parameter synthesis problems for one counter automata
play

Parameter-Synthesis Problems for One-Counter Automata Guillermo A. - PowerPoint PPT Presentation

Parameter-Synthesis Problems for One-Counter Automata Guillermo A. P erez (slides by Ritam Raha) INFINITY 2020 Outline 1. One-Counter Automata (Parametric) and Synthesis Problem 2. Previous Approach 3. Approach with Alternating Two-Way


  1. Parameter-Synthesis Problems for One-Counter Automata Guillermo A. P´ erez (slides by Ritam Raha) INFINITY 2020

  2. Outline 1. One-Counter Automata (Parametric) and Synthesis Problem 2. Previous Approach 3. Approach with Alternating Two-Way Automata for a subclass 4. (Failed) Approach with Partial Observation Games Guillermo A. P´ erez 2/27

  3. One-Counter Automata 1 n = 5 2 n = max(0, n - 5) 3 if n = 0: while n < 10: 4 n = n + 1 5 n = n + 100 6 # make progress 7 8 else: assert(False) 9 Guillermo A. P´ erez 3/27

  4. One-Counter Automata 1 n = 5 2 n = max(0, n - 5) 3 if n = 0: − 5 ≥ 10 +100 5 = 0 while n < 10: 4 ≥ 1 n = n + 1 5 +1 n = n + 100 6 # make progress false 7 8 else: assert(False) 9 ◮ Configurations: ( q , c ) , c ≥ 0; Guillermo A. P´ erez 3/27

  5. One-Counter Automata 1 def foobar(x): n = 5 2 n = max(0, n - x) 3 if n = 0: 4 while n < 10: 5 n = n + 1 6 n = n + 100 7 # make progress 8 else: 9 assert(False) 10 Guillermo A. P´ erez 4/27

  6. One-Counter Automata 1 def foobar(x): n = 5 2 n = max(0, n - x) 3 − x ≥ 10 5 = 0 +100 if n = 0: 4 while n < 10: 5 ≥ 1 n = n + 1 +1 6 n = n + 100 7 false # make progress 8 else: 9 assert(False) 10 Guillermo A. P´ erez 4/27

  7. Parametric One-Counter Automata X = { x 1 , . . . , x n } − x 1 ≥ x 2 + x 3 5 = 0 ≥ 1 +1 false Guillermo A. P´ erez 5/27

  8. Parametric One-Counter Automata X = { x 1 , . . . , x n } − x 1 ≥ x 2 + x 3 5 = 0 ≥ 1 +1 false Definition (Succinct OCA with Parameters) A = ( Q , q in , T , δ, X ) δ : T → Op with Op the union of Guillermo A. P´ erez 5/27

  9. Parametric One-Counter Automata X = { x 1 , . . . , x n } − x 1 ≥ x 2 + x 3 5 = 0 ≥ 1 +1 false Definition (Succinct OCA with Parameters) A = ( Q , q in , T , δ, X ) δ : T → Op with Op the union of • CU := { + a : a ∈ Z } • CT := { = 0 , ≥ a , = a : a ∈ Z } Guillermo A. P´ erez 5/27

  10. Parametric One-Counter Automata X = { x 1 , . . . , x n } − x 1 ≥ x 2 + x 3 5 = 0 ≥ 1 +1 false Definition (Succinct OCA with Parameters) A = ( Q , q in , T , δ, X ) δ : T → Op with Op the union of • CU := { + a : a ∈ Z } • PU := { + x , − x : x ∈ X } • CT := { = 0 , ≥ a , = a : a ∈ Z } • PT := { = x , ≥ x : x ∈ X } Guillermo A. P´ erez 5/27

  11. Parametric One-Counter Automata X = { x 1 , . . . , x n } − x 1 ≥ x 2 + x 3 5 = 0 ≥ 1 +1 false Definition (Succinct OCA with Parameters) A = ( Q , q in , T , δ, X ) δ : T → Op with Op the union of • CU := { + a : a ∈ Z } • PU := { + x , − x : x ∈ X } • CT := { = 0 , ≥ a , = a : a ∈ Z } • PT := { = x , ≥ x : x ∈ X } Non-parametric: X = ∅ Guillermo A. P´ erez 5/27

  12. One-Counter Automata Models CU PU ZT PT Non-Parametric SOCA ✓ ✘ ✓ ✘ OCAPT {− 1 , 0 , 1 } ✘ ✓ ✓ Parametric SOCAP ✓ ✓ ✓ ✓ Guillermo A. P´ erez 6/27

