ltlsynt implement polarity and gequiv after decomposition too

* bin/ltlsynt.cc: Also simplify subformulas using polarity and global
equivalence.  Add support for --polarity=before-decompose and
--global-equiv=before-decompose to restablish the previous behavior.
* spot/tl/apcollect.hh,
spot/tl/apcollect.cc (realizability_simplifier::merge_mapping): New
method.
* tests/core/ltlsynt.test: Add new test cases.

bin/ltlsynt.cc

@@ -104,12 +104,14 @@ static const argp_option options[] =
     { "decompose", OPT_DECOMPOSE, "yes|no", 0,
       "whether to decompose the specification as multiple output-disjoint "
       "problems to solve independently (enabled by default)", 0 },
-    { "polarity", OPT_POLARITY, "yes|no", 0,
+    { "polarity", OPT_POLARITY, "yes|no|before-decompose", 0,
       "whether to remove atomic propositions that always have the same "
-      "polarity in the formula to speed things up (enabled by default)", 0 },
+      "polarity in the formula to speed things up (enabled by default, "
-    { "global-equivalence", OPT_GEQUIV, "yes|no", 0,
+      "both before and after decomposition)", 0 },
+    { "global-equivalence", OPT_GEQUIV, "yes|no|before-decompose", 0,
       "whether to remove atomic propositions that are always equivalent to "
-      "another one (enabled by default)", 0 },
+      "another one (enabled by default, both before and after decomposition)",
+      0 },
     { "simplify", OPT_SIMPLIFY, "no|bisim|bwoa|sat|bisim-sat|bwoa-sat", 0,
       "simplification to apply to the controller (no) nothing, "
       "(bisim) bisimulation-based reduction, (bwoa) bisimulation-based "
@@ -252,9 +254,25 @@ static bool decompose_values[] =
     false, false, false, false,
   };
 ARGMATCH_VERIFY(decompose_args, decompose_values);
+static const char* const polarity_args[] =
+  {
+    "yes", "true", "enabled", "1",
+    "no", "false", "disabled", "0",
+    "before-decompose",
+    nullptr
+  };
+enum polarity_choice { pol_no, pol_yes, pol_before_decompose };
+static polarity_choice polarity_values[] =
+  {
+    pol_yes, pol_yes, pol_yes, pol_yes,
+    pol_no, pol_no, pol_no, pol_no,
+    pol_before_decompose
+  };
+ARGMATCH_VERIFY(polarity_args, polarity_values);
+
 bool opt_decompose_ltl = true;
-bool opt_polarity = true;
+polarity_choice opt_polarity = pol_yes;
-bool opt_gequiv = true;
+polarity_choice opt_gequiv = pol_yes;
 
 static const char* const simplify_args[] =
   {
@@ -407,12 +425,12 @@ namespace
 
     // Attempt to remove superfluous atomic propositions
     spot::realizability_simplifier* rs = nullptr;
-    if (opt_polarity || opt_gequiv)
+    if (opt_polarity != pol_no || opt_gequiv != pol_no)
       {
         unsigned opt = 0;
-        if (opt_polarity)
+        if (opt_polarity != pol_no)
           opt |= spot::realizability_simplifier::polarity;
-        if (opt_gequiv)
+        if (opt_gequiv != pol_no)
           {
             if (want_game())
               opt |= spot::realizability_simplifier::global_equiv_output_only;
@@ -420,9 +438,7 @@ namespace
               opt |= spot::realizability_simplifier::global_equiv;
           }
         rs =
-          new spot::realizability_simplifier(original_f,
+          new spot::realizability_simplifier(original_f, input_aps, opt,
-                                             input_aps,
-                                             opt,
                                              gi ? gi->verbose_stream : nullptr);
         f = rs->simplified_formula();
       }
@@ -493,6 +509,7 @@ namespace
     auto sub_f = sub_form.begin();
     auto sub_o = sub_outs_str.begin();
     std::vector<spot::mealy_like> mealy_machines;
+    unsigned numsubs = sub_form.size();
 
