X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=src%2Fmath%2Fcategoricals.c;h=d3aa372867249807618885cf697d66984a8233fc;hb=0774a97ff29339dbe87cb69c5745196d800d6584;hp=1f302a0aa78c679d17944ca779e28e84d9e85c84;hpb=46fdaed173d06e42dac6712e142bc9f7898075d7;p=pspp

diff --git a/src/math/categoricals.c b/src/math/categoricals.c
index 1f302a0aa7..d3aa372867 100644
--- a/src/math/categoricals.c
+++ b/src/math/categoricals.c
@@ -1,5 +1,5 @@
 /* PSPP - a program for statistical analysis.
-   Copyright (C) 2009 Free Software Foundation, Inc.
+   Copyright (C) 2009, 2010, 2011 Free Software Foundation, Inc.
 
    This program is free software: you can redistribute it and/or modify
    it under the terms of the GNU General Public License as published by
@@ -16,67 +16,151 @@
 
 #include <config.h>
 
-#include <stdio.h>
+#include "math/categoricals.h"
+#include "math/interaction.h"
 
-#include "categoricals.h"
+#include <stdio.h>
 
-#include <gl/xalloc.h>
-#include <data/variable.h>
-#include <data/case.h>
-#include <data/value.h>
-#include <libpspp/hmap.h>
-#include <libpspp/pool.h>
+#include "data/case.h"
+#include "data/value.h"
+#include "data/variable.h"
+#include "libpspp/array.h"
+#include "libpspp/hmap.h"
+#include "libpspp/pool.h"
+#include "libpspp/str.h"
+#include "libpspp/hash-functions.h"
 
-#include <libpspp/str.h>
+#include "gl/xalloc.h"
 
 struct value_node
 {
   struct hmap_node node;      /* Node in hash map. */
-  union value value;          /* The value being labeled. */
-  double cc;                  /* The total of the weights of cases with this value */
-  int subscript;              /* A zero based integer, unique within the variable.
-				 Can be used as an index into an array */
+
+  union value val;            /* The value */
+
+};
+
+struct interaction_value
+{
+  struct hmap_node node;        /* Node in hash map. */
+
+  struct ccase *ccase;          /* A case (probably the first in the dataset) which matches this value */
+
+  /* Total of the weights of cases matching this interaction */
+  double cc; 
+
+  void *user_data;            /* A pointer to data which the caller can store stuff */
+};
+
+static struct value_node *
+lookup_value (const struct hmap *map, const union value *val, unsigned int hash, int width)
+{
+  struct value_node *vn = NULL;
+  HMAP_FOR_EACH_WITH_HASH (vn, struct value_node, node, hash, map)
+    {
+      if (value_equal (&vn->val, val, width))
+	break;
+    }
+  
+  return vn;
+}
+
+struct variable_node
+{
+  struct hmap_node node;      /* Node in hash map. */
+  const struct variable *var; /* The variable */
+
+  struct hmap valmap;         /* A map of value nodes */
 };
 
+static void
+dump_interaction (const struct interaction *iact)
+{
+  struct string str = DS_EMPTY_INITIALIZER;
+  interaction_to_string (iact, &str);
+  printf ("Interaction: %s\n", ds_cstr (&str));
+  ds_destroy (&str);
+}
+
 
-struct var_params
+static struct variable_node *
+lookup_variable (const struct hmap *map, const struct variable *var, unsigned int hash)
 {
-  /* A map indexed by a union values */
-  struct hmap map;
+  struct variable_node *vn = NULL;
+  HMAP_FOR_EACH_WITH_HASH (vn, struct variable_node, node, hash, map)
+    {
+      if (vn->var == var)
+	break;
+      
+      fprintf (stderr, "Warning: Hash table collision\n");
+    }
+  
+  return vn;
+}
 
-  const struct variable *var;
 
-  int base_subscript;
+struct interact_params
+{
+  /* A map indexed by a interaction_value */
+  struct hmap ivmap;
 
-  /* The number of distinct values of this variable */
+  const struct interaction *iact;
+
+  int base_subscript_short;
+  int base_subscript_long;
+
+  /* The number of distinct values of this interaction */
   int n_cats;
 
