diff options
Diffstat (limited to 'src/Collapse/include')
-rw-r--r-- | src/Collapse/include/gudhi/FlagComplexSpMatrix.h | 852 | ||||
-rw-r--r-- | src/Collapse/include/gudhi/Rips_edge_list.h | 184 |
2 files changed, 1036 insertions, 0 deletions
diff --git a/src/Collapse/include/gudhi/FlagComplexSpMatrix.h b/src/Collapse/include/gudhi/FlagComplexSpMatrix.h new file mode 100644 index 00000000..ac02a46f --- /dev/null +++ b/src/Collapse/include/gudhi/FlagComplexSpMatrix.h @@ -0,0 +1,852 @@ +/* This file is part of the Gudhi Library. The Gudhi library + * (Geometric Understanding in Higher Dimensions) is a generic C++ + * library for computational topology. + * + * Author(s): Siddharth Pritam + * + * Copyright (C) 2018 INRIA Sophia Antipolis (France) + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * 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 General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see <http://www.gnu.org/licenses/>. + +*/ +#pragma once + +#include <gudhi/Rips_edge_list.h> +#include <boost/functional/hash.hpp> +// #include <boost/graph/adjacency_list.hpp> + +#include <iostream> +#include <utility> +#include <vector> +#include <queue> +#include <unordered_map> +#include <tuple> +#include <list> +#include <algorithm> +#include <chrono> + +#include <ctime> +#include <fstream> + +#include <Eigen/Sparse> + +typedef std::size_t Vertex; +using Edge = std::pair<Vertex, Vertex>; // This is an ordered pair, An edge is stored with convention of the first + // element being the smaller i.e {2,3} not {3,2}. However this is at the level + // of row indices on actual vertex lables +using EdgeFilt = std::pair<Edge, double>; +using edge_list = std::vector<Edge>; + +using MapVertexToIndex = std::unordered_map<Vertex, std::size_t>; +using Map = std::unordered_map<Vertex, Vertex>; + +using sparseRowMatrix = Eigen::SparseMatrix<double, Eigen::RowMajor>; +using rowInnerIterator = sparseRowMatrix::InnerIterator; + +using doubleVector = std::vector<double>; +using vertexVector = std::vector<Vertex>; +using boolVector = std::vector<bool>; + +using doubleQueue = std::queue<double>; + +using EdgeFiltQueue = std::queue<EdgeFilt>; +using EdgeFiltVector = std::vector<EdgeFilt>; + +typedef std::vector<std::tuple<double, Vertex, Vertex>> Filtered_sorted_edge_list; +typedef std::unordered_map<Edge, bool, boost::hash<Edge>> u_edge_map; +typedef std::unordered_map<Edge, std::size_t, boost::hash<Edge>> u_edge_to_idx_map; + +//! Class SparseMsMatrix +/*! + The class for storing the Vertices v/s MaxSimplices Sparse Matrix and performing collapses operations using the N^2() + Algorithm. +*/ +class FlagComplexSpMatrix { + private: + std::unordered_map<int, Vertex> rowToVertex; + + // Vertices strored as an unordered_set + std::unordered_set<Vertex> vertices; + //! Stores the 1-simplices(edges) of the original Simplicial Complex. + edge_list oneSimplices; + + // Unordered set of removed edges. (to enforce removal from the matrix) + std::unordered_set<Edge, boost::hash<Edge>> u_set_removed_redges; + + // Unordered set of dominated edges. (to inforce removal from the matrix) + std::unordered_set<Edge, boost::hash<Edge>> u_set_dominated_redges; + + // Map from egde to its index + u_edge_to_idx_map edge_to_index_map; + // Boolean vector to indicate if the index is critical or not. + boolVector critical_edge_indicator; // critical indicator + + // Boolean vector to indicate if the index is critical or not. + boolVector dominated_edge_indicator; // domination indicator + + //! Stores the Map between vertices<B>rowToVertex and row indices <B>rowToVertex -> row-index</B>. + /*! + \code + MapVertexToIndex = std::unordered_map<Vertex,int> + \endcode + So, if the original simplex tree had vertices 0,1,4,5 <br> + <B>rowToVertex</B> would store : <br> + \verbatim + Values = | 0 | 1 | 4 | 5 | + Indices = 0 1 2 3 + \endverbatim + And <B>vertexToRow</B> would be a map like the following : <br> + \verbatim + 0 -> 0 + 1 -> 1 + 4 -> 2 + 5 -> 3 + \endverbatim + */ + MapVertexToIndex vertexToRow; + + //! Stores the number of vertices in the original Simplicial Complex. + /*! + This stores the count of vertices (which is also the number of rows in the Matrix). + */ + std::size_t rows; + + std::size_t numOneSimplices; + + std::size_t numDomEdge; + + //! Stores the Sparse matrix of double values representing the Original Simplicial Complex. + /*! + \code + sparseRowMatrix = Eigen::SparseMatrix<double, Eigen::RowMajor> ; + \endcode + ; + */ + + sparseRowMatrix* sparse_colpsd_adj_Matrix; // Stores the collapsed sparse matrix representaion. + sparseRowMatrix sparseRowAdjMatrix; // This is row-major version of the same sparse-matrix, to facilitate easy access + // to elements when traversing the matrix row-wise. + + //! Stores <I>true</I> for