SSL nausparse.c Interaktion und
PortierbarkeitC

/*****************************************************************************
*                                                                            *
*  Sparse-graph-specific auxiliary source file for version 2.8 of nauty.     *
*                                                                            *
*   Copyright (2004-2022) Brendan McKay.  All rights reserved.               *
*   Subject to waivers and disclaimers in nauty.h.                           *
*                                                                            *
*   CHANGE HISTORY                                                           *
*       26-Oct-04 : initial creation                                         *
*       23-Nov-06 : dispatch uses targetcell_sg, not bestcell_sg             *
*        8-Dec-06 : add adjacencies_sg()                                     *
*       10-Nov-09 : remove types shortish and permutation                    *
*       14-Nov-09 : added copy_sg()                                          *
*       11-May-10 : use sorttemplates.c for sorting procedures               *
*       19-May-10 : add two *_tr procedures for traces.                      *
*       21-May-10 : fixes for nde,v fields becoming size_t                   *
*       23-May-10 : add sparsenauty()                                        *
*       15-Jan-12 : add TLS_ATTR attributes                                  *
*       17-Dec-15 : extend sortlists_sg() to sort weights                    *
*                 : add weights to copy_sg() and updatecan_sg()              *
*       11-Mar-16 : add cleanup_sg().  This can be used in the cleanup       *
*                   field of the dispatch vector to sort the lists of the    *
*                   canonical graph, but isn't there by default.             *
*       15-Oct-19 : fix static declaration of snwork[]                       *
*        6-Apr-21 : increase work space in sparsenauty()                     *
*       16-Nov-22 : fix an error in the Traces utility comparelab_tr()       *
*                                                                            *
*****************************************************************************/

#define TMP

/*   #define ONE_WORD_SETS  not sure about this!  See notes.txt.  */
#include "nausparse.h"

    /* macros for hash-codes: */
#define MASH(l,i) ((((l) ^ 065435) + (i)) & 077777)
    /* : expression whose long value depends only on long l and int/long i.
         Anything goes, preferably non-commutative. */

#define CLEANUP(l) ((int)((l) % 077777))
    /* : expression whose value depends on long l and is less than 077777
         when converted to int then short.  Anything goes. */

#if  MAXM==1
#define M 1
#else
#define M m
#endif

#define ACCUM(x,y)   x = (((x) + (y)) & 077777)

static const int fuzz1[] = {037541,061532,005257,026416};
static const int fuzz2[] = {006532,070236,035523,062437};

#define FUZZ1(x) ((x) ^ fuzz1[(x)&3])
#define FUZZ2(x) ((x) ^ fuzz2[(x)&3])

/* aproto: header new_nauty_protos.h */

dispatchvec dispatch_sparse =
  {isautom_sg,testcanlab_sg,updatecan_sg,refine_sg,refine_sg,cheapautom_sg,
   targetcell_sg,nausparse_freedyn,nausparse_check,init_sg,NULL};

#if !MAXN
DYNALLSTAT(short,vmark1,vmark1_sz);
DYNALLSTAT(short,vmark2,vmark2_sz);
DYNALLSTAT(int,work1,work1_sz);
DYNALLSTAT(int,work2,work2_sz);
DYNALLSTAT(int,work3,work3_sz);
DYNALLSTAT(int,work4,work4_sz);
DYNALLSTAT(set,snwork,snwork_sz);
#else
static TLS_ATTR short vmark1[MAXN];
static TLS_ATTR short vmark2[MAXN];
static TLS_ATTR int work1[MAXN];
static TLS_ATTR int work2[MAXN];
static TLS_ATTR int work3[MAXN];
static TLS_ATTR int work4[MAXN];
static TLS_ATTR set snwork[2*500*MAXM];
#endif

static TLS_ATTR short vmark1_val = 32000;
#define MARK1(i) vmark1[i] = vmark1_val
#define UNMARK1(i) vmark1[i] = 0
#define ISMARKED1(i) (vmark1[i] == vmark1_val)
#define ISNOTMARKED1(i) (vmark1[i] != vmark1_val)

static TLS_ATTR short vmark2_val = 32000;
#define MARK2(i) vmark2[i] = vmark2_val
#define UNMARK2(i) vmark2[i] = 0
#define ISMARKED2(i) (vmark2[i] == vmark2_val)
#define ISNOTMARKED2(i) (vmark2[i] != vmark2_val)

#if !MAXN
#define RESETMARKS1 {if (vmark1_val++ >= 32000) \
    {size_t ij; for (ij=0;ij<vmark1_sz;++ij) vmark1[ij]=0; vmark1_val=1;}}
#define PREPAREMARKS1(nn) preparemarks1(nn)
#define RESETMARKS2 {if (vmark2_val++ >= 32000) \
    {size_t ij; for (ij=0;ij<vmark2_sz;++ij) vmark2[ij]=0; vmark2_val=1;}}
#define PREPAREMARKS2(nn) preparemarks2(nn)
#else
#define RESETMARKS1 {if (vmark1_val++ >= 32000) \
    {size_t ij; for (ij=0;ij<MAXN;++ij) vmark1[ij]=0; vmark1_val=1;}}
#define PREPAREMARKS1(nn)
#define RESETMARKS2 {if (vmark2_val++ >= 32000) \
    {size_t ij; for (ij=0;ij<MAXN;++ij) vmark2[ij]=0; vmark2_val=1;}}
#define PREPAREMARKS2(nn)
#endif

/*****************************************************************************
*                                                                            *
*  preparemarks1(N) and preparemarks2(N)                                     *
*  make vmarks array large enough to mark 0..N-1 and such that               *
*  the next RESETMARKS command will work correctly                           *
*                                                                            *
*****************************************************************************/

#if !MAXN
static void
preparemarks1(size_t nn)
{
    size_t oldsize;
    short *oldpos;

    oldsize = vmark1_sz;
    oldpos = vmark1;
    DYNALLOC1(short,vmark1,vmark1_sz,nn,"preparemarks");
    if (vmark1_sz != oldsize || vmark1 != oldpos) vmark1_val = 32000;
}
#endif

#if !MAXN
static void
preparemarks2(size_t nn)
{
    size_t oldsize;
    short *oldpos;

    oldsize = vmark2_sz;
    oldpos = vmark2;
    DYNALLOC1(short,vmark2,vmark2_sz,nn,"preparemarks");
    if (vmark2_sz != oldsize || vmark2 != oldpos) vmark2_val = 32000;
}
#endif

/*****************************************************************************
*                                                                            *
*  isautom_sg(g,perm,digraph,m,n) = TRUE iff perm is an automorphism of g    *
*  (i.e., g^perm = g).  Symmetry is assumed unless digraph = TRUE.           *
*                                                                            *
*****************************************************************************/

boolean
isautom_sg(graph *g, int *p, boolean digraph, int m, int n)
{
    int *d,*e;
    size_t *v;
    int i,pi,di;
    size_t vi,vpi,j;