-  /* A map of values indexed by subscript */
-  const struct value_node **reverse_value_map;
+  /* The degrees of freedom for this interaction */
+  int df; 
 
-  /* Total of the weights of this variable */
-  double cc; 
+  /* A map of interaction_values indexed by subscript */
+  struct interaction_value **reverse_interaction_value_map;
+
+  double cc;
 };
 
 
+/* Comparison function to sort the reverse_value_map in ascending order */
+static int
+compare_interaction_value_3way (const void *vn1_, const void *vn2_, const void *aux)
+{
+  const struct interaction_value *const *vn1p = vn1_;
+  const struct interaction_value *const *vn2p = vn2_;
+
+  const struct interact_params *iap = aux;
+
+  return interaction_case_cmp_3way (iap->iact, (*vn1p)->ccase, (*vn2p)->ccase);
+}
+
 struct categoricals
 {
   /* The weight variable */
   const struct variable *wv;
 
+  /* An array of interact_params */
+  struct interact_params *iap;
 
-  /* An array of var_params */
-  struct var_params *vp;
+  /* Map whose members are the union of the variables which comprise IAP */
+  struct hmap varmap;
 
-  /* The size of VP. (ie, the number of variables involved.) */
-  size_t n_vp;
+  /* The size of IAP. (ie, the number of interactions involved.) */
+  size_t n_iap;
 
   /* The number of categorical variables which contain entries.
-     In the absence of missing values, this will be equal to N_VP */
+     In the absence of missing values, this will be equal to N_IAP */
   size_t n_vars;
 
-  /* A map to enable the lookup of variables indexed by subscript */
-  int *reverse_variable_map;
+  size_t df_sum;
+
+  /* A map to enable the lookup of variables indexed by subscript.
+     This map considers only the N - 1 of the N variables.
+   */
+  int *reverse_variable_map_short;
+
+  /* Like the above, but uses all N variables */
+  int *reverse_variable_map_long;
 
   size_t n_cats_total;
 
@@ -84,112 +168,176 @@ struct categoricals
 
   /* Missing values to be excluded */
   enum mv_class exclude;
+
+  /* Function to be called on each update */
+  update_func *update;
+
+  /* Function specified by the caller to create user_data */
+  user_data_create_func *user_data_create;
+
+  /* Auxilliary data to be passed to update and user_data_create_func*/
+  void *aux1;
+  void *aux2;
 };
 
 
-void
-categoricals_destroy ( struct categoricals *cat)
+static void
+categoricals_dump (const struct categoricals *cat)
 {
   int i;
-  for (i = 0 ; i < cat->n_vp; ++i)
-    hmap_destroy (&cat->vp[i].map);
 
-  pool_destroy (cat->pool);
-  free (cat);
-}
+  printf ("Reverse Variable Map (short):\n");
+  for (i = 0; i < cat->df_sum; ++i)
+    {
+      printf (" %d", cat->reverse_variable_map_short[i]);
+    }
+  printf ("\n");
 
+  printf ("Reverse Variable Map (long):\n");
+  for (i = 0; i < cat->n_cats_total; ++i)
+    {
+      printf (" %d", cat->reverse_variable_map_long[i]);
+    }
+  printf ("\n");
 
-void
-categoricals_dump (const struct categoricals *cat)
-{
-  int v;
 