  13. Decision Problems Non-Parametric: Parametric: Guillermo A. P´ erez 7/27

  14. Decision Problems Non-Parametric: Parametric: Reach ∃ ρ such that ρ ( q in , 0) − → q f Guillermo A. P´ erez 7/27

  15. Decision Problems Non-Parametric: Parametric: Reach ∃ ρ such that ρ ( q in , 0) − → q f [NP-complete (HKOW’09)] Guillermo A. P´ erez 7/27

  16. Decision Problems Non-Parametric: Parametric: Reach Par-Reach ∃ ρ such that ∃ V : X → N s.t. ∃ ρ , ρ ρ ( q in , 0) − → q f ( q in , 0) − → V q f [NP-complete (HKOW’09)] Guillermo A. P´ erez 7/27

  17. Decision Problems Non-Parametric: Parametric: Reach Par-Reach ∃ ρ such that ∃ V : X → N s.t. ∃ ρ , ρ ρ ( q in , 0) → q f − ( q in , 0) − → V q f [NP-complete (HKOW’09)] [in NEXP (HKOW’09,LOW’15)] Guillermo A. P´ erez 7/27

  18. Decision Problems Non-Parametric: Parametric: Reach Par-Reach ∃ ρ such that ∃ V : X → N s.t. ∃ ρ , ρ ρ ( q in , 0) − → q f ( q in , 0) − → V q f [NP-complete (HKOW’09)] [in NEXP (HKOW’09,LOW’15)] UnivReach For all infinite ρ , ρ ( q in , 0) → q f − Guillermo A. P´ erez 7/27

  19. Decision Problems Non-Parametric: Parametric: Reach Par-Reach ∃ ρ such that ∃ V : X → N s.t. ∃ ρ , ρ ρ ( q in , 0) → q f − ( q in , 0) → V q f − [NP-complete (HKOW’09)] [in NEXP (HKOW’09,LOW’15)] SynthReach UnivReach ∃ V s.t. for all infinite For all infinite ρ , ρ ρ ρ , ( q in , 0) − → V q f ( q in , 0) → q f − Guillermo A. P´ erez 7/27

  20. Decision Problems Non-Parametric: Parametric: Reach Par-Reach ∃ ρ such that ∃ V : X → N s.t. ∃ ρ , ρ ρ ( q in , 0) → q f − ( q in , 0) → V q f − [NP-complete (HKOW’09)] [in NEXP (HKOW’09,LOW’15)] SynthReach UnivReach ∃ V s.t. for all infinite For all infinite ρ , ρ ρ ρ , ( q in , 0) − → V q f ( q in , 0) → q f − Guillermo A. P´ erez 7/27

  21. UnivReach for SOCA Model and problem ◮ Op = CU ∪ ZT ◮ UnivReach : all infinite paths reach q f Guillermo A. P´ erez 8/27

  22. UnivReach for SOCA Model and problem ◮ Op = CU ∪ ZT ◮ UnivReach : all infinite paths reach q f ∃ infinite path avoiding q f = two Reach queries Guillermo A. P´ erez 8/27

  23. UnivReach for SOCA Model and problem ◮ Op = CU ∪ ZT ◮ UnivReach : all infinite paths reach q f ∃ infinite path avoiding q f = two Reach queries (coNP) Guillermo A. P´ erez 8/27

  24. UnivReach for SOCA Model and problem ◮ Op = CU ∪ ZT ◮ UnivReach : all infinite paths reach q f ∃ infinite path avoiding q f = two Reach queries (coNP) Hardness from reduction from co- subsetsum . Proposition The UnivReach problem for SOCA is coNP-complete. Guillermo A. P´ erez 8/27

  25. Decision Problems Non-Parametric: Parametric: Reach Par-Reach ∃ ρ such that ∃ V : X N s.t. ∃ ρ , ρ ρ ( q in , 0) → q f − → V q f − ( q in , 0) [NP-complete (HKOW’09)] [in NEXP (HKOW’09,LOW’15)] UnivReach SynthReach For all infinite ρ , ∃ V s.t. for all infinite ρ ρ ( q in , 0) → q f − − → V q f ρ , ( q in , 0) Guillermo A. P´ erez 9/27

  26. Decision Problems Non-Parametric: Parametric: Reach Par-Reach ∃ ρ such that ∃ V : X N s.t. ∃ ρ , ρ ρ ( q in , 0) → q f − → V q f − ( q in , 0) [NP-complete (HKOW’09)] [in NEXP (HKOW’09,LOW’15)] ? UnivReach SynthReach For all infinite ρ , ∃ V s.t. for all infinite ρ ρ ( q in , 0) − → q f − → V q f ρ , ( q in , 0) [coNP-complete] Guillermo A. P´ erez 9/27

  27. Outline 1. One-Counter Automata (Parametric) and Synthesis Problem 2. