Преглед на файлове

Merge commit '2f41fc735f20b97eda05a3f0b56d22092bab8532 [formerly 9203ec0c9278ce60ebb6310e646907746cb178d6]'

Former-commit-id: deca72e369f2cc389885d5e35a9a4f0ee8f08e63
Daniele Panozzo преди 9 години
родител
ревизия
405da96def
променени са 3 файла, в които са добавени 1589 реда и са изтрити 5 реда
  1. 1588 0
      include/igl/comiso/miq.cpp
  2. 0 1
      include/igl/comiso/miq.cpp.REMOVED.git-id
  3. 1 4
      include/igl/comiso/miq.h

+ 1588 - 0
include/igl/comiso/miq.cpp

@@ -0,0 +1,1588 @@
+// This file is part of libigl, a simple c++ geometry processing library.
+//
+// Copyright (C) 2014 Daniele Panozzo <daniele.panozzo@gmail.com>, Olga Diamanti <olga.diam@gmail.com>
+//
+// This Source Code Form is subject to the terms of the Mozilla Public License
+// v. 2.0. If a copy of the MPL was not distributed with this file, You can
+// obtain one at http://mozilla.org/MPL/2.0/.
+
+#include <igl/comiso/miq.h>
+#include <igl/local_basis.h>
+#include <igl/triangle_triangle_adjacency.h>
+#include <igl/cut_mesh.h>
+
+// includes for VertexIndexing
+#include <igl/HalfEdgeIterator.h>
+#include <igl/is_border_vertex.h>
+#include <igl/vertex_triangle_adjacency.h>
+
+// includes for PoissonSolver
+#include <igl/slice_into.h>
+#include <igl/grad.h>
+#include <igl/cotmatrix.h>
+#include <igl/doublearea.h>
+#include <gmm/gmm.h>
+#include <CoMISo/Solver/ConstrainedSolver.hh>
+#include <CoMISo/Solver/MISolver.hh>
+#include <CoMISo/Solver/GMM_Tools.hh>
+
+//
+#include <igl/cross_field_missmatch.h>
+#include <igl/comb_frame_field.h>
+#include <igl/comb_cross_field.h>
+#include <igl/cut_mesh_from_singularities.h>
+#include <igl/find_cross_field_singularities.h>
+#include <igl/compute_frame_field_bisectors.h>
+#include <igl/rotate_vectors.h>
+
+#ifndef NDEBUG
+#include <fstream>
+#endif
+#include <iostream>
+#include <igl/matlab_format.h>
+
+#warning "using namespace *; in global scope **must** be removed"
+using namespace std;
+using namespace Eigen;
+
+#define DEBUGPRINT 0
+
+
+namespace igl {
+namespace comiso {
+  struct SeamInfo
+  {
+    int v0,v0p;
+    int integerVar;
+    unsigned char MMatch;
+
+    IGL_INLINE SeamInfo(int _v0,
+                        int _v0p,
+                        int _MMatch,
+                        int _integerVar);
+
+    IGL_INLINE SeamInfo(const SeamInfo &S1);
+  };
+
+  struct MeshSystemInfo
+  {
+    ////number of vertices variables
+    int num_vert_variables;
+    ///num of integer for cuts
+    int num_integer_cuts;
+    ///this are used for drawing purposes
+    std::vector<SeamInfo> EdgeSeamInfo;
+#if 0
+    ///this are values of integer variables after optimization
+    std::vector<int> IntegerValues;
+#endif
+  };
+
+
+  template <typename DerivedV, typename DerivedF>
+  class VertexIndexing
+  {
+  public:
+    // Input:
+    const Eigen::PlainObjectBase<DerivedV> &V;
+    const Eigen::PlainObjectBase<DerivedF> &F;
+    const Eigen::PlainObjectBase<DerivedV> &Vcut;
+    const Eigen::PlainObjectBase<DerivedF> &Fcut;
+    const Eigen::PlainObjectBase<DerivedF> &TT;
+    const Eigen::PlainObjectBase<DerivedF> &TTi;
+
+    const Eigen::Matrix<int, Eigen::Dynamic, 3> &Handle_MMatch;
+    const Eigen::Matrix<int, Eigen::Dynamic, 1> &Handle_Singular; // bool
+    const Eigen::Matrix<int, Eigen::Dynamic, 3> &Handle_Seams; // 3 bool
+
+
+    ///this handle for mesh TODO: move with the other global variables
+    MeshSystemInfo Handle_SystemInfo;
+
+    IGL_INLINE VertexIndexing(const Eigen::PlainObjectBase<DerivedV> &_V,
+                              const Eigen::PlainObjectBase<DerivedF> &_F,
+                              const Eigen::PlainObjectBase<DerivedV> &_Vcut,
+                              const Eigen::PlainObjectBase<DerivedF> &_Fcut,
+                              const Eigen::PlainObjectBase<DerivedF> &_TT,
+                              const Eigen::PlainObjectBase<DerivedF> &_TTi,
+                              const Eigen::Matrix<int, Eigen::Dynamic, 3> &_Handle_MMatch,
+                              const Eigen::Matrix<int, Eigen::Dynamic, 1> &_Handle_Singular,
+                              const Eigen::Matrix<int, Eigen::Dynamic, 3> &_Handle_Seams
+                              );
+
+    // provide information about every vertex per seam
+    IGL_INLINE void InitSeamInfo();
+
+
+  private:
+    struct VertexInfo{
+      int v;  // vertex index (according to V)
+      int f0, k0; // face and local edge information of the edge that connects this vertex to the previous vertex (previous in the vector)
+      int f1, k1; // face and local edge information of the other face corresponding to the same edge
+      VertexInfo(int _v, int _f0, int _k0, int _f1, int _k1) :
+                 v(_v), f0(_f0), k0(_k0), f1(_f1), k1(_k1){}
+      bool operator==(VertexInfo const& other){
+        return other.v == v;
+      }
+    };
+
+    IGL_INLINE void GetSeamInfo(const int f0,
+                                const int f1,
+                                const int indexE,
+                                int &v0,int &v1,
+                                int &v0p,int &v1p,
+                                unsigned char &_MMatch);
+
+    IGL_INLINE std::vector<std::vector<VertexInfo> > GetVerticesPerSeam();
+  };
+
+
+  template <typename DerivedV, typename DerivedF>
+  class PoissonSolver
+  {
+
+  public:
+    IGL_INLINE void SolvePoisson(Eigen::VectorXd Stiffness,
+                                 double vector_field_scale=0.1f,
+                                 double grid_res=1.f,
+                                 bool direct_round=true,
+                                 int localIter=0,
+                                 bool _integer_rounding=true,
+                                 bool _singularity_rounding=true,
+                                 std::vector<int> roundVertices = std::vector<int>(),
+                                 std::vector<std::vector<int> > hardFeatures = std::vector<std::vector<int> >());
+
+    IGL_INLINE PoissonSolver(const Eigen::PlainObjectBase<DerivedV> &_V,
+                             const Eigen::PlainObjectBase<DerivedF> &_F,
+                             const Eigen::PlainObjectBase<DerivedV> &_Vcut,
+                             const Eigen::PlainObjectBase<DerivedF> &_Fcut,
+                             const Eigen::PlainObjectBase<DerivedF> &_TT,
+                             const Eigen::PlainObjectBase<DerivedF> &_TTi,
+                             const Eigen::PlainObjectBase<DerivedV> &_PD1,
+                             const Eigen::PlainObjectBase<DerivedV> &_PD2,
+                             const Eigen::Matrix<int, Eigen::Dynamic, 1>&_Handle_Singular,
+                             const MeshSystemInfo &_Handle_SystemInfo
+                             );
+
+    const Eigen::PlainObjectBase<DerivedV> &V;
+    const Eigen::PlainObjectBase<DerivedF> &F;
+    const Eigen::PlainObjectBase<DerivedV> &Vcut;
+    const Eigen::PlainObjectBase<DerivedF> &Fcut;
+    const Eigen::PlainObjectBase<DerivedF> &TT;
+    const Eigen::PlainObjectBase<DerivedF> &TTi;
+    const Eigen::PlainObjectBase<DerivedV> &PD1;
+    const Eigen::PlainObjectBase<DerivedV> &PD2;
+    const Eigen::Matrix<int, Eigen::Dynamic, 1> &Handle_Singular; // bool
+
+    const MeshSystemInfo &Handle_SystemInfo;
+
+    // Internal:
+    Eigen::VectorXd Handle_Stiffness;
+    std::vector<std::vector<int> > VF;
+    std::vector<std::vector<int> > VFi;
+    Eigen::MatrixXd UV; // this is probably useless
+
+    // Output:
+    // per wedge UV coordinates, 6 coordinates (1 face) per row
+    Eigen::MatrixXd WUV;
+    // per vertex UV coordinates, Vcut.rows() x 2
+    Eigen::MatrixXd UV_out;
+
+    // Matrices
+    Eigen::SparseMatrix<double> Lhs;
+    Eigen::SparseMatrix<double> Constraints;
+    Eigen::VectorXd rhs;
+    Eigen::VectorXd constraints_rhs;
+    ///vector of unknowns
+    std::vector< double > X;
+
+    ////REAL PART
+    ///number of fixed vertex
+    unsigned int n_fixed_vars;
+
+    ///the number of REAL variables for vertices
+    unsigned int n_vert_vars;
+
+    ///total number of variables of the system,
+    ///do not consider constraints, but consider integer vars
+    unsigned int num_total_vars;
+
+    //////INTEGER PART
+    ///the total number of integer variables
+    unsigned int n_integer_vars;
+
+    ///CONSTRAINT PART
+    ///number of cuts constraints
+    unsigned int num_cut_constraint;
+
+    // number of user-defined constraints