    SG_VDE(g,v,d,e);

    PREPAREMARKS1(n);

    for (i = 0; i < n; ++i)
    if (p[i] != i || digraph)
    {
        pi = p[i];
        di = d[i];
        if (d[pi] != di) return FALSE;

        vi = v[i];
        vpi = v[pi];
        RESETMARKS1;
        for (j = 0; j < di; ++j) MARK1(p[e[vi+j]]);
        for (j = 0; j < di; ++j) if (ISNOTMARKED1(e[vpi+j])) return FALSE;
    }

    return TRUE;
}

/*****************************************************************************
*                                                                            *
* aresame_sg(g1,g2) = TRUE iff g1 and g2 are identical as labelled digraphs  *
*                                                                            *
*****************************************************************************/

boolean
aresame_sg(sparsegraph *g1, sparsegraph *g2)
{
    int *d1,*e1;
    int *d2,*e2;
    int n,i,di;
    size_t vi,*v1,*v2,j;

    n = g1->nv;
    if (g2->nv != n || g2->nde != g1->nde) return FALSE;

    SG_VDE(g1,v1,d1,e1);
    SG_VDE(g2,v2,d2,e2);

    PREPAREMARKS1(n);

    for (i = 0; i < n; ++i)
    {
        di = d1[i];
        if (d2[i] != di) return FALSE;

        vi = v1[i];
        RESETMARKS1;
        for (j = 0; j < di; ++j) MARK1(e1[vi+j]);
        vi = v2[i];
        for (j = 0; j < di; ++j) if (ISNOTMARKED1(e2[vi+j])) return FALSE;
    }

    return TRUE;
}

/*****************************************************************************
*                                                                            *
*  testcanlab_sg(g,canong,lab,samerows,m,n) compares g^lab to canong,        *
*  using an ordering which is immaterial since it's only used here.  The     *
*  value returned is -1,0,1 if g^lab <,=,> canong.  *samerows is set to      *
*  the number of rows (0..n) of canong which are the same as those of g^lab. *
*                                                                            *
*****************************************************************************/

int
testcanlab_sg(graph *g, graph *canong, int *lab, int *samerows, int m, int n)
{
    int *d,*e;
    int *cd,*ce;
    int i,k,di,dli;
    size_t j,vi,vli,*v,*cv;
    int mina;

    SG_VDE(g,v,d,e);
    SG_VDE(canong,cv,cd,ce);

#if !MAXN
    DYNALLOC1(int,work1,work1_sz,n,"testcanlab_sg");
#endif
#define INVLAB work1

    PREPAREMARKS1(n);

    for (i = 0; i < n; ++i) INVLAB[lab[i]] = i;

    for (i = 0; i < n; ++i)
    {
     /* compare g[lab[i]]^INVLAB to canong[i] */
        vi = cv[i];
        di = cd[i];
        vli = v[lab[i]];
        dli = d[lab[i]];

        if (di != dli)
        {
            *samerows = i;
            if (di < dli) return -1;
            return 1;
        }

        RESETMARKS1;
        mina = n;
        for (j = 0; j < di; ++j) MARK1(ce[vi+j]);
        for (j = 0; j < di; ++j)
        {
            k = INVLAB[e[vli+j]];
            if (ISMARKED1(k))  UNMARK1(k);
            else if (k < mina) mina = k;
        }
        if (mina != n)
        {
            *samerows = i;
            for (j = 0; j < di; ++j)
            {
                k = ce[vi+j];
                if (ISMARKED1(k) && k < mina) return -1;
            }
            return 1;
        }
    }

    *samerows = n;
    return 0;
}

/*****************************************************************************
*                                                                            *
*  updatecan_sg(g,canong,lab,samerows,m,n) sets canong = g^lab, assuming     *
*  the first samerows vertices of canong are ok already.  Also assumes       *
*  contiguity and ample space in canong.                                     *
*                                                                            *
*****************************************************************************/

void
updatecan_sg(graph *g, graph *canong, int *lab, int samerows, int m, int n)
{
    int *d,*e;
    int *cd,*ce;
    int i,dli;
    size_t *v,*cv,vli,j,k;
    sg_weight *wt,*cwt;

    SWG_VDE(g,v,d,e,wt);
    SWG_VDE(canong,cv,cd,ce,cwt);

#if !MAXN
    DYNALLOC1(int,work1,work1_sz,n,"testcanlab_sg");
#endif
#define INVLAB work1

    ((sparsegraph*)canong)->nv = n;
    ((sparsegraph*)canong)->nde = ((sparsegraph*)g)->nde;

    for (i = 0; i < n; ++i) INVLAB[lab[i]] = i;

    if (samerows == 0) k = 0;
    else               k = cv[samerows-1]+cd[samerows-1];

    for (i = samerows; i < n; ++i)
    {
        cv[i] = k;
        cd[i] = dli = d[lab[i]];
        vli = v[lab[i]];
        if (wt)
        {
            for (j = 0; j < dli; ++j)
            {
                ce[k] = INVLAB[e[vli+j]];
                cwt[k] = wt[vli+j];
                ++k;
            }
        }
        else
            for (j = 0; j < dli; ++j) ce[k++] = INVLAB[e[vli+j]];
    }
}

/*****************************************************************************
*                                                                            *
*  comparelab_tr(g,lab1,invlab1,lab2,invlab2,cls,col) compares               *
*  g^lab1 to g^lab2 and returns -1,0,1 according to the comparison.          *
*  invlab1[] and invlab2[] are assumed to hold inverses of lab1,lab2.        *
*                                                                            *
*****************************************************************************/

int
comparelab_tr(sparsegraph *g,
       int *lab1, int *invlab1, int *lab2, int *invlab2, int *cls, int *col)
{
    int d1,*e1,d2,*e2;
    int i,j,k,n,c,end;
    int mina;

    n = g->nv;