-  for (v = 0 ; v < cat->n_vp; ++v)
+  printf ("Number of interactions %d\n", cat->n_iap);
+  for (i = 0 ; i < cat->n_iap; ++i)
     {
-      const struct var_params *vp = &cat->vp[v];
-      const struct hmap *m = &vp->map;
-      struct hmap_node *node ;
-      int x;
-     
-      printf ("\n%s (%d)  CC=%g n_cats=%d:\n", 
-	      var_get_name (vp->var), vp->base_subscript, vp->cc, vp->n_cats);
-
-      printf ("Reverse map\n");
-      for (x = 0 ; x < vp->n_cats; ++x)
+      int v;
+      struct string str;
+      const struct interact_params *iap = &cat->iap[i];
+      const struct interaction *iact = iap->iact;
+
+      ds_init_empty (&str);
+      interaction_to_string (iact, &str);
+
+      printf ("\nInteraction: %s (n: %d; df: %d ); ", ds_cstr (&str), iap->n_cats, iap->df);
+      ds_destroy (&str);
+      printf ("Base subscript: %d\n", iap->base_subscript_short);
+
+      printf ("\t(");
+      for (v = 0; v < hmap_count (&iap->ivmap); ++v)
 	{
-	  struct string s;
-	  const struct value_node *vn = vp->reverse_value_map[x];
-	  ds_init_empty (&s);
-	  var_append_value_name (vp->var, &vn->value, &s);
-	  printf ("Value for %d is %s\n", x, ds_cstr(&s));
-	  ds_destroy (&s);
+	  int vv;
+	  const struct interaction_value *iv = iap->reverse_interaction_value_map[v];
+	  
+	  if (v > 0)  printf ("   ");
+	  printf ("{");
+	  for (vv = 0; vv < iact->n_vars; ++vv)
+	    {
+	      const struct variable *var = iact->vars[vv];
+	      const union value *val = case_data (iv->ccase, var);
+	      
+	      printf ("%g", val->f);
+	      if (vv < iact->n_vars - 1)
+		printf (", ");
+	    }
+	  printf ("}");
 	}
+      printf (")\n");
+    }
+}
 
-      printf ("\nForward map\n");
-      for (node = hmap_first (m); node; node = hmap_next (m, node))
+
+void
+categoricals_destroy (struct categoricals *cat)
+{
+  struct variable_node *vn = NULL;
+  int i;
+  if (NULL == cat)
+    return;
+  for (i = 0; i < cat->n_iap; ++i)
+    {
+      struct interaction_value *iv = NULL;
+      /* Interate over each interaction value, and unref any cases that we reffed */
+      HMAP_FOR_EACH (iv, struct interaction_value, node, &cat->iap[i].ivmap)
 	{
-	  struct string s;
-	  ds_init_empty (&s);
-	  const struct value_node *vn = HMAP_DATA (node, struct value_node, node);
-	  var_append_value_name (vp->var, &vn->value, &s);
-	  printf ("Value: %s; Index %d; CC %g\n",
-		  ds_cstr (&s),
-		  vn->subscript, vn->cc);
-	  ds_destroy (&s);
+	  case_unref (iv->ccase);
 	}
+      hmap_destroy (&cat->iap[i].ivmap);
+    }
+
+  /* Interate over each variable and delete its value map */
+  HMAP_FOR_EACH (vn, struct variable_node, node, &cat->varmap)
+    {
+      hmap_destroy (&vn->valmap);
     }
 
-  assert (cat->n_vars <= cat->n_vp);
+  hmap_destroy (&cat->varmap);
 
-  printf ("\n");
-  printf ("Number of categorical variables: %d\n", cat->n_vp);
-  printf ("Number of non-empty categorical variables: %d\n", cat->n_vars);
-  printf ("Total number of categories: %d\n", cat->n_cats_total);
+  pool_destroy (cat->pool);
+
+  free (cat);
 }
 
 
 
-static struct value_node *
-lookup_value (const struct hmap *map, const struct variable *var, const union value *val)
+static struct interaction_value *
+lookup_case (const struct hmap *map, const struct interaction *iact, const struct ccase *c)
 {
-  struct value_node *foo;
-  unsigned int width = var_get_width (var);
-  size_t hash = value_hash (val, width, 0);
+  struct interaction_value *iv = NULL;
+  size_t hash = interaction_case_hash (iact, c);
 
-  HMAP_FOR_EACH_WITH_HASH (foo, struct value_node, node, hash, map)
+  HMAP_FOR_EACH_WITH_HASH (iv, struct interaction_value, node, hash, map)
     {
-      if (value_equal (val, &foo->value, width))
+      if (interaction_case_equal (iact, c, iv->ccase))
 	break;
 
       fprintf (stderr, "Warning: Hash table collision\n");
     }
 
-  return foo;
+  return iv;
 }
 
 
-
 struct categoricals *
-categoricals_create (const struct variable **v, size_t n_vars,
-		     const struct variable *wv, enum mv_class exclude)
+categoricals_create (struct interaction *const*inter, size_t n_inter,
+		     const struct variable *wv, enum mv_class exclude,
+		     user_data_create_func *udf,
+		     update_func *update, void *aux1, void *aux2
+		     )
 {
   size_t i;
   struct categoricals *cat = xmalloc (sizeof *cat);
   