+    unsigned int num_userdefined_constraint;
+
+    ///total number of constraints equations
+    unsigned int num_constraint_equations;
+
+    ///vector of blocked vertices
+    std::vector<int> Hard_constraints;
+
+    ///vector of indexes to round
+    std::vector<int> ids_to_round;
+
+    ///vector of indexes to round
+    std::vector<std::vector<int > > userdefined_constraints;
+
+    ///boolean that is true if rounding to integer is needed
+    bool integer_rounding;
+
+    ///START COMMON MATH FUNCTIONS
+    ///return the complex encoding the rotation
+    ///for a given missmatch interval
+    IGL_INLINE std::complex<double> GetRotationComplex(int interval);
+    ///END COMMON MATH FUNCTIONS
+
+    ///START FIXING VERTICES
+    ///set a given vertex as fixed
+    IGL_INLINE void AddFixedVertex(int v);
+
+    ///find vertex to fix in case we're using
+    ///a vector field NB: multiple components not handled
+    IGL_INLINE void FindFixedVertField();
+
+    ///find hard constraint depending if using or not
+    ///a vector field
+    IGL_INLINE void FindFixedVert();
+
+    IGL_INLINE int GetFirstVertexIndex(int v);
+
+    ///fix the vertices which are flagged as fixed
+    IGL_INLINE void FixBlockedVertex();
+    ///END FIXING VERTICES
+
+    ///HANDLING SINGULARITY
+    //set the singularity round to integer location
+    IGL_INLINE void AddSingularityRound();
+
+    IGL_INLINE void AddToRoundVertices(std::vector<int> ids);
+
+    ///START GENERIC SYSTEM FUNCTIONS
+    //build the laplacian matrix cyclyng over all rangemaps
+    //and over all faces
+    IGL_INLINE void BuildLaplacianMatrix(double vfscale=1);
+
+    ///find different sized of the system
+    IGL_INLINE void FindSizes();
+
+    IGL_INLINE void AllocateSystem();
+
+    ///intitialize the whole matrix
+    IGL_INLINE void InitMatrix();
+
+    ///map back coordinates after that
+    ///the system has been solved
+    IGL_INLINE void MapCoords();
+    ///END GENERIC SYSTEM FUNCTIONS
+
+    ///set the constraints for the inter-range cuts
+    IGL_INLINE void BuildSeamConstraintsExplicitTranslation();
+
+    ///set the constraints for the inter-range cuts
+    IGL_INLINE void BuildUserDefinedConstraints();
+
+    ///call of the mixed integer solver
+    IGL_INLINE void MixedIntegerSolve(double cone_grid_res=1,
+                                      bool direct_round=true,
+                                      int localIter=0);
+
+    IGL_INLINE void clearUserConstraint();
+
+    IGL_INLINE void addSharpEdgeConstraint(int fid, int vid);
+
+  };
+
+  template <typename DerivedV, typename DerivedF, typename DerivedU>
+  class MIQ_class
+  {
+  private:
+    const Eigen::PlainObjectBase<DerivedV> &V;
+    const Eigen::PlainObjectBase<DerivedF> &F;
+    Eigen::PlainObjectBase<DerivedV> Vcut;
+    Eigen::PlainObjectBase<DerivedF> Fcut;
+    Eigen::MatrixXd UV_out;
+    Eigen::PlainObjectBase<DerivedF> FUV_out;
+
+    // internal
+    Eigen::PlainObjectBase<DerivedF> TT;
+    Eigen::PlainObjectBase<DerivedF> TTi;
+
+    // Stiffness per face
+    Eigen::VectorXd Handle_Stiffness;
+    Eigen::PlainObjectBase<DerivedV> B1, B2, B3;
+
+  public:
+    IGL_INLINE MIQ_class(const Eigen::PlainObjectBase<DerivedV> &V_,
+                         const Eigen::PlainObjectBase<DerivedF> &F_,
+                         const Eigen::PlainObjectBase<DerivedV> &PD1_combed,
+                         const Eigen::PlainObjectBase<DerivedV> &PD2_combed,
+                         const Eigen::Matrix<int, Eigen::Dynamic, 3> &Handle_MMatch,
+                         const Eigen::Matrix<int, Eigen::Dynamic, 1> &Handle_Singular,
+                         const Eigen::Matrix<int, Eigen::Dynamic, 3> &Handle_Seams,
+                         Eigen::PlainObjectBase<DerivedU> &UV,
+                         Eigen::PlainObjectBase<DerivedF> &FUV,
+                         double GradientSize = 30.0,
+                         double Stiffness = 5.0,
+                         bool DirectRound = false,
+                         int iter = 5,
+                         int localIter = 5,
+                         bool DoRound = true,
+                         bool SingularityRound=true,
+                         std::vector<int> roundVertices = std::vector<int>(),
+                         std::vector<std::vector<int> > hardFeatures = std::vector<std::vector<int> >());
+
+
+    IGL_INLINE void extractUV(Eigen::PlainObjectBase<DerivedU> &UV_out,
+                              Eigen::PlainObjectBase<DerivedF> &FUV_out);
+
+  private:
+    IGL_INLINE int NumFlips(const Eigen::MatrixXd& WUV);
+
+    IGL_INLINE double Distortion(int f, double h, const Eigen::MatrixXd& WUV);
+
+    IGL_INLINE double LaplaceDistortion(const int f, double h, const Eigen::MatrixXd& WUV);
+
+    IGL_INLINE bool updateStiffeningJacobianDistorsion(double grad_size, const Eigen::MatrixXd& WUV);
+
+    IGL_INLINE bool IsFlipped(const Eigen::Vector2d &uv0,
+                              const Eigen::Vector2d &uv1,
+                              const Eigen::Vector2d &uv2);
+
+    IGL_INLINE bool IsFlipped(const int i, const Eigen::MatrixXd& WUV);
+
+  };
+};
+}
+
+IGL_INLINE igl::comiso::SeamInfo::SeamInfo(int _v0,
+                                   int _v0p,
+                                   int _MMatch,
+                                   int _integerVar)
+{
+  v0=_v0;
+  v0p=_v0p;
+  integerVar=_integerVar;
+  MMatch=_MMatch;
+}
+
+IGL_INLINE igl::comiso::SeamInfo::SeamInfo(const SeamInfo &S1)
+{
+  v0=S1.v0;
+  v0p=S1.v0p;
+  integerVar=S1.integerVar;
+  MMatch=S1.MMatch;
+}
+
+
+template <typename DerivedV, typename DerivedF>
+IGL_INLINE igl::comiso::VertexIndexing<DerivedV, DerivedF>::VertexIndexing(const Eigen::PlainObjectBase<DerivedV> &_V,
+                                                                   const Eigen::PlainObjectBase<DerivedF> &_F,
+                                                                   const Eigen::PlainObjectBase<DerivedV> &_Vcut,
+                                                                   const Eigen::PlainObjectBase<DerivedF> &_Fcut,
+                                                                   const Eigen::PlainObjectBase<DerivedF> &_TT,
+                                                                   const Eigen::PlainObjectBase<DerivedF> &_TTi,
+                                                                   const Eigen::Matrix<int, Eigen::Dynamic, 3> &_Handle_MMatch,
+                                                                   const Eigen::Matrix<int, Eigen::Dynamic, 1> &_Handle_Singular,
+                                                                   const Eigen::Matrix<int, Eigen::Dynamic, 3> &_Handle_Seams
+                                                                   ):
+V(_V),
+F(_F),
+Vcut(_Vcut),
+Fcut(_Fcut),
+TT(_TT),
+TTi(_TTi),
+Handle_MMatch(_Handle_MMatch),
+Handle_Singular(_Handle_Singular),
+Handle_Seams(_Handle_Seams)
+{
+  #ifdef DEBUG_PRINT
+  cerr<<igl::matlab_format(Handle_Seams,"Handle_Seams");
+#endif
+
+  Handle_SystemInfo.num_vert_variables=Vcut.rows();
+  Handle_SystemInfo.num_integer_cuts=0;
+}
+
+template <typename DerivedV, typename DerivedF>
+IGL_INLINE void igl::comiso::VertexIndexing<DerivedV, DerivedF>::GetSeamInfo(const int f0,
+                                                                     const int f1,
+                                                                     const int indexE,
+                                                                     int &v0,int &v1,
+                                                                     int &v0p,int &v1p,
+                                                                     unsigned char &_MMatch)
+{
+  int edgef0 = indexE;
+  v0 = Fcut(f0,edgef0);
+  v1 = Fcut(f0,(edgef0+1)%3);
+  ////get the index on opposite side
+  assert(TT(f0,edgef0) == f1);
+  int edgef1 = TTi(f0,edgef0);
+  v1p = Fcut(f1,edgef1);
+  v0p = Fcut(f1,(edgef1+1)%3);
+
+  _MMatch = Handle_MMatch(f0,edgef0);
+  assert(F(f0,edgef0)         == F(f1,((edgef1+1)%3)));
+  assert(F(f0,((edgef0+1)%3)) == F(f1,edgef1));
+}
+
+template <typename DerivedV, typename DerivedF>
+IGL_INLINE std::vector<std::vector<typename igl::comiso::VertexIndexing<DerivedV, DerivedF>::VertexInfo> > igl::comiso::VertexIndexing<DerivedV, DerivedF>::GetVerticesPerSeam()
+{
+  // Return value
+  std::vector<std::vector<VertexInfo> >verticesPerSeam;
+
+  // for every vertex, keep track of their adjacent vertices on seams.
