
</tr> |
| 190 |
</table> |
| 191 |
|
| 192 |
### Rips complex |
| 193 |
|
| 194 |
<table> |
| 195 |
<tr> |
| 196 |
<td width="35%" rowspan=2> |
| 197 |
\image html "rips_complex_representation.png" |
| 198 |
</td> |
| 199 |
<td width="50%"> |
| 200 |
The (Vietoris-)Rips complex is a filtered simplicial complex where every tuple of points appears with its diameter |
| 201 |
as filtration value. Alternatively, it is defined by the cliques of a graph, usually representing a metric.<br> |
| 202 |
This complex can be built from a point cloud and a distance function, or from a distance matrix. |
| 203 |
</td> |
| 204 |
<td width="15%"> |
| 205 |
<b>Author:</b> Clément Maria, Pawel Dlotko, Vincent Rouvreau, Marc Glisse<br> |
| 206 |
<b>Introduced in:</b> GUDHI 2.0.0<br> |
| 207 |
<b>License:</b> MIT<br> |
| 208 |
</td> |
| 209 |
</tr> |
| 210 |
<tr> |
| 211 |
<td colspan=2 height="25"> |
| 212 |
<b>User manual:</b> \ref rips_complex |
| 213 |
</td> |
| 214 |
</tr> |
| 215 |
</table> |
| 216 |
|
| 217 |
### Edge collapse |
| 218 |
|
| 219 |
<table> |
| 220 |
<tr> |
| 221 |
<td width="35%" rowspan=2> |
| 222 |
\image html "dominated_edge.png" |
| 223 |
</td> |
| 224 |
<td width="50%"> |
| 225 |
Edge collapse is able to reduce any flag filtration to a smaller flag filtration with the same persistence, using |
| 226 |
only the 1-skeletons of a simplicial complex. |
| 227 |
The reduction is exact and the persistence homology of the reduced sequence is identical to the persistence |
| 228 |
homology of the input sequence. The resulting method is simple and extremely efficient. |
| 229 |
|
| 230 |
Computation of edge collapse and persistent homology of a filtered flag complex via edge collapse as described in |
| 231 |
\cite edgecollapsearxiv. |
| 232 |
</td> |
| 233 |
<td width="15%"> |
| 234 |
<b>Author:</b> Siddharth Pritam, Marc Glisse<br> |
| 235 |
<b>Introduced in:</b> GUDHI 3.3.0<br> |
| 236 |
<b>License:</b> MIT |
| 237 |
</td> |
| 238 |
</tr> |
| 239 |
<tr> |
| 240 |
<td colspan=2 height="25"> |
| 241 |
<b>User manual:</b> \ref edge_collapse |
| 242 |
</td> |
| 243 |
</tr> |
| 244 |
</table> |
| 245 |
|
| 246 |
### Witness complex |
| 247 |
|
| 248 |
<table> |
| 249 |
<tr> |
| 250 |
<td width="35%" rowspan=2> |
| 251 |
\image html "Witness_complex_representation.png" |
| 252 |
</td> |
| 253 |
<td width="50%"> |
| 254 |
Witness complex \f$ Wit(W,L) \f$ is a simplicial complex defined on two sets of points in \f$\mathbb{R}^D\f$. |
| 255 |
The data structure is described in \cite boissonnatmariasimplextreealgorithmica . |
| 256 |
</td> |
| 257 |
<td width="15%"> |
| 258 |
<b>Author:</b> Siargey Kachanovich<br> |
| 259 |
<b>Introduced in:</b> GUDHI 1.3.0<br> |
| 260 |
<b>License:</b> MIT ([GPL v3](../../licensing/) for Euclidean version)<br> |
| 261 |
<b>Euclidean version requires:</b> \ref eigen and \ref cgal |
| 262 |
</td> |
| 263 |
</tr> |
| 264 |
<tr> |
| 265 |
<td colspan=2 height="25"> |
| 266 |
<b>User manual:</b> \ref witness_complex |
| 267 |
</td> |
| 268 |
</tr> |
| 269 |
</table> |
| 270 |