-  cat->n_vp = n_vars;
+  cat->n_iap = n_inter;
   cat->wv = wv;
   cat->n_cats_total = 0;
   cat->n_vars = 0;
-  cat->reverse_variable_map = NULL;
+  cat->reverse_variable_map_short = NULL;
+  cat->reverse_variable_map_long = NULL;
   cat->pool = pool_create ();
   cat->exclude = exclude;
+  cat->update = update;
+  cat->user_data_create = udf;
 
-  cat->vp = pool_calloc (cat->pool, cat->n_vp, sizeof *cat->vp);
+  cat->aux1 = aux1;
+  cat->aux2 = aux2;
 
-  for (i = 0 ; i < cat->n_vp; ++i)
+  cat->iap = pool_calloc (cat->pool, cat->n_iap, sizeof *cat->iap);
+
+  hmap_init (&cat->varmap);
+  for (i = 0 ; i < cat->n_iap; ++i)
     {
-      hmap_init (&cat->vp[i].map);
-      cat->vp[i].var = v[i];
+      int v;
+      hmap_init (&cat->iap[i].ivmap);
+      cat->iap[i].iact = inter[i];
+      cat->iap[i].cc = 0.0;
+      for (v = 0; v < inter[i]->n_vars; ++v)
+	{
+	  const struct variable *var = inter[i]->vars[v];
+	  unsigned int hash = hash_pointer (var, 0);
+	  struct variable_node *vn = lookup_variable (&cat->varmap, var, hash);
+	  if (vn == NULL)
+	    {
+	      vn = pool_malloc (cat->pool, sizeof *vn);
+	      vn->var = var;
+	      hmap_init (&vn->valmap);
+
+	      hmap_insert (&cat->varmap, &vn->node,  hash);
+	    }
+	}
     }
 
   return cat;
@@ -200,81 +348,99 @@ categoricals_create (const struct variable **v, size_t n_vars,
 void
 categoricals_update (struct categoricals *cat, const struct ccase *c)
 {
-  size_t i;
-  
+  int i;
+  struct variable_node *vn = NULL;
   const double weight = cat->wv ? case_data (c, cat->wv)->f : 1.0;
 
-  assert (NULL == cat->reverse_variable_map);
+  assert (NULL == cat->reverse_variable_map_short);
+  assert (NULL == cat->reverse_variable_map_long);
+
+  /* Interate over each variable, and add the value of that variable
+     to the appropriate map, if it's not already present. */
+  HMAP_FOR_EACH (vn, struct variable_node, node, &cat->varmap)
+    {
+      const int width = var_get_width (vn->var);
+      const union value *val = case_data (c, vn->var);
+      unsigned int hash = value_hash (val, width, 0);
+
+      struct value_node *valn = lookup_value (&vn->valmap, val, hash, width);
+      if (valn == NULL)
+	{
+	  valn = pool_malloc (cat->pool, sizeof *valn);
+	  value_init (&valn->val, width);
+	  value_copy (&valn->val, val, width);
+	  hmap_insert (&vn->valmap, &valn->node, hash);
+	}
+    }	  
+  
 
-  for (i = 0 ; i < cat->n_vp; ++i)
+  for (i = 0 ; i < cat->n_iap; ++i)
     {
-      const struct variable *var = cat->vp[i].var;
-      unsigned int width = var_get_width (var);
-      const union value *val = case_data (c, var);
-      size_t hash;
-      struct value_node *node ;
+      const struct interaction *iact = cat->iap[i].iact;
 
-      if ( var_is_value_missing (var, val, cat->exclude))
-	continue;
+      //      if ( interaction_case_is_missing (iact, c, cat->exclude))
+      //         continue;
 