+  // regular vertices have two neighbors on a seam, start- and endvertices may have any other numbers of neighbors (e.g. 1 or 3)
+  std::vector<std::list<VertexInfo> > VVSeam(V.rows());
+  Eigen::MatrixXi F_hit = Eigen::MatrixXi::Zero(F.rows(), 3);
+  for (unsigned int f=0; f<F.rows();f++)
+  {
+    int f0 = f;
+    for(int k0=0; k0<3; k0++){
+      int f1 = TT(f0,k0);
+      if(f1 == -1)
+        continue;
+
+      bool seam = Handle_Seams(f0,k0);
+      if (seam && F_hit(f0,k0) == 0)
+      {
+        int v0 = F(f0, k0);
+        int v1 = F(f0, (k0+1)%3);
+        int k1 = TTi(f0,k0);
+        VVSeam[v0].push_back(VertexInfo(v1, f0, k0, f1, k1));
+        VVSeam[v1].push_back(VertexInfo(v0, f0, k0, f1, k1));
+        F_hit(f0, k0) = 1;
+        F_hit(f1, k1) = 1;
+      }
+    }
+  }
+
+  // Find start vertices, i.e. vertices that start or end a seam branch
+  std::vector<int> startVertexIndices;
+  std::vector<bool> isStartVertex(V.rows());
+  for (unsigned int i=0;i<V.rows();i++)
+  {
+    isStartVertex[i] = false;
+    // vertices with two neighbors are regular vertices, unless the vertex is a singularity, in which case it qualifies as a start vertex
+    if (VVSeam[i].size() > 0 && VVSeam[i].size() != 2 || Handle_Singular(i) == true)
+    {
+      startVertexIndices.push_back(i);
+      isStartVertex[i] = true;
+    }
+  }
+
+  // For each startVertex, walk along its seam
+  for (unsigned int i=0;i<startVertexIndices.size();i++)
+  {
+    auto startVertexNeighbors = &VVSeam[startVertexIndices[i]];
+    const int neighborSize = startVertexNeighbors->size();
+
+    // explore every seam to which this vertex is a start vertex
+    // note: a vertex can never be a start vertex and a regular vertex simultaneously
+    for (unsigned int j=0;j<neighborSize;j++)
+    {
+      std::vector<VertexInfo> thisSeam; // temporary container
+
+      // Create vertexInfo struct for start vertex
+      auto startVertex = VertexInfo(startVertexIndices[i], -1, -1, -1, -1);// -1 values are arbitrary (will never be used)
+      auto currentVertex = startVertex;
+      // Add start vertex to the seam
+      thisSeam.push_back(currentVertex);
+
+      // advance on the seam
+      auto currentVertexNeighbors = startVertexNeighbors;
+      auto nextVertex = currentVertexNeighbors->front();
+      currentVertexNeighbors->pop_front();
+
+      auto prevVertex = startVertex; // bogus initialization to get the type
+      while (true)
+      {
+        // move to the next vertex
+        prevVertex = currentVertex;
+        currentVertex = nextVertex;
+        currentVertexNeighbors = &VVSeam[nextVertex.v];
+
+        // add current vertex to this seam
+        thisSeam.push_back(currentVertex);
+
+        // remove the previous vertex
+        auto it = std::find(currentVertexNeighbors->begin(), currentVertexNeighbors->end(), prevVertex);
+        assert(it != currentVertexNeighbors->end());
+        currentVertexNeighbors->erase(it);
+
+        if (currentVertexNeighbors->size() == 1 && !isStartVertex[currentVertex.v])
+        {
+          nextVertex = currentVertexNeighbors->front();
+          currentVertexNeighbors->pop_front();
+        }
+        else
+          break;
+      }
+      verticesPerSeam.push_back(thisSeam);
+    }
+  }
+
+  return verticesPerSeam;
+}
+
+template <typename DerivedV, typename DerivedF>
+IGL_INLINE void igl::comiso::VertexIndexing<DerivedV, DerivedF>::InitSeamInfo()
+{
+  auto verticesPerSeam = GetVerticesPerSeam();
+  Handle_SystemInfo.EdgeSeamInfo.clear();
+  int integerVar = 0;
+  // Loop over each seam
+  for(auto seam : verticesPerSeam){
+    //choose initial side of the seam such that the start vertex corresponds to Fcut(f, k) and the end vertex corresponds to Fcut(f, (k+1)%3) and not vice versa.
+    int priorVertexIdx;
+    if(seam.size() > 2){
+      auto v1 = seam[1];
+      auto v2 = seam[2];
+      if(Fcut(v1.f0, (v1.k0+1) % 3) == Fcut(v2.f0, v2.k0) || Fcut(v1.f0, (v1.k0+1) % 3) == Fcut(v2.f1, v2.k1)){
+        priorVertexIdx = Fcut(v1.f0, v1.k0);
+      }
+      else{
+        priorVertexIdx = Fcut(v1.f1, v1.k1);
+        assert(Fcut(v1.f1, (v1.k1+1) % 3) == Fcut(v2.f0, v2.k0) || Fcut(v1.f1, (v1.k1+1) % 3) == Fcut(v2.f1, v2.k1));
+      }
+    }
+    else{
+      auto v1 = seam[1];
+      priorVertexIdx = Fcut(v1.f0, v1.k0);
+    }
+
+    // Loop over each vertex of the seam
+    for(auto it=seam.begin()+1; it != seam.end(); ++it){
+      auto vertex = *it;
+      // choose the correct side of the seam
+      int f,k,ff,kk;
+      if(priorVertexIdx == Fcut(vertex.f0, vertex.k0)){
+        f = vertex.f0; ff = vertex.f1;
+        k = vertex.k0; kk = vertex.k1;
+      }
+      else{
+        f = vertex.f1; ff = vertex.f0;
+        k = vertex.k1; kk = vertex.k0;
+        assert(priorVertexIdx == Fcut(vertex.f1, vertex.k1));
+      }
+
+      int vtx0,vtx0p,vtx1,vtx1p;
+      unsigned char MM;
+      GetSeamInfo(f,ff,k,vtx0,vtx1,vtx0p,vtx1p,MM);
+      Handle_SystemInfo.EdgeSeamInfo.push_back(SeamInfo(vtx0,vtx0p,MM,integerVar));
+      if(it == seam.end() -1){
+        Handle_SystemInfo.EdgeSeamInfo.push_back(SeamInfo(vtx1,vtx1p,MM,integerVar));
+      }
+      priorVertexIdx = vtx1;
+    }
+    // use the same integer for each seam
+    integerVar++;
+  }
+  Handle_SystemInfo.num_integer_cuts = integerVar;
+
+#ifndef NDEBUG
+  int totalNVerticesOnSeams = 0;
+  for(auto seam : verticesPerSeam){
+    totalNVerticesOnSeams += seam.size();
+  }
+  assert(Handle_SystemInfo.EdgeSeamInfo.size() == totalNVerticesOnSeams);
+#endif
+}
+
+
+
+template <typename DerivedV, typename DerivedF>
+IGL_INLINE void igl::comiso::PoissonSolver<DerivedV, DerivedF>::SolvePoisson(Eigen::VectorXd Stiffness,
+                                                                     double vector_field_scale,
+                                                                     double grid_res,
+                                                                     bool direct_round,
+                                                                     int localIter,
+                                                                     bool _integer_rounding,
+                                                                     bool _singularity_rounding,
+                                                                     std::vector<int> roundVertices,
+                                                                     std::vector<std::vector<int> > hardFeatures)
+{
+  Handle_Stiffness = Stiffness;
+
+  //initialization of flags and data structures
+  integer_rounding=_integer_rounding;
+
+  ids_to_round.clear();
+
+  clearUserConstraint();
+  // copy the user constraints number
+  for (size_t i = 0; i < hardFeatures.size(); ++i)
+  {
+    addSharpEdgeConstraint(hardFeatures[i][0],hardFeatures[i][1]);
+  }
+
+  ///Initializing Matrix
+
+  int t0=clock();
+
+  ///initialize the matrix ALLOCATING SPACE
+  InitMatrix();
+  if (DEBUGPRINT)
+    printf("\n ALLOCATED THE MATRIX \n");
+
+  ///build the laplacian system
+  BuildLaplacianMatrix(vector_field_scale);
+
+  // add seam constraints