|
| 271 |
### Cover Complexes |
| 272 |
<table> |
| 273 |
<tr> |
| 274 |
<td width="35%" rowspan=2> |
| 275 |
\image html "gicvisu.jpg" |
| 276 |
</td> |
| 277 |
<td width="50%"> |
| 278 |
Nerves and Graph Induced Complexes are cover complexes, i.e. simplicial complexes that provably contain |
| 279 |
topological information about the input data. They can be computed with a cover of the |
| 280 |
data, that comes i.e. from the preimage of a family of intervals covering the image |
| 281 |
of a scalar-valued function defined on the data. <br> |
| 282 |
</td> |
| 283 |
<td width="15%"> |
| 284 |
<b>Author:</b> Mathieu Carrière<br> |
| 285 |
<b>Introduced in:</b> GUDHI 2.1.0<br> |
| 286 |
<b>License:</b> MIT [(GPL v3)](../../licensing/)<br> |
| 287 |
<b>Requires:</b> \ref cgal |
| 288 |
</td> |
| 289 |
</tr> |
| 290 |
<tr> |
| 291 |
<td colspan=2 height="25"> |
| 292 |
<b>User manual:</b> \ref cover_complex |
| 293 |
</td> |
| 294 |
</tr> |
| 295 |
</table> |
| 296 |
|
| 297 |
## Manifold reconstructions |
| 298 |
### Coxeter triangulation |
| 299 |
|
| 300 |
<table> |
| 301 |
<tr> |
| 302 |
<td width="35%" rowspan=2> |
| 303 |
\image html "manifold_tracing_on_custom_function_example.png" |
| 304 |
</td> |
| 305 |
<td width="50%"> |
| 306 |
Coxeter triangulation module is designed to provide tools for constructing a piecewise-linear approximation of an |
| 307 |
\f$m\f$-dimensional smooth manifold embedded in \f$ \mathbb{R}^d \f$ using an ambient triangulation. |
| 308 |
</td> |
| 309 |
<td width="15%"> |
| 310 |
<b>Author:</b> Siargey Kachanovich<br> |
| 311 |
<b>Introduced in:</b> GUDHI 3.4.0<br> |
| 312 |
<b>License:</b> MIT [(LGPL v3)](../../licensing/)<br> |
| 313 |
<b>Requires:</b> \ref eigen |
| 314 |
</td> |
| 315 |
</tr> |
| 316 |
<tr> |
| 317 |
<td colspan=2 height="25"> |
| 318 |
<b>User manual:</b> \ref coxeter_triangulation |
| 319 |
</td> |
| 320 |
</tr> |
| 321 |
</table> |
| 322 |
|
| 323 |
### Tangential complex |
| 324 |
|
| 325 |
<table> |
| 326 |
<tr> |
| 327 |
<td width="35%" rowspan=2> |
| 328 |
\image html "tc_examples.png" |
| 329 |
</td> |
| 330 |
<td width="50%"> |
| 331 |
A Tangential Delaunay complex is a <a target="_blank" href="https://en.wikipedia.org/wiki/Simplicial_complex">simplicial complex</a> |
| 332 |
designed to reconstruct a \f$ k \f$-dimensional manifold embedded in \f$ d \f$-dimensional Euclidean space. |
| 333 |
The input is a point sample coming from an unknown manifold. |
| 334 |
The running time depends only linearly on the extrinsic dimension \f$ d \f$ |
| 335 |
and exponentially on the intrinsic dimension \f$ k \f$. |
| 336 |
</td> |
| 337 |
<td width="15%"> |
| 338 |
<b>Author:</b> Clément Jamin<br> |
| 339 |
<b>Introduced in:</b> GUDHI 2.0.0<br> |
| 340 |
<b>License:</b> MIT [(GPL v3)](../../licensing/)<br> |
| 341 |
<b>Requires:</b> \ref eigen and \ref cgal |
| 342 |
</td> |
| 343 |
</tr> |
| 344 |
<tr> |
| 345 |
<td colspan=2 height="25"> |
| 346 |
<b>User manual:</b> \ref tangential_complex |
| 347 |
</td> |
| 348 |
</tr> |
| 349 |
</table> |
| 350 |