#if !MAXN
    DYNALLOC1(int,work1,work1_sz,n,"comparelab_tr");
#endif
#define NGHCOUNTS work1

    memset(NGHCOUNTS,0,n*sizeof(int));
    for (c=0; c<n; c+=cls[c])
    {
        if (cls[c] == 1)
        {
            end = c+cls[c];
            for (i = c; i < end; ++i)
            {
                e1 = g->e + g->v[lab1[i]];
                d1 = g->d[lab1[i]];
                e2 = g->e + g->v[lab2[i]];
                d2 = g->d[lab2[i]];
                if (d1 < d2) return -1;
                else if (d1 > d2) return 1;

                mina = n;
                for (j = 0; j < d1; ++j) {
                    (NGHCOUNTS[col[invlab1[e1[j]]]])++;
                }
                for (j = 0; j < d1; ++j)
                {
                    k = col[invlab2[e2[j]]];
                    if (NGHCOUNTS[k]) {
                        (NGHCOUNTS[k])--;
                    }
                    else {
                        if (k < mina) mina = k;
                    }
                }
                if (mina != n)
                {
                    for (j = 0; j < d1; ++j)
                    {
                        k = col[invlab1[e1[j]]];
                        if (NGHCOUNTS[k] && k < mina) return -1;
                    }
                    return 1;
                }
            }
        }
    }

    return 0;
}

/*****************************************************************************
*                                                                            *
*  testcanlab_tr(g,canong,lab,invlab,samerows) compares g^lab to canong,     *
*  using an ordering which is immaterial since it's only used here.  The     *
*  value returned is -1,0,1 if g^lab <,=,> canong.  *samerows is set to      *
*  the number of rows (0..n) of canong which are the same as those of g^lab. *
*  invlab[] is assumed to hold the inverse of lab[]                          *
*                                                                            *
*****************************************************************************/

int
testcanlab_tr(sparsegraph *g, sparsegraph *canong,
                                  int *lab, int *invlab, int *samerows)
{
    int *d,*e;
    int *cd,*ce;
    int i,di,dli;
    int k,n;
    size_t *v,*cv,vi,vli,j;
    int mina;

    SG_VDE(g,v,d,e);
    SG_VDE(canong,cv,cd,ce);
    n = g->nv;

    PREPAREMARKS1(n);

    for (i = 0; i < n; ++i)
    {
            /* compare g[lab[i]]^invlab to canong[i] */
        vi = cv[i];
        di = cd[i];
        vli = v[lab[i]];
        dli = d[lab[i]];

        if (di != dli)
        {
            *samerows = i;
            if (di < dli) return -1;
            return 1;
        }

        RESETMARKS1;
        mina = n;
        for (j = 0; j < di; ++j) MARK1(ce[vi+j]);

        for (j = 0; j < di; ++j)
        {
             k = invlab[e[vli+j]];
             if (ISMARKED1(k))  UNMARK1(k);
             else if (k < mina) mina = k;
        }
        if (mina != n)
        {
            *samerows = i;
            for (j = 0; j < di; ++j)
            {
                k = ce[vi+j];
                if (ISMARKED1(k) && k < mina) return -1;
            }
            return 1;
        }
    }

    *samerows = n;
    return 0;
}

/*****************************************************************************
*                                                                            *
*  updatecan_tr(g,canong,lab,invlab,samerows) sets canong = g^lab,           *
*  assuming the first samerows vertices of canong are ok already.            *
*  Also assumes contiguity and ample space in canong.                        *
*  Assumes invlab[] holds the inverse of lab[]                               *
*                                                                            *
*****************************************************************************/

void
updatecan_tr(sparsegraph *g, sparsegraph *canong,
                    int *lab, int *invlab, int samerows)
{
    int *d,*e;
    int *cd,*ce;
    int i,dli,n;
    size_t *v,*cv,vli,j,k;

    SG_VDE(g,v,d,e);
    SG_VDE(canong,cv,cd,ce);
    n = g->nv;

    PREPAREMARKS1(n);

    canong->nv = n;
    canong->nde = g->nde;

    if (samerows == 0) k = 0;
    else               k = cv[samerows-1]+cd[samerows-1];

    for (i = samerows; i < n; ++i)
    {
        cv[i] = k;
        cd[i] = dli = d[lab[i]];
        vli = v[lab[i]];
        for (j = 0; j < dli; ++j) ce[k++] = invlab[e[vli+j]];
    }
}

#define SORT_OF_SORT 3
#define SORT_NAME sortindirect
#define SORT_TYPE1 int
#define SORT_TYPE2 int
#include "sorttemplates.c"

#define SORT_OF_SORT 1
#define SORT_NAME sortints
#define SORT_TYPE1 int
#include "sorttemplates.c"

#define SORT_OF_SORT 2
#define SORT_NAME sortweights
#define SORT_TYPE1 int
#define SORT_TYPE2 sg_weight
#include "sorttemplates.c"

/*****************************************************************************
*                                                                            *
*  init_sg(graph *gin, graph **gout, graph *hin, graph **hout,               *
*          int *lab, int *ptn, set *active, optionblk *options,              *
*          int *status, int m, int n)                                        *
*  Initialise routine for dispatch vector.  This one just makes sure         *
*  that *hin has enough space and sets fields for n=0.                       *
*                                                                            *
*****************************************************************************/

void
init_sg(graph *gin, graph **gout, graph *hin, graph **hout, int *lab,
    int *ptn, set *active, struct optionstruct *options, int *status,
    int m, int n)
{
    sparsegraph *sg,*sh;

    if (options->getcanon)
    {
        sg = (sparsegraph*)gin;
        sh = (sparsegraph*)hin;
        SG_ALLOC(*sh,sg->nv,sg->nde,"init_sg");
        sh->nv = sg->nv;
        sh->nde = sg->nde;
    }
    *status = 0;
}

/*****************************************************************************
*                                                                            *
*  cleanup_sg(graph *gin, graph **gout, graph *hin, graph **hout,            *
*          int *lab, int *ptn, optionblk *options,                           *
*          statsblk *stats, int m, int n)                                    *
*  Cleanup routine for dispatch vector.  This one sorts the adjacency        *
*  lists for the canonical labelling.                                        *
*                                                                            *
*****************************************************************************/

void
cleanup_sg(graph *gin, graph **gout, graph *hin, graph **hout, int *lab,
           int *ptn, optionblk *options, statsblk *stats, int m, int n)
{
    sparsegraph *sh;

    if (options->getcanon
        && (stats->errstatus == 0 || stats->errstatus == NAUABORTED))
    {
        sh = (sparsegraph*)hin;
        sortlists_sg(sh);
    }
}

/*****************************************************************************
*                                                                            *
*  distvals(sparsegraph *sg, int v0, int *dist, int n) sets dist[i]          *
*  to the distance from v0 to i, for each i, or to n if there is no such     *
*  distance.  work4[] is used as a queue.                                    *
*                                                                            *
*****************************************************************************/

void
distvals(sparsegraph *g, int v0, int *dist, int n)
{
    int *d,*e;
    int i,head,tail;
    int di,k;
    size_t *v,vi,j;

    SG_VDE(g,v,d,e);
#if !MAXN
    DYNALLOC1(int,work4,work4_sz,n,"distvals");
#endif
#define QUEUE work4

    for (i = 0; i < n; ++i) dist[i] = n;

    QUEUE[0] = v0;
    dist[v0] = 0;

    head = 0;
    tail = 1;
    while (tail < n && head < tail)
    {
        i = QUEUE[head++];
        vi = v[i];
        di = d[i];
        for (j = 0; j < di; ++j)
        {
            k = e[vi+j];
            if (dist[k] == n)
            {
                dist[k] = dist[i] + 1;
                QUEUE[tail++] = k;
            }
        }
    }
}