+  BuildSeamConstraintsExplicitTranslation();
+
+  // add user defined constraints
+  BuildUserDefinedConstraints();
+
+  ////add the lagrange multiplier
+  FixBlockedVertex();
+
+  if (DEBUGPRINT)
+    printf("\n BUILT THE MATRIX \n");
+
+  if (integer_rounding)
+    AddToRoundVertices(roundVertices);
+
+  if (_singularity_rounding)
+    AddSingularityRound();
+
+  int t1=clock();
+  if (DEBUGPRINT) printf("\n time:%d \n",t1-t0);
+  if (DEBUGPRINT) printf("\n SOLVING \n");
+
+  MixedIntegerSolve(grid_res,direct_round,localIter);
+
+  int t2=clock();
+  if (DEBUGPRINT) printf("\n time:%d \n",t2-t1);
+  if (DEBUGPRINT) printf("\n ASSIGNING COORDS \n");
+
+  MapCoords();
+
+  int t3=clock();
+  if (DEBUGPRINT) printf("\n time:%d \n",t3-t2);
+  if (DEBUGPRINT) printf("\n FINISHED \n");
+}
+
+template <typename DerivedV, typename DerivedF>
+IGL_INLINE igl::comiso::PoissonSolver<DerivedV, DerivedF>
+::PoissonSolver(const Eigen::PlainObjectBase<DerivedV> &_V,
+                const Eigen::PlainObjectBase<DerivedF> &_F,
+                const Eigen::PlainObjectBase<DerivedV> &_Vcut,
+                const Eigen::PlainObjectBase<DerivedF> &_Fcut,
+                const Eigen::PlainObjectBase<DerivedF> &_TT,
+                const Eigen::PlainObjectBase<DerivedF> &_TTi,
+                const Eigen::PlainObjectBase<DerivedV> &_PD1,
+                const Eigen::PlainObjectBase<DerivedV> &_PD2,
+                const Eigen::Matrix<int, Eigen::Dynamic, 1>&_Handle_Singular,
+                const MeshSystemInfo &_Handle_SystemInfo
+):
+V(_V),
+F(_F),
+Vcut(_Vcut),
+Fcut(_Fcut),
+TT(_TT),
+TTi(_TTi),
+PD1(_PD1),
+PD2(_PD2),
+Handle_Singular(_Handle_Singular),
+Handle_SystemInfo(_Handle_SystemInfo)
+{
+  UV        = Eigen::MatrixXd(V.rows(),2);
+  WUV       = Eigen::MatrixXd(F.rows(),6);
+  UV_out    = Eigen::MatrixXd(Vcut.rows(),2);
+  igl::vertex_triangle_adjacency(V,F,VF,VFi);
+}
+
+///START COMMON MATH FUNCTIONS
+///return the complex encoding the rotation
+///for a given missmatch interval
+template <typename DerivedV, typename DerivedF>
+IGL_INLINE std::complex<double> igl::comiso::PoissonSolver<DerivedV, DerivedF>::GetRotationComplex(int interval)
+{
+  assert((interval>=0)&&(interval<4));
+
+  switch(interval)
+  {
+    case 0:return std::complex<double>(1,0);
+    case 1:return std::complex<double>(0,1);
+    case 2:return std::complex<double>(-1,0);
+    default:return std::complex<double>(0,-1);
+  }
+}
+
+///END COMMON MATH FUNCTIONS
+
+///START FIXING VERTICES
+///set a given vertex as fixed
+template <typename DerivedV, typename DerivedF>
+IGL_INLINE void igl::comiso::PoissonSolver<DerivedV, DerivedF>::AddFixedVertex(int v)
+{
+  n_fixed_vars++;
+  Hard_constraints.push_back(v);
+}
+
+///find vertex to fix in case we're using
+///a vector field NB: multiple components not handled
+template <typename DerivedV, typename DerivedF>
+IGL_INLINE void igl::comiso::PoissonSolver<DerivedV, DerivedF>::FindFixedVertField()
+{
+  Hard_constraints.clear();
+
+  n_fixed_vars=0;
+  //fix the first singularity
+  for (unsigned int v=0;v<V.rows();v++)
+  {
+    if (Handle_Singular(v))
+    {
+      AddFixedVertex(v);
+      UV.row(v) << 0,0;
+      return;
+    }
+  }
+
+  ///if anything fixed fix the first
+  AddFixedVertex(0); // TODO HERE IT ISSSSSS
+  UV.row(0) << 0,0;
+  std::cerr << "No vertices to fix, I am fixing the first vertex to the origin!" << std::endl;
+}
+
+///find hard constraint depending if using or not
+///a vector field
+template <typename DerivedV, typename DerivedF>
+IGL_INLINE void igl::comiso::PoissonSolver<DerivedV, DerivedF>::FindFixedVert()
+{
+  Hard_constraints.clear();
+  FindFixedVertField();
+}
+
+template <typename DerivedV, typename DerivedF>
+IGL_INLINE int igl::comiso::PoissonSolver<DerivedV, DerivedF>::GetFirstVertexIndex(int v)
+{
+  return Fcut(VF[v][0],VFi[v][0]);
+}
+
+///fix the vertices which are flagged as fixed
+template <typename DerivedV, typename DerivedF>
+IGL_INLINE void igl::comiso::PoissonSolver<DerivedV, DerivedF>::FixBlockedVertex()
+{
+  int offset_row = num_cut_constraint*2;
+
+  unsigned int constr_num = 0;
+  for (unsigned int i=0;i<Hard_constraints.size();i++)
+  {
+    int v = Hard_constraints[i];
+
+    ///get first index of the vertex that must blocked
+    //int index=v->vertex_index[0];
+    int index = GetFirstVertexIndex(v);
+
+    ///multiply times 2 because of uv
+    int indexvert = index*2;
+
+    ///find the first free row to add the constraint
+    int indexRow = (offset_row+constr_num*2);
+    int indexCol = indexRow;
+
+    ///add fixing constraint LHS
+    Constraints.coeffRef(indexRow,  indexvert)   += 1;
+    Constraints.coeffRef(indexRow+1,indexvert+1) += 1;
+
+    ///add fixing constraint RHS
+    constraints_rhs[indexCol]   = UV(v,0);
+    constraints_rhs[indexCol+1] = UV(v,1);
+
+    constr_num++;
+  }
+  assert(constr_num==n_fixed_vars);
+}
+///END FIXING VERTICES
+
+///HANDLING SINGULARITY
+//set the singularity round to integer location
+template <typename DerivedV, typename DerivedF>
+IGL_INLINE void igl::comiso::PoissonSolver<DerivedV, DerivedF>::AddSingularityRound()
+{
+  for (unsigned int v=0;v<V.rows();v++)
+  {
+    if (Handle_Singular(v))
+    {
+      int index0=GetFirstVertexIndex(v);
+      ids_to_round.push_back( index0*2   );
+      ids_to_round.push_back((index0*2)+1);
+    }
+  }
+}
+
+template <typename DerivedV, typename DerivedF>
+IGL_INLINE void igl::comiso::PoissonSolver<DerivedV, DerivedF>::AddToRoundVertices(std::vector<int> ids)
+{
+  for (size_t i = 0; i < ids.size(); ++i)
+  {
+    if (ids[i] < 0 || ids[i] >= V.rows())
+      std::cerr << "WARNING: Ignored round vertex constraint, vertex " << ids[i] << " does not exist in the mesh." << std::endl;
+    int index0 = GetFirstVertexIndex(ids[i]);
+    ids_to_round.push_back( index0*2   );
+    ids_to_round.push_back((index0*2)+1);
+  }
+}
+
+///START GENERIC SYSTEM FUNCTIONS
+//build the laplacian matrix cyclyng over all rangemaps
+//and over all faces
+template <typename DerivedV, typename DerivedF>
+IGL_INLINE void igl::comiso::PoissonSolver<DerivedV, DerivedF>::BuildLaplacianMatrix(double vfscale)
+{
+  Eigen::VectorXi idx  = Eigen::VectorXi::LinSpaced(Vcut.rows(), 0, 2*Vcut.rows()-2);
+  Eigen::VectorXi idx2 = Eigen::VectorXi::LinSpaced(Vcut.rows(), 1, 2*Vcut.rows()-1);
+
+  // get gradient matrix
+  Eigen::SparseMatrix<double> G(Fcut.rows() * 3, Vcut.rows());
+  igl::grad(Vcut, Fcut, G);
+
+  // get triangle weights
+  Eigen::VectorXd dblA(Fcut.rows());
+  igl::doublearea(Vcut, Fcut, dblA);
+
+  // compute intermediate result
+  Eigen::SparseMatrix<double> G2;
+  G2 = G.transpose() * dblA.replicate<3,1>().asDiagonal() * Handle_Stiffness.replicate<3,1>().asDiagonal();
+
+  ///  Compute LHS
+  Eigen::SparseMatrix<double> Cotmatrix;
+  Cotmatrix = 0.5 * G2 * G;
+  igl::slice_into(Cotmatrix, idx,  idx,  Lhs);
+  igl::slice_into(Cotmatrix, idx2, idx2, Lhs);
+
+  /// Compute RHS
+  // reshape nrosy vectors
+  const Eigen::MatrixXd u = Eigen::Map<const Eigen::MatrixXd>(PD1.data(),Fcut.rows()*3,1); // this mimics a reshape at the cost of a copy.