-      hash = value_hash (val, width, 0);
-      node = lookup_value (&cat->vp[i].map, var, val);
+      size_t hash = interaction_case_hash (iact, c);
+      struct interaction_value *node = lookup_case (&cat->iap[i].ivmap, iact, c);
 
       if ( NULL == node)
 	{
 	  node = pool_malloc (cat->pool, sizeof *node);
 
-	  value_init (&node->value, width);
-	  value_copy (&node->value, val, width);
-	  node->cc = 0.0;
+	  node->ccase = case_ref (c);
+	  node->cc = weight;
 
-	  hmap_insert (&cat->vp[i].map, &node->node,  hash);
-	  cat->n_cats_total++;
-	  
-	  if ( 0 == cat->vp[i].n_cats)
-	    cat->n_vars++;
+	  hmap_insert (&cat->iap[i].ivmap, &node->node, hash);
 
-	  node->subscript = cat->vp[i].n_cats++ ;
+	  if (cat->user_data_create)
+	    node->user_data = cat->user_data_create (cat->aux1, cat->aux2);
 	}
+      else
+	{
+	  node->cc += weight;
+	}
+      cat->iap[i].cc += weight;
 
-      node->cc += weight;
-      cat->vp[i].cc += weight;
+      if (cat->update)
+      	cat->update (node->user_data, cat->exclude, cat->wv, NULL, c, cat->aux1, cat->aux2);
     }
 }
 
-/* Return the number of categories (distinct values) for variable N */
+/* Return the number of categories (distinct values) for interction N */
 size_t
 categoricals_n_count (const struct categoricals *cat, size_t n)
 {
-  return hmap_count (&cat->vp[n].map);
+  return hmap_count (&cat->iap[n].ivmap);
 }
 
 
-/* Return the index for value VAL in the Nth variable */
-int
-categoricals_index (const struct categoricals *cat, size_t n, const union value *val)
+size_t
+categoricals_df (const struct categoricals *cat, size_t n)
 {
-  struct value_node *vn = lookup_value (&cat->vp[n].map, cat->vp[n].var, val);
-
-  if ( vn == NULL)
-    return -1;
-
-  return vn->subscript;
+  return cat->iap[n].df;
 }
 
 
 /* Return the total number of categories */
 size_t
-categoricals_total (const struct categoricals *cat)
+categoricals_n_total (const struct categoricals *cat)
 {
+  assert (cat->reverse_variable_map_long);
+
   return cat->n_cats_total;
 }
 
+size_t
+categoricals_df_total (const struct categoricals *cat)
+{
+  return cat->df_sum;
+}
 
-/* This function must be called *before* any call to categoricals_get_*_by subscript an
+/* This function must be called *before* any call to categoricals_get_*_by subscript and
  *after* all calls to categoricals_update */
 void
-categoricals_done (struct categoricals *cat)
+categoricals_done (const struct categoricals *cat_)
 {
   /* Implementation Note: Whilst this function is O(n) in cat->n_cats_total, in most
      uses it will be more efficient that using a tree based structure, since it
@@ -282,67 +448,138 @@ categoricals_done (struct categoricals *cat)
 
      1 call of O(n) + 10^9 calls of O(1) is better than 10^9 calls of O(log n).
   */
+  struct categoricals *cat = CONST_CAST (struct categoricals *, cat_);
   int v;
-  int idx = 0;
-  cat->reverse_variable_map = pool_calloc (cat->pool, cat->n_cats_total, sizeof *cat->reverse_variable_map);
-  
-  for (v = 0 ; v < cat->n_vp; ++v)
+  int i;
+  int idx_short = 0;
+  int idx_long = 0;
+  cat->df_sum = 0;
+  cat->n_cats_total = 0;
+
+  /* Calculate the degrees of freedom, and the number of categories */
+  for (i = 0 ; i < cat->n_iap; ++i)
     {
-      int i;
-      struct var_params *vp = &cat->vp[v];
-      int n_cats_total = categoricals_n_count (cat, v);
-      struct hmap_node *node ;
+      const struct interaction *iact = cat->iap[i].iact;
+
+      cat->iap[i].df = 1;
+      cat->iap[i].n_cats = 1;
 