/*****************************************************************************
*                                                                            *
*  refine_sg(g,lab,ptn,level,numcells,count,active,code,m,n) performs a      *
*  refinement operation on the partition at the specified level of the       *
*  partition nest (lab,ptn).  *numcells is assumed to contain the number of  *
*  cells on input, and is updated.  The initial set of active cells (alpha   *
*  in the paper) is specified in the set active.  Precisely, x is in active  *
*  iff the cell starting at index x in lab is active.                        *
*  The resulting partition is equitable if active is correct (see the paper  *
*  and the Guide).                                                           *
*  *code is set to a value which depends on the fine detail of the           *
*  algorithm, but which is independent of the labelling of the graph.        *
*  count is used for work space.                                             *
*                                                                            *
*****************************************************************************/

void
refine_sg(graph *g, int *lab, int *ptn, int level, int *numcells,
       int *count, set *active, int *code, int m, int n)
{
    int i,j,k,l,v1,v2,v3,isplit;
    int w1,w2,w3;
    long longcode;
    int *d,*e;
    int size,bigsize,bigpos;
    int nactive,hitcells;
    int lj,di,splitv;
    boolean trivsplit;
    size_t *v,vi,ii;

    SG_VDE(g,v,d,e);

#if !MAXN
    DYNALLOC1(int,work1,work1_sz,n,"refine_sg");
    DYNALLOC1(int,work2,work2_sz,n,"refine_sg");
    DYNALLOC1(int,work3,work3_sz,n,"refine_sg");
    DYNALLOC1(int,work4,work4_sz,n,"refine_sg");
#endif
#define CELLSTART work1
#define ACTIVE    work2
#define HITS      work3
#define HITCELL   work4

    PREPAREMARKS1(n);
    PREPAREMARKS2(n);

    longcode = *numcells;

     /* Set ACTIVE[0..nactive-1] = queue of active cell starts */

    nactive = 0;
    for (i = -1; (i = nextelement(active,m,i)) >= 0;)
        ACTIVE[nactive++] = i;

    if (nactive == 0)
    {
        *code = CLEANUP(longcode);
        return;
    }

     /* Set CELLSTART[i] = starting point in lab[] of nontrivial cell
    containing i, or n if i is a singleton */

    for (i = 0; i < n; )
    {
     /* Just here, i is a cell starting position */
        if (ptn[i] <= level)
        {
            CELLSTART[lab[i]] = n;
            ++i;
        }
        else
        {
            j = i;
            do
            {
                CELLSTART[lab[i]] = j;
            } while (ptn[i++] > level);
        }
    }

    if (level <= 2 && nactive == 1 && ptn[ACTIVE[0]] <= level
                   && *numcells <= n/8)
    {
        isplit = ACTIVE[--nactive];
        DELELEMENT(active,isplit);

        distvals((sparsegraph*)g,lab[isplit],HITS,n);

        for (v1 = 0; v1 < n; )
        {
            if (ptn[v1] <= level)
            {
                ++v1;
                continue;
            }

            longcode = MASH(longcode,v1);
            w1 = HITS[lab[v1]];

            v2 = v1+1;
            while (ptn[v2-1] > level && HITS[lab[v2]] == w1) ++v2;

            if (ptn[v2-1] <= level)
            {
                v1 = v2;
                continue;
            }

            w2 = NAUTY_INFINITY;
            v3 = j = v2;

            do
            {
                lj = lab[j];
                w3 = HITS[lj];
                if (w3 == w1)
                {
                    lab[j] = lab[v3];
                    lab[v3] = lab[v2];
                    lab[v2] = lj;
                    ++v2;
                    ++v3;
                }
                else if (w3 == w2)
                {
                    lab[j] = lab[v3];
                    lab[v3] = lj;
                    ++v3;
                }
                else if (w3 < w1)
                {
                    lab[j] = lab[v2];
                    lab[v2] = lab[v1];
                    lab[v1] = lj;
                    v3 = v2 + 1;
                    v2 = v1 + 1;
                    w2 = w1;
                    w1 = w3;
                }
                else if (w3 < w2)
                {
                    lab[j] = lab[v2];
                    lab[v2] = lj;
                    v3 = v2 + 1;
                    w2 = w3;
                }
            } while (ptn[j++] > level);

            longcode = MASH(longcode,w2);
            longcode = MASH(longcode,v2);
            if (j != v2)   /* At least two fragments
                                * v1..v2-1 = w1; v2..v3-1 = w2  */
            {
                if (v2 == v1+1)
                    CELLSTART[lab[v1]] = n;

                if (v3 == v2+1)
                    CELLSTART[lab[v2]] = n;
                else
                    for (k = v2; k < v3; ++k)
                        CELLSTART[lab[k]] = v2;
                ++*numcells;
                ptn[v2-1] = level;

                if (j == v3)
                {
                 /* Two fragments only */
                    if (v2-v1 <= v3-v2 && !ISELEMENT(active,v1))
                    {
                        ADDELEMENT(active,v1);
                        ACTIVE[nactive++] = v1;
                    }
                    else
                    {
                        ADDELEMENT(active,v2);
                        ACTIVE[nactive++] = v2;
                    }
                }
                else
                {
                 /* Extra fragments: v3..j-1 > w2 */
                    sortindirect(lab+v3,HITS,j-v3);
                    ACTIVE[nactive++] = v2;
                    ADDELEMENT(active,v2);
                    if (v2-v1 >= v3-v2)
                    {
                        bigpos = -1;
                        bigsize = v2-v1;
                    }
                    else
                    {
                        bigpos = nactive-1;
                        bigsize = v3-v2;
                    }
                    for (k = v3-1; k < j-1;)
                    {
                        ptn[k] = level;
                        longcode = MASH(longcode,k);
                        ++*numcells;
                        l = k+1;
                        ADDELEMENT(active,l);
                        ACTIVE[nactive++] = l;
                        w3 = HITS[lab[l]];
                        for (k = l; k < j-1
                                    && HITS[lab[k+1]] == w3; ++k)
                            CELLSTART[lab[k+1]] = l;
                        size = k-l+1;
                        if (size == 1)
                            CELLSTART[lab[l]] = n;
                        else
                        {
                            CELLSTART[lab[l]] = l;
                            if (size > bigsize)
                            {
                                bigsize = size;
                                bigpos = nactive-1;
                            }
                        }
                    }

                    if (bigpos >= 0 && !ISELEMENT(active,v1))
                    {
                        longcode = MASH(longcode,bigpos);
                        DELELEMENT(active,ACTIVE[bigpos]);
                        ADDELEMENT(active,v1);
                        ACTIVE[bigpos] = v1;
                    }
                }
            }
            v1 = j;
        }
    }

     /* Iterate until complete */
    while (nactive > 0 && *numcells < n)
    {
        for (i = 0; i < nactive && i < 10; ++i)
            if (ptn[ACTIVE[i]] <= level) break;

        if (i < nactive && i < 10)
        {
            trivsplit = TRUE;
            isplit = ACTIVE[i];
            ACTIVE[i] = ACTIVE[--nactive];
        }
        else
        {
            isplit = ACTIVE[--nactive];
            trivsplit = ptn[isplit] <= level;
        }

        DELELEMENT(active,isplit);
        longcode = MASH(longcode,isplit);

        if (trivsplit)
        {
            RESETMARKS1;
            RESETMARKS2;
            hitcells = 0;
            splitv = lab[isplit];
            vi = v[splitv];
            di = d[splitv];
            for (ii = 0; ii < di; ++ii)
            {
                j = e[vi+ii];
                MARK2(j);
                k = CELLSTART[j];
                if (k != n && ISNOTMARKED1(k))
                {
                    MARK1(k);
                    HITCELL[hitcells++] = k;
                }
            }

            if (hitcells > 1) sortints(HITCELL,hitcells);
            longcode = MASH(longcode,hitcells);

         /* divide cells according to which vertices are hit */

            for (i = 0; i < hitcells; ++i)
            {
                j = v1 = v2 = HITCELL[i];
                longcode = MASH(longcode,v2);
                k = 0;
                do
                {
                    lj = lab[j];
                    if (ISMARKED2(lj))
                        HITS[k++] = lj;
                    else
                        lab[v2++] = lj;
                } while (ptn[j++] > level);

                longcode = MASH(longcode,k);
                v3 = v2;
                while (--k >= 0)
                {
                    j = HITS[k];
                    CELLSTART[j] = v2;
                    lab[v3++] = j;
                }

                if (v2 != v3 && v2 != v1)
                {
                    ++*numcells;
                    if (v2 == v1+1) CELLSTART[lab[v1]] = n;
                    if (v3 == v2+1) CELLSTART[lab[v2]] = n;
                    ptn[v2-1] = level;
                    longcode = MASH(longcode,v2);
                    if (v2-v1 <= v3-v2 && !ISELEMENT(active,v1))
                    {
                        ADDELEMENT(active,v1);
                        ACTIVE[nactive++] = v1;
                    }
                    else
                    {
                        ADDELEMENT(active,v2);
                        ACTIVE[nactive++] = v2;
                    }
                }
            }
        }
        else  /* non-trivial splitting */
        {
         /* isplit is the start of the splitting cell.
            Set HITS[i] = hits of i for i in non-trivial cells,
            HITCELL[0..hitcells-1] = starts of hit non-trivial cells */

            RESETMARKS1;
            hitcells = 0;
            do
            {
                vi = v[lab[isplit]];
                di = d[lab[isplit]];
                for (ii = 0; ii < di; ++ii)
                {
                    j = e[vi+ii];
                    k = CELLSTART[j];
                    if (k != n)
                    {
                        if (ISNOTMARKED1(k))
                        {
                            MARK1(k);
                            HITCELL[hitcells++] = k;
                            do
                            {
                                HITS[lab[k]] = 0;
                            } while (ptn[k++] > level);
                        }
                        ++HITS[j];
                    }
                }
            } while (ptn[isplit++] > level);

            if (hitcells > 1) sortints(HITCELL,hitcells);

         /* divide cells according to hit counts */

            longcode = MASH(longcode,hitcells);
            for (i = 0; i < hitcells; ++i)
            {
                v1 = HITCELL[i];
                w1 = HITS[lab[v1]];
                longcode = MASH(longcode,v1);

                v2 = v1+1;
                while (ptn[v2-1] > level && HITS[lab[v2]] == w1) ++v2;

                if (ptn[v2-1] <= level) continue;
                w2 = NAUTY_INFINITY;
                v3 = j = v2;

                do
                {
                    lj = lab[j];
                    w3 = HITS[lj];
                    if (w3 == w1)
                    {
                        lab[j] = lab[v3];
                        lab[v3] = lab[v2];
                        lab[v2] = lj;
                        ++v2;
                        ++v3;
                    }
                    else if (w3 == w2)
                    {
                        lab[j] = lab[v3];
                        lab[v3] = lj;
                        ++v3;
                    }
                    else if (w3 < w1)
                    {
                        lab[j] = lab[v2];
                        lab[v2] = lab[v1];
                        lab[v1] = lj;
                        v3 = v2 + 1;
                        v2 = v1 + 1;
                        w2 = w1;
                        w1 = w3;
                    }
                    else if (w3 < w2)
                    {
                        lab[j] = lab[v2];
                        lab[v2] = lj;
                        v3 = v2 + 1;
                        w2 = w3;
                    }
                } while (ptn[j++] > level);

                longcode = MASH(longcode,w1);
                longcode = MASH(longcode,v2);
                if (j != v2)   /* At least two fragments
                                * v1..v2-1 = w1; v2..v3-1 = w2  */
                {
                    if (v2 == v1+1)
                        CELLSTART[lab[v1]] = n;

                    if (v3 == v2+1)
                        CELLSTART[lab[v2]] = n;
                    else
                        for (k = v2; k < v3; ++k)
                            CELLSTART[lab[k]] = v2;
                    ++*numcells;
                    ptn[v2-1] = level;

                    if (j == v3)
                    {
                     /* Two fragments only */
                        if (v2-v1 <= v3-v2 && !ISELEMENT(active,v1))
                        {
                            ADDELEMENT(active,v1);
                            ACTIVE[nactive++] = v1;
                        }
                        else
                        {
                            ADDELEMENT(active,v2);
                            ACTIVE[nactive++] = v2;
                        }
                    }
                    else
                    {
                     /* Extra fragments: v3..j-1 > w2 */
                        longcode = MASH(longcode,v3);
                        sortindirect(lab+v3,HITS,j-v3);
                        ACTIVE[nactive++] = v2;
                        ADDELEMENT(active,v2);
                        if (v2-v1 >= v3-v2)
                        {
                            bigpos = -1;
                            bigsize = v2-v1;
                        }
                        else
                        {
                            bigpos = nactive-1;
                            bigsize = v3-v2;
                            longcode = MASH(longcode,bigsize);
                        }
                        for (k = v3-1; k < j-1;)
                        {
                            ptn[k] = level;
                            ++*numcells;
                            l = k+1;
                            ADDELEMENT(active,l);
                            ACTIVE[nactive++] = l;
                            w3 = HITS[lab[l]];
                            longcode = MASH(longcode,w3);
                            for (k = l; k < j-1
                                        && HITS[lab[k+1]] == w3; ++k)
                                CELLSTART[lab[k+1]] = l;
                            size = k-l+1;
                            if (size == 1)
                                CELLSTART[lab[l]] = n;
                            else
                            {
                                CELLSTART[lab[l]] = l;
                                if (size > bigsize)
                                {
                                    bigsize = size;
                                    bigpos = nactive-1;
                                }
                            }
                        }