+  const Eigen::MatrixXd v = Eigen::Map<const Eigen::MatrixXd>(PD2.data(),Fcut.rows()*3,1); // this mimics a reshape at the cost of a copy.
+
+  // multiply with weights
+  Eigen::VectorXd rhs1 =  G2 * u * 0.5 * vfscale;
+  Eigen::VectorXd rhs2 = -G2 * v * 0.5 * vfscale;
+  igl::slice_into(rhs1, idx,  1, rhs);
+  igl::slice_into(rhs2, idx2, 1, rhs);
+}
+
+///find different sized of the system
+template <typename DerivedV, typename DerivedF>
+IGL_INLINE void igl::comiso::PoissonSolver<DerivedV, DerivedF>::FindSizes()
+{
+  ///find the vertex that need to be fixed
+  FindFixedVert();
+
+  ///REAL PART
+  n_vert_vars = Handle_SystemInfo.num_vert_variables;
+
+  ///INTEGER PART
+  ///the total number of integer variables
+  n_integer_vars = Handle_SystemInfo.num_integer_cuts;
+
+  ///CONSTRAINT PART
+  num_cut_constraint = Handle_SystemInfo.EdgeSeamInfo.size();
+
+  num_constraint_equations = num_cut_constraint * 2 + n_fixed_vars * 2 + num_userdefined_constraint;
+
+  ///total variable of the system
+  num_total_vars = (n_vert_vars+n_integer_vars) * 2;
+
+  ///initialize matrix size
+
+  if (DEBUGPRINT)     printf("\n*** SYSTEM VARIABLES *** \n");
+  if (DEBUGPRINT)     printf("* NUM REAL VERTEX VARIABLES %d \n",n_vert_vars);
+
+  if (DEBUGPRINT)     printf("\n*** INTEGER VARIABLES *** \n");
+  if (DEBUGPRINT)     printf("* NUM INTEGER VARIABLES %d \n",(int)n_integer_vars);
+
+  if (DEBUGPRINT)     printf("\n*** CONSTRAINTS *** \n ");
+  if (DEBUGPRINT)     printf("* NUM FIXED CONSTRAINTS %d\n",n_fixed_vars);
+  if (DEBUGPRINT)     printf("* NUM CUTS CONSTRAINTS %d\n",num_cut_constraint);
+  if (DEBUGPRINT)     printf("* NUM USER DEFINED CONSTRAINTS %d\n",num_userdefined_constraint);
+
+  if (DEBUGPRINT)     printf("\n*** TOTAL SIZE *** \n");
+  if (DEBUGPRINT)     printf("* TOTAL VARIABLE SIZE (WITH INTEGER TRASL) %d \n",num_total_vars);
+  if (DEBUGPRINT)     printf("* TOTAL CONSTRAINTS %d \n",num_constraint_equations);
+}
+
+template <typename DerivedV, typename DerivedF>
+IGL_INLINE void igl::comiso::PoissonSolver<DerivedV, DerivedF>::AllocateSystem()
+{
+  Lhs.resize(n_vert_vars * 2, n_vert_vars * 2);
+  Constraints.resize(num_constraint_equations, num_total_vars);
+  rhs.resize(n_vert_vars * 2);
+  constraints_rhs.resize(num_constraint_equations);
+
+  printf("\n INITIALIZED SPARSE MATRIX OF %d x %d \n",n_vert_vars*2, n_vert_vars*2);
+  printf("\n INITIALIZED SPARSE MATRIX OF %d x %d \n",num_constraint_equations, num_total_vars);
+  printf("\n INITIALIZED VECTOR OF %d x 1 \n",n_vert_vars*2);
+  printf("\n INITIALIZED VECTOR OF %d x 1 \n",num_constraint_equations);
+}
+
+///intitialize the whole matrix
+template <typename DerivedV, typename DerivedF>
+IGL_INLINE void igl::comiso::PoissonSolver<DerivedV, DerivedF>::InitMatrix()
+{
+  FindSizes();
+  AllocateSystem();
+}
+
+///map back coordinates after that
+///the system has been solved
+template <typename DerivedV, typename DerivedF>
+IGL_INLINE void igl::comiso::PoissonSolver<DerivedV, DerivedF>::MapCoords()
+{
+  ///map coords to faces
+  for (unsigned int f=0;f<Fcut.rows();f++)
+  {
+
+    for (int k=0;k<3;k++)
+    {
+      //get the index of the variable in the system
+      int indexUV = Fcut(f,k);
+      ///then get U and V coords
+      double U=X[indexUV*2];
+      double V=X[indexUV*2+1];
+
+      WUV(f,k*2 + 0) = U;
+      WUV(f,k*2 + 1) = V;
+    }
+
+  }
+
+  for(int i = 0; i < Vcut.rows(); i++){
+    UV_out(i,0) = X[i*2];
+    UV_out(i,1) = X[i*2+1];
+  }
+#if 0
+  ///initialize the vector of integer variables to return their values
+  Handle_SystemInfo.IntegerValues.resize(n_integer_vars*2);
+  int baseIndex = (n_vert_vars)*2;
+  int endIndex  = baseIndex+n_integer_vars*2;
+  int index     = 0;
+  for (int i=baseIndex; i<endIndex; i++)
+  {
+    ///assert that the value is an integer value
+    double value=X[i];
+    double diff = value-(int)floor(value+0.5);
+    assert(diff<0.00000001);
+    Handle_SystemInfo.IntegerValues[index] = value;
+    index++;
+  }
+#endif
+}
+
+///END GENERIC SYSTEM FUNCTIONS
+
+///set the constraints for the inter-range cuts
+template <typename DerivedV, typename DerivedF>
+IGL_INLINE void igl::comiso::PoissonSolver<DerivedV, DerivedF>::BuildSeamConstraintsExplicitTranslation()
+{
+  ///current constraint row
+  int constr_row = 0;
+
+  for (unsigned int i=0; i<num_cut_constraint; i++)
+  {
+    unsigned char interval = Handle_SystemInfo.EdgeSeamInfo[i].MMatch;
+    if (interval==1)
+      interval=3;
+    else
+      if(interval==3)
+        interval=1;
+
+    int p0  = Handle_SystemInfo.EdgeSeamInfo[i].v0;
+    int p0p = Handle_SystemInfo.EdgeSeamInfo[i].v0p;
+
+    std::complex<double> rot = GetRotationComplex(interval);
+
+    ///get the integer variable
+    int integerVar = n_vert_vars + Handle_SystemInfo.EdgeSeamInfo[i].integerVar;
+
+    if (integer_rounding)
+    {
+      ids_to_round.push_back(integerVar*2);
+      ids_to_round.push_back(integerVar*2+1);
+    }
+
+    // cross boundary compatibility conditions
+    Constraints.coeffRef(constr_row,   2*p0)   +=  rot.real();
+    Constraints.coeffRef(constr_row,   2*p0+1) += -rot.imag();
+    Constraints.coeffRef(constr_row+1, 2*p0)   +=  rot.imag();
+    Constraints.coeffRef(constr_row+1, 2*p0+1) +=  rot.real();
+
+    Constraints.coeffRef(constr_row,   2*p0p)   += -1;
+    Constraints.coeffRef(constr_row+1, 2*p0p+1) += -1;
+
+    Constraints.coeffRef(constr_row,   2*integerVar)   += 1;
+    Constraints.coeffRef(constr_row+1, 2*integerVar+1) += 1;
+
+    constraints_rhs[constr_row]   = 0;
+    constraints_rhs[constr_row+1] = 0;
+
+    constr_row += 2;
+  }
+
+}
+
+///set the constraints for the inter-range cuts
+template <typename DerivedV, typename DerivedF>
+IGL_INLINE void igl::comiso::PoissonSolver<DerivedV, DerivedF>::BuildUserDefinedConstraints()
+{
+  /// the user defined constraints are at the end
+  int offset_row = num_cut_constraint*2 + n_fixed_vars*2;
+
+  ///current constraint row
+  int constr_row = offset_row;
+
+  assert(num_userdefined_constraint == userdefined_constraints.size());
+
+  for (unsigned int i=0; i<num_userdefined_constraint; i++)
+  {
+    for (unsigned int j=0; j<userdefined_constraints[i].size()-1; ++j)
+    {
+      Constraints.coeffRef(constr_row, j) = userdefined_constraints[i][j];
+    }
+
+    constraints_rhs[constr_row] = userdefined_constraints[i][userdefined_constraints[i].size()-1];
+    constr_row +=1;
+  }
+}
+
+///call of the mixed integer solver
+template <typename DerivedV, typename DerivedF>
+IGL_INLINE void igl::comiso::PoissonSolver<DerivedV, DerivedF>::MixedIntegerSolve(double cone_grid_res,
+                                                                          bool direct_round,
+                                                                          int localIter)
+{
+  X = std::vector<double>((n_vert_vars+n_integer_vars)*2);
+  if (DEBUGPRINT)
+    printf("\n ALLOCATED X \n");
+
+  ///variables part
+  int ScalarSize = n_vert_vars*2;
+  int SizeMatrix = (n_vert_vars+n_integer_vars)*2;
+
+  ///matrix A
+  gmm::col_matrix< gmm::wsvector< double > > A(SizeMatrix,SizeMatrix); // lhs matrix variables