dominated rows and <I>false</I> for undominated rows. + /*! + Initialised to a vector of length equal to the value of the variable <B>rows</B> with all <I>false</I> values. + Subsequent removal of dominated vertices is reflected by concerned entries changing to <I>true</I> in this vector. + */ + boolVector vertDomnIndicator; //(domination indicator) + + boolVector contractionIndicator; //(contraction indicator) + + //! Stores the indices of the rows to-be checked for domination in the current iteration. + /*! + Initialised with all rows for the first iteration. + Subsequently once a dominated row is found, its non-dominated neighbhour indices are inserted. + */ + // doubleQueue rowIterator; + + doubleQueue rowIterator; + + // Queue of filtered edges, for edge-collapse, the indices of the edges are the row-indices. + EdgeFiltQueue filteredEgdeIter; + + // Vector of filtered edges, for edge-collapse, the indices of the edges are the row-indices. + EdgeFiltVector fEgdeVector; + + // List of non-dominated edges, the indices of the edges are the vertex lables!!. + Filtered_sorted_edge_list criticalCoreEdges; + // Stores the indices from the sorted filtered edge vector. + // std::set<std::size_t> recurCriticalCoreIndcs; + + //! Stores <I>true</I> if the current row is inserted in the queue <B>rowIterator<B> otherwise its value is + //! <I>false<I>. + /*! + Initialised to a boolean vector of length equal to the value of the variable <B>rows</B> with all <I>true</I> + values. Subsequent removal/addition of a row from <B>rowIterator<B> is reflected by concerned entries changing to + <I>false</I>/<I>true</I> in this vector. + */ + boolVector rowInsertIndicator; //(current iteration row insertion indicator) + + //! Map that stores the Reduction / Collapse of vertices. + /*! + \code + Map = std::unordered_map<Vertex,Vertex> + \endcode + This is empty to begin with. As and when collapses are done (let's say from dominated vertex <I>v</I> to dominating + vertex <I>v'</I>) : <br> <B>ReductionMap</B>[<I>v</I>] = <I>v'</I> is entered into the map. <br> <I>This does not + store uncollapsed vertices. What it means is that say vertex <I>x</I> was never collapsed onto any other vertex. + Then, this map <B>WILL NOT</B> have any entry like <I>x</I> -> <I>x</I>. Basically, it will have no entry + corresponding to vertex <I>x</I> at all. </I> + */ + Map ReductionMap; + + bool vertexCollapsed; + bool edgeCollapsed; + // Variable to indicate if filtered-edge-collapse has to be performed. + int expansion_limit; + + void init() { + rowToVertex.clear(); + vertexToRow.clear(); + oneSimplices.clear(); + ReductionMap.clear(); + + vertDomnIndicator.clear(); + rowInsertIndicator.clear(); + rowIterator.push(0); + rowIterator.pop(); + + filteredEgdeIter.push({{0, 0}, 0}); + filteredEgdeIter.pop(); + fEgdeVector.clear(); + + rows = 0; + numDomEdge = 0; + + numOneSimplices = 0; + expansion_limit = 2; + + vertexCollapsed = false; + edgeCollapsed = false; + } + + //! Function for computing the sparse-matrix corresponding to the core of the complex. It also prepares the working + //!list filteredEgdeIter for edge collapses + void after_vertex_collapse() { + sparse_colpsd_adj_Matrix = new sparseRowMatrix(rows, rows); // Just for debugging purpose. + oneSimplices.clear(); + if (not filteredEgdeIter.empty()) { + std::cout << "Working list for edge collapses are not empty before the edge-collapse." << std::endl; + } + for (std::size_t rw = 0; rw < rows; ++rw) { + if (not vertDomnIndicator[rw]) // If the current column is not dominated + { + auto nbhrs_to_insert = closed_neighbours_row_index(rw); // returns row indices of the non-dominated vertices. + for (auto& v : nbhrs_to_insert) { + sparse_colpsd_adj_Matrix->insert(rw, v) = 1; // This creates the full matrix + if (rw < v) { + oneSimplices.push_back({rowToVertex[rw], rowToVertex[v]}); + filteredEgdeIter.push({{rw, v}, 1}); + // if(rw == v) + // std::cout << "Pushed the edge {" << rw << ", " << v << "} " << std::endl; + } + } + } + } + // std::cout << "Total number of non-zero elements before domination check are: " << + // sparse_colpsd_adj_Matrix->nonZeros() << std::endl; std::cout << "Total number of edges for domination check are: + // " << filteredEgdeIter.size() << std::endl; + // std::cout << *sparse_colpsd_adj_Matrix << std::endl; + return; + } + + //! Function to fully compact a particular vertex of the ReductionMap. + /*! + It takes as argument the iterator corresponding to a particular vertex pair (key-value) stored in the ReductionMap. + <br> It then checks if the second element of this particular vertex pair is present as a first element of some other + key-value pair in the map. If no, then the first element of the vertex pair in consideration is fully compact. If + yes, then recursively