     for (; sub_f != sub_form.end(); ++sub_f, ++sub_o)
     {
@@ -502,6 +519,28 @@ namespace
                 nullptr,
                 bddfalse
               };
+
+      if (numsubs > 1 && (opt_polarity == pol_yes || opt_gequiv == pol_yes))
+        {
+          unsigned opt = 0;
+          if (opt_polarity == pol_yes)
+            opt |= spot::realizability_simplifier::polarity;
+          if (opt_gequiv == pol_yes)
+            {
+              if (want_game())
+                opt |= spot::realizability_simplifier::global_equiv_output_only;
+              else
+                opt |= spot::realizability_simplifier::global_equiv;
+            }
+          if (gi->verbose_stream)
+            *gi->verbose_stream << "working on subformula " << *sub_f << '\n';
+          spot::realizability_simplifier rsub(*sub_f, input_aps, opt,
+                                              gi ?
+                                              gi->verbose_stream : nullptr);
+          *sub_f = rsub.simplified_formula();
+          rs->merge_mapping(rsub);
+        }
+
       // If we want to print a game,
       // we never use the direct approach
       if (!want_game() && opt_bypass)
@@ -1049,7 +1088,7 @@ parse_opt(int key, char *arg, struct argp_state *)
       break;
     case OPT_GEQUIV:
       opt_gequiv = XARGMATCH("--global-equivalence", arg,
-                               decompose_args, decompose_values);
+                               polarity_args, polarity_values);
       break;
     case OPT_HIDE:
       show_status = false;
@@ -1068,7 +1107,7 @@ parse_opt(int key, char *arg, struct argp_state *)
       }
     case OPT_POLARITY:
       opt_polarity = XARGMATCH("--polarity", arg,
-                               decompose_args, decompose_values);
+                               polarity_args, polarity_values);
       break;
     case OPT_PRINT:
       opt_print_pg = true;

spot/tl/apcollect.cc

@@ -429,6 +429,13 @@ namespace spot
     while (oldf != f_);
   }
 
+  void
+  realizability_simplifier::merge_mapping(const realizability_simplifier& other)
+  {
+    for (auto [from, from_is_input, to]: other.get_mapping())
+      mapping_.emplace_back(from, from_is_input, to);
+  }
+
   void realizability_simplifier::patch_mealy(twa_graph_ptr mealy) const
   {
     bdd add = bddtrue;

spot/tl/apcollect.hh

@@ -107,6 +107,10 @@ namespace spot
       return mapping_;
     }
 
+    /// \brief Augment the current mapping with output variable renaming from
+    /// another realizability_simplifier.
+    void merge_mapping(const realizability_simplifier& other);
+
     /// \brief Patch a Mealy machine to add the missing APs.
     void patch_mealy(twa_graph_ptr mealy) const;
 