+
+  ///constraints part
+  int CsizeX = num_constraint_equations;
+  int CsizeY = SizeMatrix+1;
+  gmm::row_matrix< gmm::wsvector< double > > C(CsizeX,CsizeY); // constraints
+
+  if (DEBUGPRINT)
+    printf("\n ALLOCATED QMM STRUCTURES \n");
+
+  std::vector<double> B(SizeMatrix,0);  // rhs
+
+  if (DEBUGPRINT)
+    printf("\n ALLOCATED RHS STRUCTURES \n");
+
+  //// copy LHS
+  for (int k=0; k < Lhs.outerSize(); ++k){
+    for (Eigen::SparseMatrix<double>::InnerIterator it(Lhs,k); it; ++it){
+      int row = it.row();
+      int col = it.col();
+      A(row, col) += it.value();
+    }
+  }
+  //// copy Constraints
+  for (int k=0; k < Constraints.outerSize(); ++k){
+    for (Eigen::SparseMatrix<double>::InnerIterator it(Constraints,k); it; ++it){
+      int row = it.row();
+      int col = it.col();
+      C(row, col) += it.value();
+    }
+  }
+
+  if (DEBUGPRINT)
+    printf("\n SET %d INTEGER VALUES \n",n_integer_vars);
+
+  ///add penalization term for integer variables
+  double penalization = 0.000001;
+  int offline_index   = ScalarSize;
+  for(unsigned int i = 0; i < (n_integer_vars)*2; ++i)
+  {
+    int index=offline_index+i;
+    A(index,index)=penalization;
+  }
+
+  if (DEBUGPRINT)
+    printf("\n SET RHS \n");
+
+  // copy RHS
+  for(int i = 0; i < (int)ScalarSize; ++i)
+  {
+    B[i] = rhs[i] * cone_grid_res;
+  }
+
+  // copy constraint RHS
+  if (DEBUGPRINT)
+    printf("\n SET %d CONSTRAINTS \n",num_constraint_equations);
+
+  for(unsigned int i = 0; i < num_constraint_equations; ++i)
+  {
+    C(i, SizeMatrix) = -constraints_rhs[i] * cone_grid_res;
+  }
+
+  COMISO::ConstrainedSolver solver;
+
+  solver.misolver().set_local_iters(localIter);
+
+  solver.misolver().set_direct_rounding(direct_round);
+
+  std::sort(ids_to_round.begin(),ids_to_round.end());
+  std::vector<int>::iterator new_end=std::unique(ids_to_round.begin(),ids_to_round.end());
+  int dist=distance(ids_to_round.begin(),new_end);
+  ids_to_round.resize(dist);
+
+  solver.solve( C, A, X, B, ids_to_round, 0.0, false, false);
+}
+
+template <typename DerivedV, typename DerivedF>
+IGL_INLINE void igl::comiso::PoissonSolver<DerivedV, DerivedF>::clearUserConstraint()
+{
+  num_userdefined_constraint = 0;
+  userdefined_constraints.clear();
+}
+
+template <typename DerivedV, typename DerivedF>
+IGL_INLINE void igl::comiso::PoissonSolver<DerivedV, DerivedF>::addSharpEdgeConstraint(int fid, int vid)
+{
+  // prepare constraint
+  std::vector<int> c(Handle_SystemInfo.num_vert_variables*2 + 1);
+
+  for (size_t i = 0; i < c.size(); ++i)
+  {
+    c[i] = 0;
+  }
+
+  int v1 = Fcut(fid,vid);
+  int v2 = Fcut(fid,(vid+1)%3);
+
+  Eigen::Matrix<typename DerivedV::Scalar, 3, 1> e = Vcut.row(v2) - Vcut.row(v1);
+  e = e.normalized();
+
+  double d1 = fabs(e.dot(PD1.row(fid).normalized()));
+  double d2 = fabs(e.dot(PD2.row(fid).normalized()));
+
+  int offset = 0;
+
+  if (d1>d2)
+    offset = 1;
+
+  ids_to_round.push_back((v1 * 2) + offset);
+  ids_to_round.push_back((v2 * 2) + offset);
+
+  // add constraint
+  c[(v1 * 2) + offset] =  1;
+  c[(v2 * 2) + offset] = -1;
+
+  // add to the user-defined constraints
+  num_userdefined_constraint++;
+  userdefined_constraints.push_back(c);
+
+}
+
+
+
+template <typename DerivedV, typename DerivedF, typename DerivedU>
+IGL_INLINE igl::comiso::MIQ_class<DerivedV, DerivedF, DerivedU>::MIQ_class(const Eigen::PlainObjectBase<DerivedV> &V_,
+                                                                   const Eigen::PlainObjectBase<DerivedF> &F_,
+                                                                   const Eigen::PlainObjectBase<DerivedV> &PD1_combed,
+                                                                   const Eigen::PlainObjectBase<DerivedV> &PD2_combed,
+                                                                   const Eigen::Matrix<int, Eigen::Dynamic, 3> &Handle_MMatch,
+                                                                   const Eigen::Matrix<int, Eigen::Dynamic, 1> &Handle_Singular,
+                                                                   const Eigen::Matrix<int, Eigen::Dynamic, 3> &Handle_Seams,
+                                                                   Eigen::PlainObjectBase<DerivedU> &UV,
+                                                                   Eigen::PlainObjectBase<DerivedF> &FUV,
+                                                                   double GradientSize,
+                                                                   double Stiffness,
+                                                                   bool DirectRound,
+                                                                   int iter,
+                                                                   int localIter,
+                                                                   bool DoRound,
+                                                                   bool SingularityRound,
+                                                                   std::vector<int> roundVertices,
+                                                                   std::vector<std::vector<int> > hardFeatures):
+V(V_),
+F(F_)
+{
+  igl::cut_mesh(V, F, Handle_Seams, Vcut, Fcut);
+
+  igl::local_basis(V,F,B1,B2,B3);
+  igl::triangle_triangle_adjacency(V,F,TT,TTi);
+
+  // Prepare indexing for the linear system
+  VertexIndexing<DerivedV, DerivedF> VInd(V, F, Vcut, Fcut, TT, TTi, Handle_MMatch, Handle_Singular, Handle_Seams);
+
+  VInd.InitSeamInfo();
+
+  // Eigen::PlainObjectBase<DerivedV> PD1_combed_for_poisson, PD2_combed_for_poisson;
+  // // Rotate by 90 degrees CCW
+  // PD1_combed_for_poisson.setZero(PD1_combed.rows(),3);
+  // PD2_combed_for_poisson.setZero(PD2_combed.rows(),3);
+  // for (unsigned i=0; i<PD1_combed.rows();++i)
+  // {
+  //   double n1 = PD1_combed.row(i).norm();
+  //   double n2 = PD2_combed.row(i).norm();
+  //
+  //   double a1 = atan2(B2.row(i).dot(PD1_combed.row(i)),B1.row(i).dot(PD1_combed.row(i)));
+  //   double a2 = atan2(B2.row(i).dot(PD2_combed.row(i)),B1.row(i).dot(PD2_combed.row(i)));
+  //
+  //   // a1 += M_PI/2;
+  //   // a2 += M_PI/2;
+  //
+  //
+  //   PD1_combed_for_poisson.row(i) = cos(a1) * B1.row(i) + sin(a1) * B2.row(i);
+  //   PD2_combed_for_poisson.row(i) = cos(a2) * B1.row(i) + sin(a2) * B2.row(i);
+  //
+  //   PD1_combed_for_poisson.row(i) = PD1_combed_for_poisson.row(i).normalized() * n1;
+  //   PD2_combed_for_poisson.row(i) = PD2_combed_for_poisson.row(i).normalized() * n2;
+  // }
+
+
+  // Assemble the system and solve
+  PoissonSolver<DerivedV, DerivedF> PSolver(V,
+                                            F,
+                                            Vcut,
+                                            Fcut,
+                                            TT,
+                                            TTi,
+                                            PD1_combed,
+                                            PD2_combed,
+                                            Handle_Singular,