call fully_compact_this_vertex() on the second element of the original pair in consideration + and assign its resultant image as the image of the first element of the original pair in consideration as well. + */ + void fully_compact_this_vertex(Map::iterator iter) { + Map::iterator found = ReductionMap.find(iter->second); + if (found == ReductionMap.end()) return; + + fully_compact_this_vertex(found); + iter->second = ReductionMap[iter->second]; + } + + //! Function to fully compact the Reduction Map. + /*! + While doing strong collapses, we store only the immediate collapse of a vertex. Which means that in one round, + vertex <I>x</I> may collapse to vertex <I>y</I>. And in some later round it may be possible that vertex <I>y</I> + collapses to <I>z</I>. In which case our map stores : <br> <I>x</I> -> <I>y</I> and also <I>y</I> -> <I>z</I>. But + it really should store : <I>x</I> -> <I>z</I> and <I>y</I> -> <I>z</I>. This function achieves the same. <br> It + basically calls fully_compact_this_vertex() for each entry in the map. + */ + void fully_compact() { + Map::iterator it = ReductionMap.begin(); + while (it != ReductionMap.end()) { + fully_compact_this_vertex(it); + it++; + } + } + + void sparse_strong_vertex_collapse() { + complete_vertex_domination_check( + rowIterator, rowInsertIndicator, + vertDomnIndicator); // Complete check for rows in rowIterator, rowInsertIndicator is a list of boolean + // indicator if a vertex is already inserted in the working row_queue (rowIterator) + if (not rowIterator.empty()) + sparse_strong_vertex_collapse(); + else + return; + } + + void complete_vertex_domination_check(doubleQueue& iterator, boolVector& insertIndicator, boolVector& domnIndicator) { + double k; + doubleVector nonZeroInnerIdcs; + while (not iterator.empty()) // "iterator" contains list(FIFO) of rows to be considered for domination check + { + k = iterator.front(); + iterator.pop(); + insertIndicator[k] = false; + if (not domnIndicator[k]) // Check if is already dominated + { + nonZeroInnerIdcs = closed_neighbours_row_index(k); + for (doubleVector::iterator it = nonZeroInnerIdcs.begin(); it != nonZeroInnerIdcs.end(); it++) { + int checkDom = vertex_domination_check(k, *it); // "true" for row domination comparison + if (checkDom == 1) // row k is dominated by *it, k <= *it; + { + setZero(k, *it); + break; + } else if (checkDom == -1) // row *it is dominated by k, *it <= k; + setZero(*it, k); + } + } + } + } + + bool check_edge_domination(Edge e) // Edge e is the actual edge (u,v). Not the row ids in the matrixs + { + auto u = std::get<0>(e); + auto v = std::get<1>(e); + + auto rw_u = vertexToRow[u]; + auto rw_v = vertexToRow[v]; + auto rw_e = std::make_pair(rw_u, rw_v); + // std::cout << "The edge {" << u << ", " << v << "} is going for domination check." << std::endl; + auto commonNeighbours = closed_common_neighbours_row_index(rw_e); + // std::cout << "And its common neighbours are." << std::endl; + // for (doubleVector::iterator it = commonNeighbours.begin(); it!=commonNeighbours.end(); it++) { + // std::cout << rowToVertex[*it] << ", " ; + // } + // std::cout<< std::endl; + if (commonNeighbours.size() > 2) { + if (commonNeighbours.size() == 3) + return true; + else + for (doubleVector::iterator it = commonNeighbours.begin(); it != commonNeighbours.end(); it++) { + auto rw_c = *it; // Typecasting + if (rw_c != rw_u and rw_c != rw_v) { + auto neighbours_c = closed_neighbours_row_index(rw_c); + if (std::includes(neighbours_c.begin(), neighbours_c.end(), commonNeighbours.begin(), + commonNeighbours.end())) // If neighbours_c contains the common neighbours. + return true; + } + } + } + return false; + } + + bool check_domination_indicator(Edge e) // The edge should be sorted by the indices and indices are original + { + return dominated_edge_indicator[edge_to_index_map[e]]; + } + + std::set<std::size_t> three_clique_indices(std::size_t crit) { + std::set<std::size_t> edge_indices; + + EdgeFilt fe = fEgdeVector.at(crit); + Edge e = std::get<0>(fe); + Vertex u = std::get<0>(e); + Vertex v = std::get<1>(e); + + // std::cout << "The current critical edge to re-check criticality with filt value is : {" << u << "," << v << "}; + // "<< std::get<1>(fe) << std::endl; + auto rw_u = vertexToRow[u]; + auto rw_v = vertexToRow[v]; + auto rw_critical_edge = std::make_pair(rw_u, rw_v); + + doubleVector commonNeighbours = closed_common_neighbours_row_index(rw_critical_edge); + + if (commonNeighbours.size() > 2) { + for (doubleVector::iterator it = commonNeighbours.begin(); it != commonNeighbours.end(); it++) { + auto rw_c = *it; + if (rw_c != rw_u and rw_c != rw_v) { + auto e_with_new_nbhr_v = std::minmax(u, rowToVertex[rw_c]); + auto e_with_new_nbhr_u = std::minmax(v, rowToVertex[rw_c]); + edge_indices.emplace(edge_to_index_map[e_with_new_nbhr_v]); + edge_indices.emplace(edge_to_index_map[e_with_new_nbhr_u]); + } + } + } + return edge_indices; + } + + void set_edge_critical(std::size_t indx, double filt) { + // std::cout << "The curent index with filtration value " << indx << ", " << filt << " is primary critical" << + // std::endl; + std::set<std::size_t> effectedIndcs = three_clique_indices(indx); + if (effectedIndcs.size() > 0) { + for (auto idx = indx - 1; idx > 0; idx--) { + EdgeFilt fec = fEgdeVector.at(idx); + Edge e = std::get<0>(fec); + Vertex u = std::get<0>(e); + Vertex v = std::get<1>(e); + if (not critical_edge_indicator.at( + idx)) { // If idx is not critical so it should be proceses, otherwise it stays in the graph // prev + // code : recurCriticalCoreIndcs.find(idx) == recurCriticalCoreIndcs.end() + if (effectedIndcs.find(idx) != effectedIndcs.end()) { // If idx is affected + if (not check_edge_domination(e)) { + // std::cout << "The curent index is became critical " << idx << std::endl; + critical_edge_indicator.at(idx) = true; + criticalCoreEdges.push_back({filt, u, v}); + std::set<std::size_t> inner_effected_indcs = three_clique_indices(idx); + for (auto inr_idx = inner_effected_indcs.rbegin(); inr_idx != inner_effected_indcs.rend(); inr_idx++) { + if (*inr_idx < idx) effectedIndcs.emplace(*inr_idx); + } + inner_effected_indcs.clear(); + // std::cout << "The following edge is critical with filt value: {" << std::get<0>(e) << "," << + // std::get<1>(e) << "}; " << filt << std::endl; + } else + u_set_dominated_redges.emplace(std::minmax(vertexToRow[u], vertexToRow[v])); + } else // Idx is not affected hence dominated. + u_set_dominated_redges.emplace(std::minmax(vertexToRow[u], vertexToRow[v])); + } + } + } + effectedIndcs.clear(); + u_set_dominated_redges.clear(); + } + + void critical_core_edges() { + std::size_t totEdges = fEgdeVector.size(); + + std::size_t endIdx = 0; + + u_set_removed_redges.clear(); + u_set_dominated_redges.clear(); + critical_edge_indicator.clear(); + + while (endIdx < totEdges) { + EdgeFilt fec = fEgdeVector.at(endIdx); + + insert_new_edges(std::get<0>(std::get<0>(fec)), std::get<1>(std::get<0>(fec)), + std::get<1>(fec)); // Inserts the edge in the sparse matrix to update the graph (G_i) + + Edge e = std::get<0>(fec); + Vertex u = std::get<0>(e); + Vertex v = std::get<1>(e); + edge_to_index_map.emplace(std::minmax(u, v), endIdx); + critical_edge_indicator.push_back(false); + dominated_edge_indicator.push_back(false); + + if (not check_edge_domination(e)) { + critical_edge_indicator.at(endIdx) = true; + dominated_edge_indicator.at(endIdx) = false; + criticalCoreEdges.push_back({std::get<1>(fec), u, v}); + if (endIdx > 1) set_edge_critical(endIdx, std::get<1>(fec)); + + } else + dominated_edge_indicator.at(endIdx) = true; + endIdx++; + } + + std::cout << "The total number of critical edges is: " << criticalCoreEdges.size() << std::endl; + std::cout << "The total number of non-critical edges is: " << totEdges - criticalCoreEdges.size() << std::endl; + } + + int vertex_domination_check(double i, double j) // True for row comparison, false for column comparison + { + if (i != j) { + doubleVector Listi = closed_neighbours_row_index(i); + doubleVector Listj = closed_neighbours_row_index(j); + if (Listj.size() <= Listi.size()) { + if (std::includes(Listi.begin(), Listi.end(), Listj.begin(), Listj.end())) // Listj is a subset of Listi + return -1; + } + + else if (std::includes(Listj.begin(), Listj.end(), Listi.begin(), Listi.end())) // Listi is a subset of Listj + return 1; + } + return 0; + } + + doubleVector closed_neighbours_row_index(double indx) // Returns list of non-zero columns of the particular indx. + { + doubleVector nonZeroIndices; + Vertex u = indx; + Vertex v; + // std::cout << "The neighbours of the vertex: " << rowToVertex[u] << " are. " << std::endl; + if (not vertDomnIndicator[indx]) { + for (rowInnerIterator it(sparseRowAdjMatrix, indx); it; ++it) { // Iterate over the non-zero columns + v = it.index(); + if (not vertDomnIndicator[v] and u_set_removed_redges.find(std::minmax(u, v)) == u_set_removed_redges.end() and + u_set_dominated_redges.find(std::minmax(u, v)) == + u_set_dominated_redges + .end()) { // If the vertex v is not dominated and the edge {u,v} is still in the matrix + nonZeroIndices.push_back(it.index()); // inner index, here it is equal to it.columns() + // std::cout << rowToVertex[it.index()] << ", " ; + } + } + // std::cout << std::endl; + } + return nonZeroIndices; + } + + doubleVector closed_common_neighbours_row_index(Edge e) // Returns the list of closed neighbours of the edge :{u,v}. + { + doubleVector common; + doubleVector nonZeroIndices_u; + doubleVector nonZeroIndices_v; + double u = std::get<0>(e); + double v = std::get<1>(e); + + nonZeroIndices_u = closed_neighbours_row_index(u); + nonZeroIndices_v = closed_neighbours_row_index(v); + std::set_intersection(nonZeroIndices_u.begin(), nonZeroIndices_u.end(), nonZeroIndices_v.begin(), + nonZeroIndices_v.end(), std::inserter(common, common.begin())); + + return common; + } + + void setZero(double dominated, double dominating) { + for (auto& v : closed_neighbours_row_index(dominated)) + if (not rowInsertIndicator[v]) // Checking if the row is already inserted + { + rowIterator.push(v); + rowInsertIndicator[v] = true; + } + vertDomnIndicator[dominated] = true; + ReductionMap[rowToVertex[dominated]] = rowToVertex[dominating]; + + vertexToRow.erase(rowToVertex[dominated]); + vertices.erase(rowToVertex[dominated]); + rowToVertex.erase(dominated); + } + + vertexVector closed_neighbours_vertex_index(double rowIndx) // Returns list of non-zero "vertices" of the particular + // colIndx. the difference is in the return type + { + vertexVector colmns; + for (auto& v : closed_neighbours_row_index(rowIndx)) // Iterate over the non-zero columns + colmns.push_back(rowToVertex[v]); + std::sort(colmns.begin(), colmns.end()); + return colmns; + } + + vertexVector vertex_closed_active_neighbours( + double rowIndx) // Returns list of all non-zero "vertices" of the particular colIndx which are currently active. + // the difference is in the return type. + { + vertexVector colmns; + for (auto& v : closed_neighbours_row_index(rowIndx)) // Iterate over the non-zero columns + if (not contractionIndicator[v]) // Check if the row corresponds to a contracted vertex + colmns.push_back(rowToVertex[v]); + std::sort(colmns.begin(), colmns.end()); + return colmns; + } + + vertexVector closed_all_neighbours_row_index( + double rowIndx) // Returns list of all non-zero "vertices" of the particular colIndx whether dominated or not. + // the difference is in the return type. + { + vertexVector colmns; + for (rowInnerIterator itCol(sparseRowAdjMatrix, rowIndx); itCol; ++itCol) // Iterate over the non-zero columns + colmns.push_back(rowToVertex[itCol.index()]); // inner index, here it is equal to it.row() + std::sort(colmns.begin(), colmns.end()); + return colmns; + } + + void swap_rows(const Vertex& v, + const Vertex& w) { // swap the rows of v and w. Both should be members of the skeleton + if (membership(v) && membership(w)) { + auto rw_v = vertexToRow[v]; + auto rw_w = vertexToRow[w]; + vertexToRow[v] = rw_w; + vertexToRow[w] = rw_v; + rowToVertex[rw_v] = w; + rowToVertex[rw_w] = v; + } + } + + public: + //! Default Constructor + /*! + Only initialises all Data Members of the class to empty/Null values as appropriate. + One <I>WILL</I> have to create the matrix using the Constructor that has an object of the Simplex_tree class as + argument. + */ + + FlagComplexSpMatrix() { init(); } + + FlagComplexSpMatrix(std::size_t expRows) { + init(); + sparseRowAdjMatrix = sparseRowMatrix( + expansion_limit * expRows, + expansion_limit * expRows); // Initializing sparseRowAdjMatrix, This is a row-major sparse matrix. + } + + //! Main Constructor + /*! + Argument is an instance of Filtered_sorted_edge_list. <br> + This is THE function that initialises all data members to appropriate values. <br> + <B>rowToVertex</B>, <B>vertexToRow</B>, <B>rows</B>, <B>cols</B>, <B>sparseRowAdjMatrix</B> are initialised here. + <B>vertDomnIndicator</B>, <B>rowInsertIndicator</B> ,<B>rowIterator<B> are initialised by init() function which is + called at the begining of this. <br> + */ + FlagComplexSpMatrix(const size_t& num_vertices, const Filtered_sorted_edge_list& edge_t) { + init(); + sparseRowAdjMatrix = sparseRowMatrix( + num_vertices, num_vertices); // Initializing sparseRowAdjMatrix, This is a row-major sparse matrix. + + for (size_t bgn_idx = 0; bgn_idx < edge_t.size(); bgn_idx++) { + // std::vector<size_t> s = {std::get<1>(edge_t.at(bgn_idx)), std::get<2>(edge_t.at(bgn_idx))}; + // insert_new_edges(std::get<1>(edge_t.at(bgn_idx)), std::get<2>(edge_t.at(bgn_idx)), 1); + fEgdeVector.push_back( + {{std::get<1>(edge_t.at(bgn_idx)), std::get<2>(edge_t.at(bgn_idx))}, std::get<0>(edge_t.at(bgn_idx))}); + } + + // sparseRowAdjMatrix.makeCompressed(); + } + + //! Destructor. + /*! + Frees up memory locations on the heap. + */ + ~FlagComplexSpMatrix() {} + + //! Function for performing strong collapse. + /*! + calls sparse_strong_vertex_collapse(), and + Then, it compacts the ReductionMap by calling the function fully_compact(). + */ + double strong_vertex_collapse() { + auto begin_collapse = std::chrono::high_resolution_clock::now(); + sparse_strong_vertex_collapse(); + vertexCollapsed = true; + auto end_collapse = std::chrono::high_resolution_clock::now(); + + auto collapseTime = std::chrono::duration<double, std::milli>(end_collapse - begin_collapse).count(); + // std::cout << "Time of Collapse : " << collapseTime << " ms\n" << std::endl; + + // Now we complete the Reduction Map + fully_compact(); + // Post processing... + after_vertex_collapse(); + return collapseTime; + } + + // Performs edge collapse in a decreasing sequence of the filtration value. + Filtered_sorted_edge_list filtered_edge_collapse() { + critical_core_edges(); + vertexCollapsed = false; + edgeCollapsed = true; + return criticalCoreEdges; + } + + // double strong_vertex_edge_collapse() { + // auto begin_collapse = std::chrono::high_resolution_clock::now(); + // strong_vertex_collapse(); + // strong_edge_collapse(); + // // strong_vertex_collapse(); + // auto end_collapse = std::chrono::high_resolution_clock::now(); + + // auto collapseTime = std::chrono::duration<double, std::milli>(end_collapse- begin_collapse).count(); + // return collapseTime; + // } + + bool membership(const Vertex& v) { + auto rw = vertexToRow.find(v); + if (rw != vertexToRow.end()) + return true; + else + return false; + } + + bool membership(const Edge& e) { + auto u = std::get<0>(e); + auto v = std::get<1>(e); + if (membership(u) && membership(v)) { + auto rw_u = vertexToRow[u]; + auto rw_v = vertexToRow[v]; + if (rw_u <= rw_v) + for (auto x : closed_neighbours_row_index(rw_v)) { // Taking advantage of sorted lists. + if (rw_u == x) + return true; + else if (rw_u < x) + return false; + } + else + for (auto x : closed_neighbours_row_index(rw_u)) { // Taking advantage of sorted lists. + if (rw_v == x) + return true; + else if (rw_v < x) + return false; + } + } + return false; + } + void insert_vertex(const Vertex& vertex, double filt_val) { + auto rw = vertexToRow.find(vertex); + if (rw == vertexToRow.end()) { + sparseRowAdjMatrix.insert(rows, rows) = + filt_val; // Initializing the diagonal element of the adjency matrix corresponding to rw_b. + vertDomnIndicator.push_back(false); + rowInsertIndicator.push_back(true); + contractionIndicator.push_back(false); + rowIterator.push(rows); + vertexToRow.insert(std::make_pair(vertex, rows)); + rowToVertex.insert(std::make_pair(rows, vertex)); + vertices.emplace(vertex); + rows++; + } + } + + void insert_new_edges(const Vertex& u, const Vertex& v, + double filt_val) // The edge must not be added before, it should be a new edge. + { + insert_vertex(u, filt_val); + if (u != v) { + insert_vertex(v, filt_val); + // std::cout << "Insertion of the edge begins " << u <<", " << v << std::endl; + + auto rw_u = vertexToRow.find(u); + auto rw_v = vertexToRow.find(v); + // std::cout << "Inserting the edge " << u <<", " << v << std::endl; + sparseRowAdjMatrix.insert(rw_u->second, rw_v->second) = filt_val; + sparseRowAdjMatrix.insert(rw_v->second, rw_u->second) = filt_val; + oneSimplices.emplace_back(u, v); + numOneSimplices++; + } + // else + // std::cout << "Already a member simplex, skipping..." << std::endl; + } + + std::size_t num_vertices() const { return vertices.size(); } + + //! Function for returning the ReductionMap. + /*! + This is the (stl's unordered) map that stores all the collapses of vertices. <br> + It is simply returned. + */ + + Map reduction_map() const { return ReductionMap; } + std::unordered_set<Vertex> vertex_set() const { return vertices; } + sparseRowMatrix collapsed_matrix() const { return *sparse_colpsd_adj_Matrix; } + + sparseRowMatrix uncollapsed_matrix() const { return sparseRowAdjMatrix; } + + edge_list all_edges() const { return oneSimplices; } + + vertexVector active_neighbors(const Vertex& v) { + vertexVector nb; + auto rw_v = vertexToRow.find(v); + if (rw_v != vertexToRow.end()) { + nb = vertex_closed_active_neighbours(rw_v->second); + } + return nb; + } + + vertexVector neighbors(const Vertex& v) { + vertexVector nb; + auto rw_v = vertexToRow.find(v); + if (rw_v != vertexToRow.end()) nb = closed_neighbours_vertex_index(rw_v->second); + + return nb; + } + + vertexVector active_relative_neighbors(const Vertex& v, const Vertex& w) { + std::vector<Vertex> diff; + if (membership(v) && membership(w)) { + auto nbhrs_v = active_neighbors(v); + auto nbhrs_w = active_neighbors(w); + std::set_difference(nbhrs_v.begin(), nbhrs_v.end(), nbhrs_w.begin(), nbhrs_w.end(), + std::inserter(diff, diff.begin())); + } + return diff; + } + + void contraction(const Vertex& del, const Vertex& keep) { + if (del != keep) { + bool del_mem = membership(del); + bool keep_mem = membership(keep); + if (del_mem && keep_mem) { + doubleVector del_indcs, keep_indcs, diff; + auto row_del = vertexToRow[del]; + auto row_keep = vertexToRow[keep]; + del_indcs = closed_neighbours_row_index(row_del); + keep_indcs = closed_neighbours_row_index(row_keep); + std::set_difference(del_indcs.begin(), del_indcs.end(), keep_indcs.begin(), keep_indcs.end(), + std::inserter(diff, diff.begin())); + for (auto& v : diff) { + if (v != row_del) { + sparseRowAdjMatrix.insert(row_keep, v) = 1; + sparseRowAdjMatrix.insert(v, row_keep) = 1; + } + } + vertexToRow.erase(del); + vertices.erase(del); + rowToVertex.erase(row_del); + // setZero(row_del->second, row_keep->second); + } else if (del_mem && not keep_mem) { + vertexToRow.insert(std::make_pair(keep, vertexToRow.find(del)->second)); + rowToVertex[vertexToRow.find(del)->second] = keep; + vertices.emplace(keep); + vertices.erase(del); + vertexToRow.erase(del); + + } else { + std::cerr << "The first vertex entered in the method contraction() doesn't exist in the skeleton." << std::endl; + exit(-1); + } + } + } + + void relable(const Vertex& v, const Vertex& w) { // relable v as w. + if (membership(v) and v != w) { + auto rw_v = vertexToRow[v]; + rowToVertex[rw_v] = w; + vertexToRow.insert(std::make_pair(w, rw_v)); + vertices.emplace(w); + vertexToRow.erase(v); + vertices.erase(v); + // std::cout<< "Re-named the vertex " << v << " as " << w << std::endl; + } + } + + // Returns the contracted edge. along with the contracted vertex in the begining of the list at {u,u} or {v,v} + + void active_strong_expansion(const Vertex& v, const Vertex& w, double filt_val) { + if (membership(v) && membership(w) && v != w) { + // std::cout << "Strong expansion of the vertex " << v << " and " << w << " begins. " << std::endl; + auto active_list_v_w = active_relative_neighbors(v, w); + auto active_list_w_v = active_relative_neighbors(w, v); + if (active_list_w_v.size() < + active_list_v_w.size()) { // simulate the contraction of w by expanding the star of v + for (auto& x : active_list_w_v) { + active_edge_insertion(v, x, filt_val); + // std::cout << "Inserted the edge " << v << " , " << x << std::endl; + } + swap_rows(v, w); + // std::cout << "A swap of the vertex " << v << " and " << w << "took place." << std::endl; + } else { + for (auto& y : active_list_v_w) { + active_edge_insertion(w, y, filt_val); + // std::cout << "Inserted the edge " << w << ", " << y << std::endl; + } + } + auto rw_v = vertexToRow.find(v); + contractionIndicator[rw_v->second] = true; + } + if (membership(v) && !membership(w)) { + relable(v, w); + } + } + void active_edge_insertion(const Vertex& v, const Vertex& w, double filt_val) { + insert_new_edges(v, w, filt_val); + // update_active_indicator(v,w); + } + + void print_sparse_skeleton() { std::cout << sparseRowAdjMatrix << std::endl; } +};
\ No newline at end of file diff --git a/src/Collapse/include/gudhi/Rips_edge_list.h b/src/Collapse/include/gudhi/Rips_edge_list.h new file mode 100644 index 00000000..b7c4dcff --- /dev/null +++ b/src/Collapse/include/gudhi/Rips_edge_list.h @@ -0,0 +1,184 @@ +/* This file is part of the Gudhi Library. The Gudhi library + * (Geometric Understanding in Higher Dimensions) is a generic C++ + * library for computational topology. + * + * Author(s): Clément Maria, Pawel Dlotko, Vincent Rouvreau Siddharth Pritam + * + * Copyright (C) 2016 INRIA + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * 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 General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see <http://www.gnu.org/licenses/>. + */ + +#ifndef RIPS_edge_list_H_ +#define RIPS_edge_list_H_ + +#include <gudhi/Debug_utils.h> +#include <gudhi/graph_simplicial_complex.h> +#include <boost/graph/adjacency_list.hpp> + +#include <iostream> +#include <vector> +#include <map> +#include <string> +#include <limits> // for numeric_limits +#include <utility> // for pair<> + + +namespace Gudhi { + +namespace rips_edge_list { + +/** + * \class Rips_complex + * \brief Rips complex data structure. + * + * \ingroup rips_complex + * + * \details + * The data structure is a one skeleton graph, or Rips graph, containing edges when the edge length is less or equal + * to a given threshold. Edge length is computed from a user given point cloud with a given distance function, or a + * distance matrix. + * + * \tparam Filtration_value is the type used to store the filtration values of the simplicial complex. + */ +template<typename Filtration_value> +class Rips_edge_list { + public: + /** + * \brief Type of the one skeleton graph stored inside the Rips complex structure. + */ + // typedef typename boost::adjacency_list < boost::vecS, boost::vecS, boost::undirectedS + // , boost::property < vertex_filtration_t, Filtration_value > + // , boost::property < edge_filtration_t, Filtration_value >> OneSkeletonGraph; + + private: + typedef int Vertex_handle; + + public: + /** \brief Rips_complex constructor from a list of points. + * + * @param[in] points Range of points. + * @param[in] threshold Rips value. + * @param[in] distance distance function that returns a `Filtration_value` from 2 given points. + * + * \tparam ForwardPointRange