|
| 351 |
## Topological descriptors computation {#TopologicalDescriptorsComputation} |
| 352 |
|
| 353 |
### Persistent Cohomology |
| 354 |
|
| 355 |
<table> |
| 356 |
<tr> |
| 357 |
<td width="35%" rowspan=2> |
| 358 |
\image html "plain_homology.png" |
| 359 |
</td> |
| 360 |
<td width="50%"> |
| 361 |
The theory of homology consists in attaching to a topological space a sequence of (homology) groups, capturing |
| 362 |
global topological features like connected components, holes, cavities, etc. Persistent homology studies the |
| 363 |
evolution -- birth, life and death -- of these features when the topological space is changing. Consequently, the |
| 364 |
theory is essentially composed of three elements: topological spaces, their homology groups and an evolution |
| 365 |
scheme. |
| 366 |
Computation of persistent cohomology using the algorithm of \cite DBLP:journals/dcg/SilvaMV11 and |
| 367 |
\cite DBLP:conf/compgeom/DeyFW14 and the Compressed Annotation Matrix implementation of |
| 368 |
\cite DBLP:conf/esa/BoissonnatDM13 . |
| 369 |
</td> |
| 370 |
<td width="15%"> |
| 371 |
<b>Author:</b> Clément Maria<br> |
| 372 |
<b>Introduced in:</b> GUDHI 1.0.0<br> |
| 373 |
<b>License:</b> MIT<br> |
| 374 |
</td> |
| 375 |
</tr> |
| 376 |
<tr> |
| 377 |
<td colspan=2 height="25"> |
| 378 |
<b>User manual:</b> \ref persistent_cohomology |
| 379 |
</td> |
| 380 |
</tr> |
| 381 |
</table> |
| 382 |
|
| 383 |
### Persistence Matrix |
| 384 |
|
| 385 |
<table> |
| 386 |
<tr> |
| 387 |
<td width="35%" rowspan=2> |
| 388 |
\image html "persistence_matrix.png" |
| 389 |
</td> |
| 390 |
<td width="50%"> |
| 391 |
Matrix structure for filtered complexes with multiple functionnalities related to persistence homology, |
| 392 |
such as representative cycles computation or vineyards. |
| 393 |
<dl class="section note"> |
| 394 |
<dt>Note</dt> |
| 395 |
<dd>Matrix API is in a beta version and may change in incompatible ways in the near future.</dd> |
| 396 |
</dl> |
| 397 |
|
| 398 |
</td> |
| 399 |
<td width="15%"> |
| 400 |
<b>Author:</b> Hannah Schreiber<br> |
| 401 |
<b>Introduced in:</b> GUDHI 3.10.0<br> |
| 402 |
<b>Copyright:</b> MIT<br> |
| 403 |
</td> |
| 404 |
</tr> |
| 405 |
<tr> |
| 406 |
<td colspan=2 height="25"> |
| 407 |
<b>User manual:</b> \ref persistence_matrix and \ref persistence_fields |
| 408 |
</td> |
| 409 |
</tr> |
| 410 |
</table> |
| 411 |
|
| 412 |
### Zigzag Persistent Homology |
| 413 |
|
| 414 |
<table> |
| 415 |
<tr> |
| 416 |
<td width="35%" rowspan=2> |
| 417 |
\image html "zigzag_ex.png" |
| 418 |
</td> |
| 419 |
<td width="50%"> |
| 420 |
Zigzag filtrations are a generalization of (standard) filtrations: the inclusion maps can not only go forward, |
| 421 |
but also backward. In other words, simplices can also be removed from the complex which is build. |
| 422 |
The implemented algorithm is from \cite zigzag. |
| 423 |
</td> |
| 424 |
<td width="15%"> |
| 425 |
<b>Author:</b> Clément Maria, Hannah Schreiber<br> |
| 426 |
<b>Introduced in:</b> GUDHI 3.11.0<br> |
| 427 |