+                                            VInd.Handle_SystemInfo);
+  Handle_Stiffness = Eigen::VectorXd::Constant(F.rows(),1);
+
+
+  if (iter > 0) // do stiffening
+  {
+    for (int i=0;i<iter;i++)
+    {
+      PSolver.SolvePoisson(Handle_Stiffness, GradientSize,1.f,DirectRound,localIter,DoRound,SingularityRound,roundVertices,hardFeatures);
+      int nflips=NumFlips(PSolver.WUV);
+      bool folded = updateStiffeningJacobianDistorsion(GradientSize,PSolver.WUV);
+      printf("ITERATION %d FLIPS %d \n",i,nflips);
+      if (!folded)break;
+    }
+  }
+  else
+  {
+    PSolver.SolvePoisson(Handle_Stiffness,GradientSize,1.f,DirectRound,localIter,DoRound,SingularityRound,roundVertices,hardFeatures);
+  }
+
+  int nflips=NumFlips(PSolver.WUV);
+  printf("**** END OPTIMIZING #FLIPS %d  ****\n",nflips);
+
+  UV_out = PSolver.UV_out;
+  FUV_out = PSolver.Fcut;
+  fflush(stdout);
+}
+
+template <typename DerivedV, typename DerivedF, typename DerivedU>
+IGL_INLINE void igl::comiso::MIQ_class<DerivedV, DerivedF, DerivedU>::extractUV(Eigen::PlainObjectBase<DerivedU> &UV_out,
+                                                                        Eigen::PlainObjectBase<DerivedF> &FUV_out)
+{
+  UV_out = this->UV_out;
+  FUV_out = this->FUV_out;
+}
+
+template <typename DerivedV, typename DerivedF, typename DerivedU>
+IGL_INLINE int igl::comiso::MIQ_class<DerivedV, DerivedF, DerivedU>::NumFlips(const Eigen::MatrixXd& WUV)
+{
+  int numFl=0;
+  for (unsigned int i=0;i<F.rows();i++)
+  {
+    if (IsFlipped(i, WUV))
+      numFl++;
+  }
+  return numFl;
+}
+
+template <typename DerivedV, typename DerivedF, typename DerivedU>
+IGL_INLINE double igl::comiso::MIQ_class<DerivedV, DerivedF, DerivedU>::Distortion(int f, double h, const Eigen::MatrixXd& WUV)
+{
+  assert(h > 0);
+
+  Eigen::Vector2d uv0,uv1,uv2;
+
+  uv0 << WUV(f,0), WUV(f,1);
+  uv1 << WUV(f,2), WUV(f,3);
+  uv2 << WUV(f,4), WUV(f,5);
+
+  Eigen::Matrix<typename DerivedV::Scalar, 3, 1> p0 = V.row(Fcut(f,0));
+  Eigen::Matrix<typename DerivedV::Scalar, 3, 1> p1 = V.row(Fcut(f,1));
+  Eigen::Matrix<typename DerivedV::Scalar, 3, 1> p2 = V.row(Fcut(f,2));
+
+  Eigen::Matrix<typename DerivedV::Scalar, 3, 1> norm = (p1 - p0).cross(p2 - p0);
+  double area2 = norm.norm();
+  double area2_inv  = 1.0 / area2;
+  norm *= area2_inv;
+
+  if (area2 > 0)
+  {
+    // Singular values of the Jacobian
+    Eigen::Matrix<typename DerivedV::Scalar, 3, 1> neg_t0 = norm.cross(p2 - p1);
+    Eigen::Matrix<typename DerivedV::Scalar, 3, 1> neg_t1 = norm.cross(p0 - p2);
+    Eigen::Matrix<typename DerivedV::Scalar, 3, 1> neg_t2 = norm.cross(p1 - p0);
+
+    Eigen::Matrix<typename DerivedV::Scalar, 3, 1> diffu =  (neg_t0 * uv0(0) +neg_t1 *uv1(0) +  neg_t2 * uv2(0) )*area2_inv;
+    Eigen::Matrix<typename DerivedV::Scalar, 3, 1> diffv = (neg_t0 * uv0(1) + neg_t1*uv1(1) +  neg_t2*uv2(1) )*area2_inv;
+
+    // first fundamental form
+    double I00 = diffu.dot(diffu);  // guaranteed non-neg
+    double I01 = diffu.dot(diffv);  // I01 = I10
+    double I11 = diffv.dot(diffv);  // guaranteed non-neg
+
+    // eigenvalues of a 2x2 matrix
+    // [a00 a01]
+    // [a10 a11]
+    // 1/2 * [ (a00 + a11) +/- sqrt((a00 - a11)^2 + 4 a01 a10) ]
+    double trI = I00 + I11;                     // guaranteed non-neg
+    double diffDiag = I00 - I11;                // guaranteed non-neg
+    double sqrtDet = sqrt(std::max(0.0, diffDiag*diffDiag +
+                                   4 * I01 * I01)); // guaranteed non-neg
+    double sig1 = 0.5 * (trI + sqrtDet); // higher singular value
+    double sig2 = 0.5 * (trI - sqrtDet); // lower singular value
+
+    // Avoid sig2 < 0 due to numerical error
+    if (fabs(sig2) < 1.0e-8)
+      sig2 = 0;
+
+    assert(sig1 >= 0);
+    assert(sig2 >= 0);
+
+    if (sig2 < 0) {
+      printf("Distortion will be NaN! sig1^2 is negative (%lg)\n",
+             sig2);
+    }
+
+    // The singular values of the Jacobian are the sqrts of the
+    // eigenvalues of the first fundamental form.
+    sig1 = sqrt(sig1);
+    sig2 = sqrt(sig2);
+
+    // distortion
+    double tao = IsFlipped(f,WUV) ? -1 : 1;
+    double factor = tao / h;
+    double lam = fabs(factor * sig1 - 1) + fabs(factor * sig2 - 1);
+    return lam;
+  }
+  else {
+    return 10; // something "large"
+  }
+}
+
+////////////////////////////////////////////////////////////////////////////
+// Approximate the distortion laplacian using a uniform laplacian on
+//  the dual mesh:
+//      ___________
+//      \-1 / \-1 /
+//       \ / 3 \ /
+//        \-----/
+//         \-1 /
+//          \ /
+//
+//  @param[in]  f   facet on which to compute distortion laplacian
+//  @param[in]  h   scaling factor applied to cross field
+//  @return     distortion laplacian for f
+///////////////////////////////////////////////////////////////////////////
+template <typename DerivedV, typename DerivedF, typename DerivedU>
+IGL_INLINE double igl::comiso::MIQ_class<DerivedV, DerivedF, DerivedU>::LaplaceDistortion(const int f, double h, const Eigen::MatrixXd& WUV)
+{
+  double mydist = Distortion(f, h, WUV);
+  double lapl=0;
+  for (int i=0;i<3;i++)
+  {
+    if (TT(f,i) != -1)
+      lapl += (mydist - Distortion(TT(f,i), h, WUV));
+  }
+  return lapl;
+}
+
+template <typename DerivedV, typename DerivedF, typename DerivedU>
+IGL_INLINE bool igl::comiso::MIQ_class<DerivedV, DerivedF, DerivedU>::updateStiffeningJacobianDistorsion(double grad_size, const Eigen::MatrixXd& WUV)
+{
+  bool flipped = NumFlips(WUV)>0;
+
+  if (!flipped)
+    return false;
+
+  double maxL=0;
+  double maxD=0;
+
+  if (flipped)
+  {
+    const double c = 1.0;
+    const double d = 5.0;
+
+    for (unsigned int i = 0; i < Fcut.rows(); ++i)
+    {
+      double dist=Distortion(i,grad_size,WUV);
+      if (dist > maxD)
+        maxD=dist;
+
+      double absLap=fabs(LaplaceDistortion(i, grad_size,WUV));
+      if (absLap > maxL)
+        maxL = absLap;
+
+      double stiffDelta = std::min(c * absLap, d);
+
+      Handle_Stiffness[i]+=stiffDelta;
+    }
+  }
+  printf("Maximum Distorsion %4.4f \n",maxD);
+  printf("Maximum Laplacian %4.4f \n",maxL);
+  return flipped;
+}
+
+template <typename DerivedV, typename DerivedF, typename DerivedU>
+IGL_INLINE bool igl::comiso::MIQ_class<DerivedV, DerivedF, DerivedU>::IsFlipped(const Eigen::Vector2d &uv0,
+                                                                        const Eigen::Vector2d &uv1,
+                                                                        const Eigen::Vector2d &uv2)
+{
+  Eigen::Vector2d e0 = (uv1-uv0);
+  Eigen::Vector2d e1 = (uv2-uv0);
+
+  double Area = e0(0)*e1(1) - e0(1)*e1(0);
+  return (Area<=0);
+}
+
+template <typename DerivedV, typename DerivedF, typename DerivedU>
+IGL_INLINE bool igl::comiso::MIQ_class<DerivedV, DerivedF, DerivedU>::IsFlipped(
+  const int i, const Eigen::MatrixXd& WUV)
+{
+  Eigen::Vector2d uv0,uv1,uv2;
+  uv0 << WUV(i,0), WUV(i,1);
+  uv1 << WUV(i,2), WUV(i,3);
+  uv2 << WUV(i,4), WUV(i,5);
+
+  return (IsFlipped(uv0,uv1,uv2));
+}
+
+
+
+
+template <typename DerivedV, typename DerivedF, typename DerivedU>