                        if (bigpos >= 0 && !ISELEMENT(active,v1))
                        {
                            DELELEMENT(active,ACTIVE[bigpos]);
                            ADDELEMENT(active,v1);
                            ACTIVE[bigpos] = v1;
                        }
                    }
                }
            }
        }
    }

    longcode = MASH(longcode,*numcells);
    *code = CLEANUP(longcode);
}

/*****************************************************************************
*                                                                            *
*  cheapautom_sg(ptn,level,digraph,n) returns TRUE if the partition at the   *
*  specified level in the partition nest (lab,ptn) {lab is not needed here}  *
*  satisfies a simple sufficient condition for its cells to be the orbits of *
*  some subgroup of the automorphism group.  Otherwise it returns FALSE.     *
*  It always returns FALSE if digraph!=FALSE.                                *
*                                                                            *
*  nauty assumes that this function will always return TRUE for any          *
*  partition finer than one for which it returns TRUE.                       *
*                                                                            *
*****************************************************************************/

boolean
cheapautom_sg(int *ptn, int level, boolean digraph, int n)
{
    int i,k,nnt;

    if (digraph) return FALSE;

    k = n;
    nnt = 0;
    for (i = 0; i < n; ++i)
    {
        --k;
        if (ptn[i] > level)
        {
            ++nnt;
            while (ptn[++i] > level) {}
        }
    }

    return (k <= nnt + 1 || k <= 4);
}

/*****************************************************************************
*                                                                            *
*  bestcell_sg(g,lab,ptn,level,tc_level,m,n) returns the index in lab of     *
*  the start of the "best non-singleton cell" for fixing.  If there is no    *
*  non-singleton cell it returns n.                                          *
*  This implementation finds the first cell which is non-trivially joined    *
*  to the greatest number of other cells, assuming equitability.             *
*  This is not good for digraphs!                                            *
*                                                                            *
*****************************************************************************/

static int
bestcell_sg(graph *g, int *lab, int *ptn, int level,
                                          int tc_level, int m, int n)
{
    int nnt;
    int *d,*e;
    int i,k,di;
    int *work1b;
    int maxcnt;
    size_t *v,vi,j;

    SG_VDE(g,v,d,e);

#if !MAXN
    DYNALLOC1(int,work1,work1_sz,n,"bestcell_sg");
    DYNALLOC1(int,work2,work2_sz,n,"bestcell_sg");
    DYNALLOC1(int,work3,work3_sz,n,"bestcell_sg");
    DYNALLOC1(int,work4,work4_sz,n,"bestcell_sg");
#endif
    work1b = work1 + (n/2);
#define START    work1
#define SIZE     work1b
#define NNTCELL  work2
#define HITS     work3
#define COUNT    work4

   /* find non-singleton cells: put starts in START[0..nnt-1],
      sizes in SIZE[0..nnt-1].
      Also NNTCELL[i] = n if {i} is a singelton, else index of
      nontriv cell containing i. */

    i = nnt = 0;

    while (i < n)
    {
        if (ptn[i] > level)
        {
            START[nnt] = i;
            j = i;
            do
                NNTCELL[lab[j]] = nnt;
            while (ptn[j++] > level);
            SIZE[nnt] = j-i;
            ++nnt;
            i = j;
        }
        else
        {
            NNTCELL[lab[i]] = n;
            ++i;
        }
    }

    if (nnt == 0) return n;

     /* set COUNT[i] to # non-trivial neighbours of n.s. cell i */

    for (i = 0; i < nnt; ++i) HITS[i] = COUNT[i] = 0;

    for (i = 0; i < nnt; ++i)
    {
        vi = v[lab[START[i]]];
        di = d[lab[START[i]]];

        for (j = 0; j < di; ++j)
        {
            k = NNTCELL[e[vi+j]];
            if (k != n) ++HITS[k];
        }
        for (j = 0; j < di; ++j)
        {
            k = NNTCELL[e[vi+j]];
            if (k != n)
            {
                if (HITS[k] > 0 && HITS[k] < SIZE[k]) ++COUNT[i];
                HITS[k] = 0;
            }
        }
    }

     /* find first greatest bucket value */

    j = 0;
    maxcnt = COUNT[0];
    for (i = 1; i < nnt; ++i)
        if (COUNT[i] > maxcnt)
        {
            j = i;
            maxcnt = COUNT[i];
        }

    return (int)START[j];
}
/*****************************************************************************
*                                                                            *
*  targetcell_sg(g,lab,ptn,level,tc_level,digraph,hint,m,n) returns the      *
*  index in lab of the next cell to split.                                   *
*  hint is a suggestion for the answer, which is obeyed if it is valid.      *
*  Otherwise we use bestcell() up to tc_level and the first non-trivial      *
*  cell after that.                                                          *
*                                                                            *
*****************************************************************************/

int
targetcell_sg(graph *g, int *lab, int *ptn, int level, int tc_level,
       boolean digraph, int hint, int m, int n)
{
    int i;

    if (hint >= 0 && ptn[hint] > level &&
                     (hint == 0 || ptn[hint-1] <= level))
        return hint;
    else if (level <= tc_level)
        return bestcell_sg(g,lab,ptn,level,tc_level,m,n);
    else
    {
        for (i = 0; i < n && ptn[i] <= level; ++i) {}
        return (i == n ? 0 : i);
    }
}

/*****************************************************************************
*                                                                            *
*  sortlists_sg(g) sorts the adjacency lists into numerical order            *
*                                                                            *
*****************************************************************************/

void
sortlists_sg(sparsegraph *g)
{
    int *d,*e;
    int n,i;
    size_t *v;
    sg_weight *wt;

    SWG_VDE(g,v,d,e,wt);
    n = g->nv;

    if (wt)
    {
        for (i = 0; i < n; ++i)
            if (d[i] > 1) sortweights(e+v[i],wt+v[i],d[i]);
    }
    else
    {
        for (i = 0; i < n; ++i)
            if (d[i] > 1) sortints(e+v[i],d[i]);
    }
}

/*****************************************************************************
*                                                                            *
*  put_sg(f,sg,digraph,linelength) writes the sparse graph to file f using   *
*  at most linelength characters per line.  If digraph then all directed     *
*  edges are written; else one v-w for w>=v is written.                      *
*                                                                            *
*****************************************************************************/

void
put_sg(FILE *f, sparsegraph *sg, boolean digraph, int linelength)
{
    int *d,*e;
    int n,di;
    int i,curlen,slen;
    size_t *v,vi,j;
    char s[12];