tests/core/ltlsynt.test

@@ -732,6 +732,7 @@ diff outx exp
 
 cat >exp <<EOF
 there are 2 subformulas
+working on subformula (b & (b | y)) -> y
 trying to create strategy directly for (b & (b | y)) -> y
 direct strategy might exist but was not found.
 translating formula done in X seconds
@@ -743,6 +744,7 @@ automaton has 4 states
 solving game with acceptance: all
 game solved in X seconds
 simplification took X seconds
+working on subformula (a | x) -> x
 trying to create strategy directly for (a | x) -> x
 direct strategy might exist but was not found.
 translating formula done in X seconds
@@ -784,10 +786,12 @@ diff outx exp
 # Here, G!(!x | !y) should be Gx & Gy
 cat >exp <<EOF
 there are 2 subformulas
+working on subformula Gx
 trying to create strategy directly for Gx
 direct strategy was found.
 direct strat has 1 states, 1 edges and 0 colors
 simplification took X seconds
+working on subformula Gy
 trying to create strategy directly for Gy
 direct strategy was found.
 direct strat has 1 states, 1 edges and 0 colors
@@ -802,6 +806,7 @@ diff outx exp
 # Here, !F(a | b) should be G(!a) & G(!b)
 cat >exp <<EOF
 there are 2 subformulas
+working on subformula G!a
 trying to create strategy directly for G!a
 no strategy exists.
 EOF
@@ -813,6 +818,7 @@ diff outx exp
 # Here, G!(a -> b) should be G(a) & G(!b)
 cat >exp <<EOF
 there are 2 subformulas
+working on subformula Ga
 trying to create strategy directly for Ga
 no strategy exists.
 EOF
@@ -846,6 +852,7 @@ diff outx exp
 # (a => b) & (a => c) & (a => d)
 cat >exp <<EOF
 there are 3 subformulas
+working on subformula a -> b
 trying to create strategy directly for a -> b
 direct strategy might exist but was not found.
 translating formula done in X seconds
@@ -857,6 +864,7 @@ automaton has 4 states
 solving game with acceptance: all
 game solved in X seconds
 simplification took X seconds
+working on subformula a -> c
 trying to create strategy directly for a -> c
 direct strategy might exist but was not found.
 translating formula done in X seconds
@@ -868,6 +876,7 @@ automaton has 4 states
 solving game with acceptance: all
 game solved in X seconds
 simplification took X seconds
+working on subformula a -> d
 trying to create strategy directly for a -> d
 direct strategy might exist but was not found.
 translating formula done in X seconds
@@ -889,6 +898,7 @@ diff outx exp
 # Here, !(F(a | b)) should be G!a & G!b
 cat >exp <<EOF
 there are 2 subformulas
+working on subformula G!a
 trying to create strategy directly for G!a
 no strategy exists.
 EOF
@@ -921,8 +931,10 @@ ltlsynt -f 'G(c) & (G(a) <-> GFb)' --outs=b,c --decompose=yes\
                                    --verbose --pol=no --realizability 2> out
 cat >exp <<EOF
 there are 2 subformulas
+working on subformula Gc
 trying to create strategy directly for Gc
 direct strategy was found.
+working on subformula Ga <-> GFb
 trying to create strategy directly for Ga <-> GFb
 direct strategy was found.
 EOF
@@ -932,6 +944,7 @@ ltlsynt -f 'G(c) & (G(a) <-> GFb)' --outs=b,c --decompose=yes --pol=no \
                                    --verbose --realizability --bypass=no 2> out
 cat >exp <<EOF
 there are 2 subformulas
+working on subformula Gc
 translating formula done in X seconds
 automaton has 1 states and 0 colors
 LAR construction done in X seconds
@@ -940,6 +953,7 @@ split inputs and outputs done in X seconds
 automaton has 2 states
 solving game with acceptance: all
 game solved in X seconds
+working on subformula Ga <-> GFb
 translating formula done in X seconds
 automaton has 2 states and 2 colors
 LAR construction done in X seconds
@@ -956,6 +970,7 @@ diff outx exp
 # ACD verbose
 cat >exp <<EOF
 there are 2 subformulas
+working on subformula GFa <-> GFb
 translating formula done in X seconds
 automaton has 1 states and 2 colors
 ACD construction done in X seconds
@@ -965,6 +980,7 @@ automaton has 6 states
 solving game with acceptance: generalized-Streett 1 1
 game solved in X seconds
 simplification took X seconds
+working on subformula Gc
 translating formula done in X seconds
 automaton has 1 states and 0 colors
 ACD construction done in X seconds
@@ -1147,3 +1163,140 @@ grep 'controlenv.*matches both' err
 