+IGL_INLINE void igl::comiso::miq(
+  const Eigen::PlainObjectBase<DerivedV> &V,
+  const Eigen::PlainObjectBase<DerivedF> &F,
+  const Eigen::PlainObjectBase<DerivedV> &PD1_combed,
+  const Eigen::PlainObjectBase<DerivedV> &PD2_combed,
+  const Eigen::Matrix<int, Eigen::Dynamic, 3> &Handle_MMatch,
+  const Eigen::Matrix<int, Eigen::Dynamic, 1> &Handle_Singular,
+  const Eigen::Matrix<int, Eigen::Dynamic, 3> &Handle_Seams,
+  Eigen::PlainObjectBase<DerivedU> &UV,
+  Eigen::PlainObjectBase<DerivedF> &FUV,
+  double GradientSize,
+  double Stiffness,
+  bool DirectRound,
+  int iter,
+  int localIter,
+  bool DoRound,
+  bool SingularityRound,
+  std::vector<int> roundVertices,
+  std::vector<std::vector<int> > hardFeatures)
+{
+  GradientSize = GradientSize/(V.colwise().maxCoeff()-V.colwise().minCoeff()).norm();
+
+  igl::comiso::MIQ_class<DerivedV, DerivedF, DerivedU> miq(V,
+    F,
+    PD1_combed,
+    PD2_combed,
+    Handle_MMatch,
+    Handle_Singular,
+    Handle_Seams,
+    UV,
+    FUV,
+    GradientSize,
+    Stiffness,
+    DirectRound,
+    iter,
+    localIter,
+    DoRound,
+    SingularityRound,
+    roundVertices,
+    hardFeatures);
+
+  miq.extractUV(UV,FUV);
+}
+
+template <typename DerivedV, typename DerivedF, typename DerivedU>
+IGL_INLINE void igl::comiso::miq(
+    const Eigen::PlainObjectBase<DerivedV> &V,
+    const Eigen::PlainObjectBase<DerivedF> &F,
+    const Eigen::PlainObjectBase<DerivedV> &PD1,
+    const Eigen::PlainObjectBase<DerivedV> &PD2,
+    Eigen::PlainObjectBase<DerivedU> &UV,
+    Eigen::PlainObjectBase<DerivedF> &FUV,
+    double GradientSize,
+    double Stiffness,
+    bool DirectRound,
+    int iter,
+    int localIter,
+    bool DoRound,
+    bool SingularityRound,
+    std::vector<int> roundVertices,
+    std::vector<std::vector<int> > hardFeatures)
+{
+  // Eigen::MatrixXd PD2i = PD2;
+  // if (PD2i.size() == 0)
+  // {
+  // Eigen::MatrixXd B1, B2, B3;
+  // igl::local_basis(V,F,B1,B2,B3);
+  // PD2i = igl::rotate_vectors(V,Eigen::VectorXd::Constant(1,M_PI/2),B1,B2);
+  // }
+
+  Eigen::PlainObjectBase<DerivedV> BIS1, BIS2;
+  igl::compute_frame_field_bisectors(V, F, PD1, PD2, BIS1, BIS2);
+
+  Eigen::PlainObjectBase<DerivedV> BIS1_combed, BIS2_combed;
+  igl::comb_cross_field(V, F, BIS1, BIS2, BIS1_combed, BIS2_combed);
+
+  Eigen::PlainObjectBase<DerivedF> Handle_MMatch;
+  igl::cross_field_missmatch(V, F, BIS1_combed, BIS2_combed, true, Handle_MMatch);
+
+  Eigen::Matrix<int, Eigen::Dynamic, 1> isSingularity, singularityIndex;
+  igl::find_cross_field_singularities(V, F, Handle_MMatch, isSingularity, singularityIndex);
+
+  Eigen::Matrix<int, Eigen::Dynamic, 3> Handle_Seams;
+  igl::cut_mesh_from_singularities(V, F, Handle_MMatch, Handle_Seams);
+
+  Eigen::PlainObjectBase<DerivedV> PD1_combed, PD2_combed;
+  igl::comb_frame_field(V, F, PD1, PD2, BIS1_combed, BIS2_combed, PD1_combed, PD2_combed);
+
+  igl::comiso::miq(V,
+           F,
+           PD1_combed,
+           PD2_combed,
+           Handle_MMatch,
+           isSingularity,
+           Handle_Seams,
+           UV,
+           FUV,
+           GradientSize,
+           Stiffness,
+           DirectRound,
+           iter,
+           localIter,
+           DoRound,
+           SingularityRound,
+           roundVertices,
+           hardFeatures);
+
+}
+
+#ifdef IGL_STATIC_LIBRARY
+// Explicit template specialization
+template void igl::comiso::miq<Eigen::Matrix<double, -1, 3, 0, -1, 3>, Eigen::Matrix<int, -1, 3, 0, -1, 3>, Eigen::Matrix<double, -1, -1, 0, -1, -1> >(Eigen::PlainObjectBase<Eigen::Matrix<double, -1, 3, 0, -1, 3> > const&, Eigen::PlainObjectBase<Eigen::Matrix<int, -1, 3, 0, -1, 3> > const&, Eigen::PlainObjectBase<Eigen::Matrix<double, -1, 3, 0, -1, 3> > const&, Eigen::PlainObjectBase<Eigen::Matrix<double, -1, 3, 0, -1, 3> > const&, Eigen::PlainObjectBase<Eigen::Matrix<double, -1, -1, 0, -1, -1> >&, Eigen::PlainObjectBase<Eigen::Matrix<int, -1, 3, 0, -1, 3> >&, double, double, bool, int, int, bool, bool, std::vector<int, std::allocator<int> >, std::vector<std::vector<int, std::allocator<int> >, std::allocator<std::vector<int, std::allocator<int> > > >);
+template void igl::comiso::miq<Eigen::Matrix<double, -1, -1, 0, -1, -1>, Eigen::Matrix<int, -1, -1, 0, -1, -1>, Eigen::Matrix<double, -1, -1, 0, -1, -1> >(Eigen::PlainObjectBase<Eigen::Matrix<double, -1, -1, 0, -1, -1> > const&, Eigen::PlainObjectBase<Eigen::Matrix<int, -1, -1, 0, -1, -1> > const&, Eigen::PlainObjectBase<Eigen::Matrix<double, -1, -1, 0, -1, -1> > const&, Eigen::PlainObjectBase<Eigen::Matrix<double, -1, -1, 0, -1, -1> > const&, Eigen::Matrix<int, -1, 3, 0, -1, 3> const&, Eigen::Matrix<int, -1, 1, 0, -1, 1> const&, Eigen::Matrix<int, -1, 3, 0, -1, 3> const&, Eigen::PlainObjectBase<Eigen::Matrix<double, -1, -1, 0, -1, -1> >&, Eigen::PlainObjectBase<Eigen::Matrix<int, -1, -1, 0, -1, -1> >&, double, double, bool, int, int, bool, bool, std::vector<int, std::allocator<int> >, std::vector<std::vector<int, std::allocator<int> >, std::allocator<std::vector<int, std::allocator<int> > > >);
+template void igl::comiso::miq<Eigen::Matrix<double, -1, -1, 0, -1, -1>, Eigen::Matrix<int, -1, -1, 0, -1, -1>, Eigen::Matrix<double, -1, -1, 0, -1, -1> >(Eigen::PlainObjectBase<Eigen::Matrix<double, -1, -1, 0, -1, -1> > const&, Eigen::PlainObjectBase<Eigen::Matrix<int, -1, -1, 0, -1, -1> > const&, Eigen::PlainObjectBase<Eigen::Matrix<double, -1, -1, 0, -1, -1> > const&, Eigen::PlainObjectBase<Eigen::Matrix<double, -1, -1, 0, -1, -1> > const&, Eigen::PlainObjectBase<Eigen::Matrix<double, -1, -1, 0, -1, -1> >&, Eigen::PlainObjectBase<Eigen::Matrix<int, -1, -1, 0, -1, -1> >&, double, double, bool, int, int, bool, bool, std::vector<int, std::allocator<int> >, std::vector<std::vector<int, std::allocator<int> >, std::allocator<std::vector<int, std::allocator<int> > > >);
+#endif

+ 0 - 1
include/igl/comiso/miq.cpp.REMOVED.git-id

@@ -1 +0,0 @@
-52e740e1c4b8ff9631c7f0a84e89f2fa45c35e11

+ 1 - 4
include/igl/comiso/miq.h

@@ -15,7 +15,7 @@ namespace igl
 {
   namespace comiso
   {
-    // Global seamless parametrization aligned with a given per-face jacobian (PD1,PD2).
+  // Global seamless parametrization aligned with a given per-face jacobian (PD1,PD2).
     // The algorithm is based on
     // "Mixed-Integer Quadrangulation" by D. Bommes, H. Zimmer, L. Kobbelt
     // ACM SIGGRAPH 2009, Article No. 77 (http://dl.acm.org/citation.cfm?id=1531383)
@@ -75,11 +75,8 @@ namespace igl
       const Eigen::PlainObjectBase<DerivedF> &F,
       const Eigen::PlainObjectBase<DerivedV> &PD1_combed,
       const Eigen::PlainObjectBase<DerivedV> &PD2_combed,
-      // const Eigen::PlainObjectBase<DerivedV> &BIS1_combed,
-      // const Eigen::PlainObjectBase<DerivedV> &BIS2_combed,
       const Eigen::Matrix<int, Eigen::Dynamic, 3> &MMatch,
       const Eigen::Matrix<int, Eigen::Dynamic, 1> &Singular,
-      // const Eigen::Matrix<int, Eigen::Dynamic, 1> &SingularDegree,
       const Eigen::Matrix<int, Eigen::Dynamic, 3> &Seams,
       Eigen::PlainObjectBase<DerivedU> &UV,
       Eigen::PlainObjectBase<DerivedF> &FUV,