<b>Copyright:</b> MIT<br> |
| 428 |
</td> |
| 429 |
</tr> |
| 430 |
<tr> |
| 431 |
<td colspan=2 height="25"> |
| 432 |
<b>User manual:</b> \ref zigzag_persistence |
| 433 |
</td> |
| 434 |
</tr> |
| 435 |
</table> |
| 436 |
|
| 437 |
## Topological descriptors tools {#TopologicalDescriptorsTools} |
| 438 |
|
| 439 |
### Bottleneck distance |
| 440 |
|
| 441 |
<table> |
| 442 |
<tr> |
| 443 |
<td width="35%" rowspan=2> |
| 444 |
\image html "perturb_pd.png" |
| 445 |
</td> |
| 446 |
<td width="50%"> |
| 447 |
Bottleneck distance measures the similarity between two persistence diagrams. |
| 448 |
It's the shortest distance b for which there exists a perfect matching between |
| 449 |
the points of the two diagrams (+ all the diagonal points) such that |
| 450 |
any couple of matched points are at distance at most b, |
| 451 |
where the distance between points is the sup norm in \f$\mathbb{R}^2\f$ |
| 452 |
(not the Euclidean distance). |
| 453 |
</td> |
| 454 |
<td width="15%"> |
| 455 |
<b>Author:</b> François Godi<br> |
| 456 |
<b>Introduced in:</b> GUDHI 2.0.0<br> |
| 457 |
<b>License:</b> MIT [(GPL v3)](../../licensing/)<br> |
| 458 |
<b>Requires:</b> \ref cgal |
| 459 |
</td> |
| 460 |
</tr> |
| 461 |
<tr> |
| 462 |
<td colspan=2 height="25"> |
| 463 |
<b>User manual:</b> \ref bottleneck_distance |
| 464 |
</td> |
| 465 |
</tr> |
| 466 |
</table> |
| 467 |
|
| 468 |
### Persistence representations |
| 469 |
|
| 470 |
<table> |
| 471 |
<tr> |
| 472 |
<td width="35%" rowspan=2> |
| 473 |
\image html "average_landscape.png" |
| 474 |
</td> |
| 475 |
<td width="50%"> |
| 476 |
It contains implementation of various representations of persistence diagrams; diagrams themselves, persistence |
| 477 |
landscapes (rigorous and grid version), persistence heat maps, vectors and others. It implements basic |
| 478 |
functionalities which are necessary to use persistence in statistics and machine learning. |
| 479 |
</td> |
| 480 |
<td width="15%"> |
| 481 |
<b>Author:</b> Pawel Dlotko<br> |
| 482 |
<b>Introduced in:</b> GUDHI 2.1.0<br> |
| 483 |
<b>License:</b> MIT<br> |
| 484 |
</td> |
| 485 |
</tr> |
| 486 |
<tr> |
| 487 |
<td colspan=2 height="25"> |
| 488 |
<b>User manual:</b> \ref Persistence_representations |
| 489 |
</td> |
| 490 |
</tr> |
| 491 |
</table> |
| 492 |
|
| 493 |
## Point cloud utilities {#PointCloudUtils} |
| 494 |
|
| 495 |
<table> |
| 496 |
<tr> |
| 497 |
<td width="35%" rowspan=2> |
| 498 |
\f$(x_1,\ldots,x_d)\f$ |
| 499 |
</td> |
| 500 |
<td width="50%"> |
| 501 |
This contains various tools to handle point clouds: spatial searching, subsampling, etc. |
| 502 |
</td> |
| 503 |
<td width="15%"> |
| 504 |
<b>Author:</b>Clément Jamin, Siargey Kachanovich, Marc Glisse<br> |
| 505 |
<b>Introduced in:</b> GUDHI 1.3.0<br> |
| 506 |
<b>License:</b> MIT [(GPL v3)](../../licensing/)<br> |
| 507 |
</td> |
| 508 |
</tr> |
| 509 |
<tr> |
| 510 |
<td colspan=2 height="25"> |
| 511 |
<b>Manuals:</b> \ref spatial_searching, \ref subsampling |
| 512 |
</td> |
| 513 |
</tr> |
| 514 |
</table> |