 ltlsynt --polarity=1 --global-e=1 -f 'G(i -> Xo) & G(!i -> F!o)' --real
 ltlsynt --polarity=0 --global-e=0 -f 'G(i -> Xo) & G(!i -> F!o)' --real
+
+cat >exp <<EOF
+there are 2 subformulas
+working on subformula G(i1 -> (o1 | !o2)) & G(i2 -> X(!o1 | o2))
+the following signals can be temporarily removed:
+  i1 := 1
+  i2 := 1
+new formula: G(o1 | !o2) & GX(!o1 | o2)
+trying to create strategy directly for G(o1 | !o2) & GX(!o1 | o2)
+direct strategy might exist but was not found.
+translating formula done in X seconds
+automaton has 2 states and 0 colors
+LAR construction done in X seconds
+DPA has 2 states, 0 colors
+split inputs and outputs done in X seconds
+automaton has 4 states
+solving game with acceptance: all
+game solved in X seconds
+simplification took X seconds
+working on subformula G(!i1 -> (o3 | !o4)) & G(!i2 -> X(!o3 | o4))
+the following signals can be temporarily removed:
+  i1 := 0
+  i2 := 0
+new formula: G(o3 | !o4) & GX(!o3 | o4)
+trying to create strategy directly for G(o3 | !o4) & GX(!o3 | o4)
+direct strategy might exist but was not found.
+translating formula done in X seconds
+automaton has 2 states and 0 colors
+LAR construction done in X seconds
+DPA has 2 states, 0 colors
+split inputs and outputs done in X seconds
+automaton has 4 states
+solving game with acceptance: all
+game solved in X seconds
+simplification took X seconds
+REALIZABLE
+HOA: v1
+States: 1
+Start: 0
+AP: 6 "o1" "o2" "o3" "o4" "i1" "i2"
+acc-name: all
+Acceptance: 0 t
+properties: trans-labels explicit-labels state-acc deterministic
+controllable-AP: 0 1 2 3
+--BODY--
+State: 0
+[!0&!1&!2&!3 | !0&!1&2&3 | 0&1&!2&!3 | 0&1&2&3] 0
+--END--
+EOF
+f1='G(i1->(o1|!o2)) & G(!i1->(o3|!o4)) & G(i2->X(!o1|o2)) & G(!i2->X(!o3|o4))'
+ltlsynt -f "$f1" --verbose 2>out 1>&2
+sed 's/ [0-9.e-]* seconds/ X seconds/g' out > outx
+diff outx exp
+
+gg='G(i2 -> (!o1 | o2)) & G(!i2 -> (!o3 | o4))'
+cat >exp <<EOF
+the following signals can be temporarily removed:
+  o5 := 1
+new formula: G(i1 -> (o1 | !o2)) & G(!i1 -> (o3 | !o4)) & $gg
+there are 2 subformulas
+working on subformula G(i1 -> (o1 | !o2)) & G(i2 -> (!o1 | o2))
+the following signals can be temporarily removed:
+  i1 := 1
+  i2 := 1
+new formula: G(o1 | !o2) & G(!o1 | o2)
+  o2 := o1
+new formula: G(o1 | !o1)
+trying to create strategy directly for G(o1 | !o1)
+direct strategy was found.
+direct strat has 1 states, 1 edges and 0 colors
+simplification took X seconds
+working on subformula G(!i1 -> (o3 | !o4)) & G(!i2 -> (!o3 | o4))
+the following signals can be temporarily removed:
+  i1 := 0
+  i2 := 0
+new formula: G(o3 | !o4) & G(!o3 | o4)
+  o4 := o3
+new formula: G(o3 | !o3)
+trying to create strategy directly for G(o3 | !o3)
+direct strategy was found.
+direct strat has 1 states, 1 edges and 0 colors
+simplification took X seconds
+REALIZABLE
+HOA: v1
+States: 1
+Start: 0
+AP: 7 "o1" "o2" "o3" "o4" "o5" "i1" "i2"
+acc-name: all
+Acceptance: 0 t
+properties: trans-labels explicit-labels state-acc deterministic weak
+controllable-AP: 0 1 2 3 4
+--BODY--
+State: 0
+[!0&!1&!2&!3&4 | !0&!1&2&3&4 | 0&1&!2&!3&4 | 0&1&2&3&4] 0
+--END--
+EOF
+f2='G(i1->(o1|!o2)) & G(!i1->(o3|!o4)) & G(i2->(!o1|o2)) & G(!i2->(!o3|o4))&Go5'
+ltlsynt -f "$f2" --verbose 2>out 1>&2
+sed 's/ [0-9.e-]* seconds/ X seconds/g' out > outx
+diff outx exp
+
+gg='G(i2->(!o1 | o2)) & G(!i2->(!o3 | o4))'
+hh='0&1&3&!5&6 | 0&!3&!4&!5&6 | !1&2&!3&5&6 | !1&!3&!4&!5&6 | '
+ii='1&!2&3&!5&6 | 1&!2&4&5&6 | !2&!3&!4&!5&6 | !2&!3&4&5&6'
+cat >exp <<EOF
+the following signals can be temporarily removed:
+  o5 := 1
+new formula: G(i1->(o1 | !o2)) & G(!i1->(o3 | !o4)) & $gg
+there are 2 subformulas
+working on subformula G(i1->(o1 | !o2)) & G(i2->(!o1 | o2))
+trying to create strategy directly for G(i1->(o1 | !o2)) & G(i2->(!o1 | o2))
+direct strategy was found.
+direct strat has 1 states, 1 edges and 0 colors
+simplification took X seconds
+working on subformula G(!i1->(o3 | !o4)) & G(!i2->(!o3 | o4))
+trying to create strategy directly for G(!i1->(o3 | !o4)) & G(!i2->(!o3 | o4))
+direct strategy was found.
+direct strat has 1 states, 1 edges and 0 colors
+simplification took X seconds
+REALIZABLE
+HOA: v1
+States: 1
+Start: 0
+AP: 7 "o1" "o2" "i1" "i2" "o3" "o4" "o5"
+acc-name: all
+Acceptance: 0 t
+properties: trans-labels explicit-labels state-acc deterministic weak
+controllable-AP: 0 1 4 5 6
+--BODY--
+State: 0
+[!0&!1&2&5&6 | !0&!1&3&!5&6 | !0&!2&4&5&6 | 0&1&2&5&6 | $hh$ii] 0
+--END--
+EOF
+f2='G(i1->(o1|!o2)) & G(!i1->(o3|!o4)) & G(i2->(!o1|o2)) & G(!i2->(!o3|o4))&Go5'
+ltlsynt -f "$f2" --polarity=before-decom --verbose 2>out 1>&2
+sed 's/ [0-9.e-]* seconds/ X seconds/g;s/ -> /->/g;' out > outx
+diff outx exp