    SG_VDE(sg,v,d,e);
    n = sg->nv;

    for (i = 0; i < n; ++i)
    {
        vi = v[i];
        di = d[i];
        if (di == 0) continue;
        slen = itos(i+labelorg,s);
        putstring(f,s);
        putstring(f," :");
        curlen = slen + 2;

        for (j = 0; j < di; ++j)
        {
            if (!digraph && e[vi+j] < i) continue;
            slen = itos(e[vi+j]+labelorg,s);
            if (linelength && curlen + slen + 1 >= linelength)
            {
                putstring(f,"\n ");
                curlen = 2;
            }
            PUTC(' ',f);
            putstring(f,s);
            curlen += slen + 1;
        }
        PUTC('\n',f);
    }
}

/*****************************************************************************
*                                                                            *
*  sg_to_nauty(sg,g,reqm,&m) creates a nauty-format graph from a sparse      *
*  graph.  reqm is the required m value (computed from n if reqm=0), and     *
*  m is the actual value used.  g is dynamically generated if NULL is given. *
*  A pointer to g is returned.                                               *
*                                                                            *
*****************************************************************************/

graph*
sg_to_nauty(sparsegraph *sg, graph *g, int reqm, int *pm)
{
    int *d,*e;
    int m,n,i,di;
    size_t *v,vi,j;
    set *gi;

    SG_VDE(sg,v,d,e);
    n = sg->nv;
    if (reqm != 0 && reqm*WORDSIZE < n)
    {
        fprintf(ERRFILE,"sg_to_nauty: reqm is impossible\n");
        exit(1);
    }

    if (reqm != 0) m = reqm;
    else           m = (n+WORDSIZE-1)/WORDSIZE;

    *pm = m;

    if (g == NULL)
    {
        if ((g = (graph*)ALLOCS(n,m*sizeof(graph))) == NULL)
        {
            fprintf(ERRFILE,"sg_to_nauty: malloc failed\n");
            exit(1);
        }
    }

    for (i = 0, gi = g; i < n; ++i, gi += m)
    {
        vi = v[i];
        di = d[i];
        EMPTYSET(gi,m);
        for (j = 0; j < di; ++j) ADDELEMENT(gi,e[vi+j]);
    }

    return g;
}

/*****************************************************************************
*                                                                            *
*  copy_sg(sg1,sg2) makes a copy of sg1 into sg2.                            *
*  If sg2 is not NULL, it is assumed that the vlen,dlen,elen fields are      *
*  correct and v,d,e are dynamically allocated (or NULL); they are           *
*  reallocated if necessary.  If sg2==NULL, a new structure is allocated.    *
*  A pointer to the copy is returned.                                        *
*  The new graph e component is the same, no compression is done.            *
*                                                                            *
*****************************************************************************/

sparsegraph*
copy_sg(sparsegraph *sg1, sparsegraph *sg2)
{
    int *d1,*e1,*d2,*e2;
    int i,n;
    size_t *v1,*v2,k;
    sg_weight *wt1,*wt2;

    if (!sg2)
    {
        if ((sg2 = (sparsegraph*)ALLOCS(1,sizeof(sparsegraph))) == NULL)
        {
            fprintf(ERRFILE,"copy_sg: malloc failed\n");
            exit(1);
        }
        SG_INIT(*sg2);
    }

    SWG_VDE(sg1,v1,d1,e1,wt1);

    n = sg1->nv;

    k = 0;
    for (i = 0; i < n; ++i)
       if (v1[i]+d1[i]>k) k = v1[i] + d1[i];

    if (wt1)
        SWG_ALLOC(*sg2,n,k,"copy_sg malloc");
    else
    {
        SG_ALLOC(*sg2,n,k,"copy_sg malloc");
        DYNFREE(sg2->w,sg2->wlen);
    }
    SWG_VDE(sg2,v2,d2,e2,wt2);

    sg2->nv = n;
    sg2->nde = sg1->nde;
    memcpy(v2,v1,n*sizeof(size_t));
    memcpy(d2,d1,n*sizeof(int));
    memcpy(e2,e1,k*sizeof(int));
    if (wt1) memcpy(wt2,wt1,k*sizeof(sg_weight));

    return sg2;
}

/*****************************************************************************
*                                                                            *
*  nauty_to_sg(g,sg,m,n) creates a sparse graph from a nauty format graph    *
*  If sg is not NULL, it is assumed that the vlen,dlen,elen fields are       *
*  correct and v,d,e are dynamically allocated (or NULL); they are           *
*  reallocated if necessary.  If sg==NULL, a new structure is allocated.     *
*  A pointer to the sparse graph is returned.                                *
*                                                                            *
*****************************************************************************/

sparsegraph*
nauty_to_sg(graph *g, sparsegraph *sg, int m, int n)
{
    int *d,*e;
    int i,k;
    set *gi;
    size_t j,*v,nde;

    if (!sg)
    {
        if ((sg = (sparsegraph*)ALLOCS(1,sizeof(sparsegraph))) == NULL)
        {
            fprintf(ERRFILE,"nauty_to_sg: malloc failed\n");
            exit(1);
        }
        SG_INIT(*sg);
    }

    nde = 0;
    for (gi = g + (size_t)m*(size_t)n; --gi >= g; )
        if (*gi != 0) nde += POPCOUNT(*gi);

    sg->nv = n;
    sg->nde = nde;

    SG_ALLOC(*sg,n,nde,"nauty_to_sg");

    SG_VDE(sg,v,d,e);

    j = 0;
    for (i = 0, gi = g; i < n; ++i, gi += m)
    {
        v[i] = j;
        for (k = -1; (k = nextelement(gi,m,k)) >= 0; )
            e[j++] = k;
        d[i] = j - v[i];
    }

    return sg;
}

/*****************************************************************************
*                                                                            *
*  distances_sg() assigns to each vertex v a value depending on the number   *
*  of vertices at each distance from v, and what cells they lie in.          *
*  If we find any cell which is split in this manner, we don't try any       *
*  further cells.                                                            *
*                                                                            *
*****************************************************************************/

void
distances_sg(graph *g, int *lab, int *ptn, int level, int numcells, int tvpos,
         int *invar, int invararg, boolean digraph, int m, int n)
{
    int *d,*e;
    int i,k,dlim,wt;
    int di;
    int cell1,cell2,iv,liv,kcode;
    int head,tail;
    long longcode;
    size_t *v,vi,j;
    boolean success;

    SG_VDE(g,v,d,e);

