| 1 | /* $Id$ $Revision$ */ |
| 2 | /* vim:set shiftwidth=4 ts=8: */ |
| 3 | |
| 4 | /************************************************************************* |
| 5 | * Copyright (c) 2011 AT&T Intellectual Property |
| 6 | * All rights reserved. This program and the accompanying materials |
| 7 | * are made available under the terms of the Eclipse Public License v1.0 |
| 8 | * which accompanies this distribution, and is available at |
| 9 | * http://www.eclipse.org/legal/epl-v10.html |
| 10 | * |
| 11 | * Contributors: See CVS logs. Details at http://www.graphviz.org/ |
| 12 | *************************************************************************/ |
| 13 | |
| 14 | #include "digcola.h" |
| 15 | #ifdef DIGCOLA |
| 16 | #include "matrix_ops.h" |
| 17 | #include "conjgrad.h" |
| 18 | |
| 19 | static void construct_b(vtx_data * graph, int n, double *b) |
| 20 | { |
| 21 | /* construct a vector - b s.t. -b[i]=\sum_j -w_{ij}*\delta_{ij} |
| 22 | * (the "balance vector") |
| 23 | * Note that we build -b and not b, since our matrix is not the |
| 24 | * real laplacian L, but its negation: -L. |
| 25 | * So instead of solving Lx=b, we will solve -Lx=-b |
| 26 | */ |
| 27 | int i, j; |
| 28 | |
| 29 | double b_i = 0; |
| 30 | |
| 31 | for (i = 0; i < n; i++) { |
| 32 | b_i = 0; |
| 33 | if (graph[0].edists == NULL) { |
| 34 | continue; |
| 35 | } |
| 36 | for (j = 1; j < graph[i].nedges; j++) { /* skip the self loop */ |
| 37 | b_i += graph[i].ewgts[j] * graph[i].edists[j]; |
| 38 | } |
| 39 | b[i] = b_i; |
| 40 | } |
| 41 | } |
| 42 | |
| 43 | #define hierarchy_cg_tol 1e-3 |
| 44 | |
| 45 | int |
| 46 | compute_y_coords(vtx_data * graph, int n, double *y_coords, |
| 47 | int max_iterations) |
| 48 | { |
| 49 | /* Find y coords of a directed graph by solving L*x = b */ |
| 50 | int i, j, rv = 0; |
| 51 | double *b = N_NEW(n, double); |
| 52 | double tol = hierarchy_cg_tol; |
| 53 | int nedges = 0; |
| 54 | float *uniform_weights; |
| 55 | float *old_ewgts = graph[0].ewgts; |
| 56 | |
| 57 | construct_b(graph, n, b); |
| 58 | |
| 59 | init_vec_orth1(n, y_coords); |
| 60 | |
| 61 | for (i = 0; i < n; i++) { |
| 62 | nedges += graph[i].nedges; |
| 63 | } |
| 64 | |
| 65 | /* replace original edge weights (which are lengths) with uniform weights */ |
| 66 | /* for computing the optimal arrangement */ |
| 67 | uniform_weights = N_GNEW(nedges, float); |
| 68 | for (i = 0; i < n; i++) { |
| 69 | graph[i].ewgts = uniform_weights; |
| 70 | uniform_weights[0] = (float) -(graph[i].nedges - 1); |
| 71 | for (j = 1; j < graph[i].nedges; j++) { |
| 72 | uniform_weights[j] = 1; |
| 73 | } |
| 74 | uniform_weights += graph[i].nedges; |
| 75 | } |
| 76 | |
| 77 | if (conjugate_gradient(graph, y_coords, b, n, tol, max_iterations) < 0) { |
| 78 | rv = 1; |
| 79 | } |
| 80 | |
| 81 | /* restore original edge weights */ |
| 82 | free(graph[0].ewgts); |
| 83 | for (i = 0; i < n; i++) { |
| 84 | graph[i].ewgts = old_ewgts; |
| 85 | old_ewgts += graph[i].nedges; |
| 86 | } |
| 87 | |
| 88 | free(b); |
| 89 | return rv; |
| 90 | } |
| 91 | |
| 92 | #endif /* DIGCOLA */ |
| 93 | |