-      vp->reverse_value_map = pool_calloc (cat->pool, n_cats_total, sizeof *vp->reverse_value_map);
+      for (v = 0 ; v < iact->n_vars; ++v)
+	{
+	  const struct variable *var = iact->vars[v];
 
-      vp->base_subscript = idx;
+	  struct variable_node *vn = lookup_variable (&cat->varmap, var, hash_pointer (var, 0));
 
-      for (node = hmap_first (&vp->map); node; node = hmap_next (&vp->map, node))
+	  cat->iap[i].df *= hmap_count (&vn->valmap) - 1;
+	  cat->iap[i].n_cats *= hmap_count (&vn->valmap);
+	}
+
+      cat->df_sum += cat->iap[i].df;
+      cat->n_cats_total += cat->iap[i].n_cats;
+    }
+
+
+  cat->reverse_variable_map_short = pool_calloc (cat->pool,
+						 cat->df_sum,
+						 sizeof *cat->reverse_variable_map_short);
+
+  cat->reverse_variable_map_long = pool_calloc (cat->pool,
+						cat->n_cats_total,
+						sizeof *cat->reverse_variable_map_long);
+
+  for (i = 0 ; i < cat->n_iap; ++i)
+    {
+      struct interaction_value *ivn = NULL;
+      int x = 0;
+      int ii;
+      struct interact_params *iap = &cat->iap[i];
+
+      iap->base_subscript_short = idx_short;
+      iap->base_subscript_long = idx_long;
+
+      iap->reverse_interaction_value_map = pool_calloc (cat->pool, iap->n_cats,
+							sizeof *iap->reverse_interaction_value_map);
+
+      HMAP_FOR_EACH (ivn, struct interaction_value, node, &iap->ivmap)
 	{
-	  const struct value_node *vn = HMAP_DATA (node, struct value_node, node);
-	  vp->reverse_value_map[vn->subscript] = vn;
+	  iap->reverse_interaction_value_map[x++] = ivn;
 	}
 
-      for (i = 0; i < vp->n_cats; ++i)
-	cat->reverse_variable_map[idx++] = v;
+      assert (x <= iap->n_cats);
+
+      /* For some purposes (eg CONTRASTS in ONEWAY) the values need to be sorted */
+      sort (iap->reverse_interaction_value_map, x, sizeof (*iap->reverse_interaction_value_map),
+	    compare_interaction_value_3way, iap);
+
+      /* Fill the remaining values with null */
+      for (ii = x ; ii < iap->n_cats; ++ii)
+	iap->reverse_interaction_value_map[ii] = NULL;
+
+      /* Populate the reverse variable maps. */
+      for (ii = 0; ii < iap->df; ++ii)
+	cat->reverse_variable_map_short[idx_short++] = i;
+
+      for (ii = 0; ii < iap->n_cats; ++ii)
+	cat->reverse_variable_map_long[idx_long++] = i;
     }
 
-  assert (cat->n_vars <= cat->n_vp);
+  assert (cat->n_vars <= cat->n_iap);
+
+  //  categoricals_dump (cat);
 }
 
 
+static int
+reverse_variable_lookup_short (const struct categoricals *cat, int subscript)
+{
+  assert (cat->reverse_variable_map_short);
+  assert (subscript >= 0);
+  assert (subscript < cat->df_sum);
+
+  return cat->reverse_variable_map_short[subscript];
+}
 
-/* Return the categorical variable corresponding to SUBSCRIPT */
-const struct variable *
-categoricals_get_variable_by_subscript (const struct categoricals *cat, int subscript)
+static int
+reverse_variable_lookup_long (const struct categoricals *cat, int subscript)
 {
-  int index;
+  assert (cat->reverse_variable_map_long);
+  assert (subscript >= 0);
+  assert (subscript < cat->n_cats_total);
+
+  return cat->reverse_variable_map_long[subscript];
+}
 
-  assert (cat->reverse_variable_map);
-  
-  index = cat->reverse_variable_map[subscript];
 
-  return cat->vp[index].var;
+/* Return the interaction corresponding to SUBSCRIPT */
+const struct interaction *
+categoricals_get_interaction_by_subscript (const struct categoricals *cat, int subscript)
+{
+  int index = reverse_variable_lookup_short (cat, subscript);
+
+  return cat->iap[index].iact;
 }
 