#if !MAXN
    DYNALLOC1(int,work1,work1_sz,n,"distances_sg");
    DYNALLOC1(int,work4,work4_sz,n,"distances_sg");
    DYNALLOC1(int,work3,work3_sz,n,"distances_sg");
#endif
#define CELLCODE work1
#define QUEUE work4
#define DIST work3

    for (i = n; --i >= 0;) invar[i] = 0;

    wt = 1;
    for (i = 0; i < n; ++i)
    {
        CELLCODE[lab[i]] = FUZZ1(wt);
        if (ptn[i] <= level) ++wt;
    }

    if (invararg > n || invararg == 0) dlim = n;
    else                               dlim = invararg+1;

    success = FALSE;
    for (cell1 = 0; cell1 < n; cell1 = cell2 + 1)
    {
        for (cell2 = cell1; ptn[cell2] > level; ++cell2) {}
        if (cell2 == cell1) continue;

        for (iv = cell1; iv <= cell2; ++iv)
        {
            liv = lab[iv];
            QUEUE[0] = liv;
            DIST[liv] = 0;
            RESETMARKS1;
            MARK1(liv);
            longcode = 0;
            head = 0;
            tail = 1;

            while (tail < n && head < tail)
            {
                i = QUEUE[head++];
                if (DIST[i] >= dlim) break;
                vi = v[i];
                di = d[i];

                for (j = 0; j < di; ++j)
                {
                    k = e[vi+j];
                    if (ISNOTMARKED1(k))
                    {
                        MARK1(k);
                        DIST[k] = DIST[i] + 1;
                        kcode = DIST[k]+CELLCODE[k];
                        ACCUM(longcode,FUZZ1(kcode));
                        QUEUE[tail++] = k;
                    }
                }
            }
            invar[liv] = CLEANUP(longcode);
            if (invar[liv] != invar[lab[cell1]]) success = TRUE;
        }
        if (success) break;
    }
}

/*****************************************************************************
*                                                                            *
*  adjacencies_sg() assigns to each vertex v a code depending on which cells *
*  it is joined to and from, and how many times.  It is intended to provide  *
*  better partitioning that the normal refinement routine for digraphs.      *
*  It will not help with undirected graphs in nauty at all.                  *
*                                                                            *
*****************************************************************************/

void
adjacencies_sg(graph *g, int *lab, int *ptn, int level, int numcells,
               int tvpos, int *invar, int invararg, boolean digraph,
               int m, int n)
{
    int *d,*e;
    int vwt,wwt;
    int *ei,di,i;
    size_t *v,j;

    SG_VDE(g,v,d,e);

#if !MAXN
    DYNALLOC1(int,work2,work2_sz,n,"adjacencies_sg");
#endif

    vwt = 1;
    for (i = 0; i < n; ++i)
    {
        work2[lab[i]] = vwt;
        if (ptn[i] <= level) ++vwt;
        invar[i] = 0;
    }

    for (i = 0; i < n; ++i)
    {
        vwt = FUZZ1(work2[i]);
        wwt = 0;
        di = d[i];
        ei = e + v[i];
        for (j = 0; j < di; ++j)
        {
            ACCUM(wwt,FUZZ2(work2[ei[j]]));
            ACCUM(invar[ei[j]],vwt);
        }
        ACCUM(invar[i],wwt);
    }
}

/*****************************************************************************
*                                                                            *
*  sparsenauty(g,lab,ptn,orbits,&options,&stats,h)                           *
*  is a slightly simplified interface to nauty().  It allocates enough       *
*  workspace for 500 automorphisms and checks that the sparsegraph dispatch  *
*  vector is in use.                                                         *
*                                                                            *
*****************************************************************************/

void
sparsenauty(sparsegraph *g, int *lab, int *ptn, int *orbits,
            optionblk *options, statsblk *stats, sparsegraph *h)
{
    int m,n;

    if (options->dispatch != &dispatch_sparse)
    {
        fprintf(ERRFILE,"Error: sparsenauty() needs standard options block\n");
        exit(1);
    }

    n = g->nv;
    m = SETWORDSNEEDED(n);

#if !MAXN
  /*  Don't increase 2*500*m in the following without also increasing
           the static decalaration of snwork[] above. */
    DYNALLOC1(set,snwork,snwork_sz,2*500*m,"densenauty malloc");
#endif

    nauty((graph*)g,lab,ptn,NULL,orbits,options,stats,
          snwork,2*500*m,m,n,(graph*)h);
}

/*****************************************************************************
*                                                                            *
*  nausparse_check() checks that this file is compiled compatibly with the   *
*  given parameters.   If not, call exit(1).                                 *
*                                                                            *
*****************************************************************************/

void
nausparse_check(int wordsize, int m, int n, int version)
{
    if (wordsize != WORDSIZE)
    {
        fprintf(ERRFILE,"Error: WORDSIZE mismatch in nausparse.c\n");
        exit(1);
    }

#if MAXN
    if (m > MAXM)
    {
        fprintf(ERRFILE,"Error: MAXM inadequate in nausparse.c\n");
        exit(1);
    }

    if (n > MAXN)
    {
        fprintf(ERRFILE,"Error: MAXN inadequate in nausparse.c\n");
        exit(1);
    }
#endif

    if (version < NAUTYREQUIRED)
    {
        fprintf(ERRFILE,"Error: nausparse.c version mismatch\n");
        exit(1);
    }
}

/*****************************************************************************
*                                                                            *
*  nausparse_freedyn() - free the dynamic memory in this module              *
*                                                                            *
*****************************************************************************/

void
nausparse_freedyn(void)
{
#if !MAXN
    DYNFREE(vmark1,vmark1_sz);
    DYNFREE(vmark2,vmark2_sz);
    DYNFREE(work1,work1_sz);
    DYNFREE(work2,work2_sz);
    DYNFREE(work3,work3_sz);
    DYNFREE(work4,work4_sz);
    DYNFREE(snwork,snwork_sz);
#endif
}

quality73%

¤ Diese beiden folgenden Angebotsgruppen bietet das Unternehmen0.65Angebot Wie Sie bei der Firma Beratungs- und Dienstleistungen beauftragen können ¤

*Eine klare Vorstellung vom Zielzustand

Wurzel

Suchen

Beweissystem der NASA

Beweissystem Isabelle

NIST Cobol Testsuite

Cephes Mathematical Library

Wiener Entwicklungsmethode

Haftungshinweis

Die Informationen auf dieser Webseite wurden nach bestem Wissen sorgfältig zusammengestellt. Es wird jedoch weder Vollständigkeit, noch Richtigkeit, noch Qualität der bereit gestellten Informationen zugesichert.

Bemerkung:

Die farbliche Syntaxdarstellung ist noch experimentell.

SSL nausparse.c Interaktion und PortierbarkeitC

¤ Diese beiden folgenden Angebotsgruppen bietet das Unternehmen0.65Angebot Wie Sie bei der Firma Beratungs- und Dienstleistungen beauftragen können ¤

Haftungshinweis

Bemerkung:

SSL nausparse.c Interaktion und
PortierbarkeitC