must be a range for which `std::begin` and `std::end` return input iterators on a + * point. + * + * \tparam Distance furnishes `operator()(const Point& p1, const Point& p2)`, where + * `Point` is a point from the `ForwardPointRange`, and that returns a `Filtration_value`. + */ + template<typename ForwardPointRange, typename Distance > + Rips_edge_list(const ForwardPointRange& points, Filtration_value threshold, Distance distance) { + compute_proximity_graph(points, threshold, distance); + } + + /** \brief Rips_complex constructor from a distance matrix. + * + * @param[in] distance_matrix Range of distances. + * @param[in] threshold Rips value. + * + * \tparam DistanceMatrix must have a `size()` method and on which `distance_matrix[i][j]` returns + * the distance between points \f$i\f$ and \f$j\f$ as long as \f$ 0 \leqslant i < j \leqslant + * distance\_matrix.size().\f$ + */ + template<typename DistanceMatrix> + Rips_edge_list(const DistanceMatrix& distance_matrix, Filtration_value threshold) { + compute_proximity_graph(boost::irange((size_t)0, distance_matrix.size()), threshold, + [&](size_t i, size_t j){return distance_matrix[j][i];}); + } + + /** \brief Initializes the egde list (one skeleton) complex from the Rips graph + * + * \tparam EdgeListForRips must meet `EdgeListForRips` concept. + * + * @param[in] edges EdgeListForRips to be created. + * @param[in] dim_max graph expansion for Rips until this given maximal dimension. + * @exception std::invalid_argument In debug mode, if `edges.num_vertices()` does not return 0. + * + */ + template <typename EdgeListForRips> + void create_edges(EdgeListForRips& edge_list) { + GUDHI_CHECK(edges.num_vertices() == 0, + std::invalid_argument("Rips_complex::create_complex - edge list is not empty")); + + // sort the tuple (filteration_valuem, (v1,v2){edge}) + //By default the sort is done on the first element, so here it's filteration value. + std::sort(edges.begin(),edges.end()); + for(size_t i = 0; i < edges.size(); i++ ) + edge_list.emplace_back(edges.at(i)); + + } + + private: + /** \brief Computes the proximity graph of the points. + * + * If points contains n elements, the proximity graph is the graph with n vertices, and an edge [u,v] iff the + * distance function between points u and v is smaller than threshold. + * + * \tparam ForwardPointRange furnishes `.begin()` and `.end()` + * methods. + * + * \tparam Distance furnishes `operator()(const Point& p1, const Point& p2)`, where + * `Point` is a point from the `ForwardPointRange`, and that returns a `Filtration_value`. + */ + template< typename ForwardPointRange, typename Distance > + void compute_proximity_graph(const ForwardPointRange& points, Filtration_value threshold, + Distance distance) { + edges.clear(); + // Compute the proximity graph of the points. + // If points contains n elements, the proximity graph is the graph with n vertices, and an edge [u,v] iff the + // distance function between points u and v is smaller than threshold. + // -------------------------------------------------------------------------------------------- + // Creates the vector of edges and its filtration values (returned by distance function) + Vertex_handle idx_u = 0; + for (auto it_u = std::begin(points); it_u != std::end(points); ++it_u, ++idx_u) { + Vertex_handle idx_v = idx_u + 1; + for (auto it_v = it_u + 1; it_v != std::end(points); ++it_v, ++idx_v) { + Filtration_value fil = distance(*it_u, *it_v); + if (fil <= threshold) { + edges.emplace_back(fil, idx_u, idx_v); + } + } + } + + // -------------------------------------------------------------------------------------------- + // Creates the proximity graph from edges and sets the property with the filtration value. + // Number of points is labeled from 0 to idx_u-1 + // -------------------------------------------------------------------------------------------- + // Do not use : rips_skeleton_graph_ = OneSkeletonGraph(...) -> deep copy of the graph (boost graph is not + // move-enabled) + // rips_skeleton_graph_.~OneSkeletonGraph(); + // new(&rips_skeleton_graph_)OneSkeletonGraph(edges.begin(), edges.end(), edges_fil.begin(), idx_u); + + // auto vertex_prop = boost::get(vertex_filtration_t(), rips_skeleton_graph_); + + // using vertex_iterator = typename boost::graph_traits<OneSkeletonGraph>::vertex_iterator; + // vertex_iterator vi, vi_end; + // for (std::tie(vi, vi_end) = boost::vertices(rips_skeleton_graph_); + // vi != vi_end; ++vi) { + // boost::put(vertex_prop, *vi, 0.); + // } + } + + private: + // OneSkeletonGraph rips_skeleton_graph_; + std::vector< std::tuple < Filtration_value, Vertex_handle, Vertex_handle > > edges; + // std::vector< Filtration_value > edges_fil; +}; + +} // namespace rips_complex + +} // namespace Gudhi + +#endif // RIPS_COMPLEX_H_ |