 
-/* Return the value corresponding to SUBSCRIPT */
-const union value *
-categoricals_get_value_by_subscript (const struct categoricals *cat, int subscript)
+/* Return the case corresponding to SUBSCRIPT */
+static const struct ccase *
+categoricals_get_case_by_subscript (const struct categoricals *cat, int subscript)
 {
-  int vindex = cat->reverse_variable_map[subscript];
-  const struct var_params *vp = &cat->vp[vindex];
-  const struct value_node *vn = vp->reverse_value_map [subscript - vp->base_subscript];
+  int vindex = reverse_variable_lookup_short (cat, subscript);
+  const struct interact_params *vp = &cat->iap[vindex];
+  const struct interaction_value *vn = vp->reverse_interaction_value_map [subscript - vp->base_subscript_short];
 
-  return &vn->value;
+  if ( vn == NULL)
+    return NULL;
+
+  return vn->ccase;
 }
 
 
 double
 categoricals_get_weight_by_subscript (const struct categoricals *cat, int subscript)
 {
-  int vindex = cat->reverse_variable_map[subscript];
-  const struct var_params *vp = &cat->vp[vindex];
+  int vindex = reverse_variable_lookup_short (cat, subscript);
+  const struct interact_params *vp = &cat->iap[vindex];
 
   return vp->cc;
 }
@@ -350,13 +587,16 @@ categoricals_get_weight_by_subscript (const struct categoricals *cat, int subscr
 double
 categoricals_get_sum_by_subscript (const struct categoricals *cat, int subscript)
 {
-  int vindex = cat->reverse_variable_map[subscript];
-  const struct var_params *vp = &cat->vp[vindex];
+  int vindex = reverse_variable_lookup_short (cat, subscript);
+  const struct interact_params *vp = &cat->iap[vindex];
 
-  const struct value_node *vn = vp->reverse_value_map [subscript - vp->base_subscript];
-  return vn->cc;
-}
+  const struct interaction_value *iv = vp->reverse_interaction_value_map [subscript - vp->base_subscript_short];
 
+  if (iv == NULL)
+    return 0;
+
+  return iv->cc;
+}
 
 /* Returns unity if the value in case C at SUBSCRIPT is equal to the category
    for that subscript */
@@ -364,17 +604,41 @@ double
 categoricals_get_binary_by_subscript (const struct categoricals *cat, int subscript,
 				      const struct ccase *c)
 {
-  const struct variable *var = categoricals_get_variable_by_subscript (cat, subscript);
-  int width = var_get_width (var);
+  const struct interaction *iact = categoricals_get_interaction_by_subscript (cat, subscript);
 
-  const union value *val = case_data (c, var);
+  const struct ccase *c2 =  categoricals_get_case_by_subscript (cat, subscript);
 
-  return value_equal (val, categoricals_get_value_by_subscript (cat, subscript), width);
+  if ( c2 == NULL)
+    return 0;
+
+  return interaction_case_equal (iact, c, c2);
 }
 
 
 size_t
 categoricals_get_n_variables (const struct categoricals *cat)
 {
+  printf ("%s\n", __FUNCTION__);
   return cat->n_vars;
 }
+
+/* Return a case containing the set of values corresponding to SUBSCRIPT */
+const struct ccase *
+categoricals_get_case_by_category (const struct categoricals *cat, int subscript)
+{
+  int vindex = reverse_variable_lookup_long (cat, subscript);
+  const struct interact_params *vp = &cat->iap[vindex];
+  const struct interaction_value *vn = vp->reverse_interaction_value_map [subscript - vp->base_subscript_long];
+
+  return vn->ccase;
+}
+
+void *
+categoricals_get_user_data_by_category (const struct categoricals *cat, int subscript)
+{
+  int vindex = reverse_variable_lookup_long (cat, subscript);
+  const struct interact_params *vp = &cat->iap[vindex];
+
+  const struct interaction_value *iv = vp->reverse_interaction_value_map [subscript - vp->base_subscript_long];
+  return iv->user_data;
+}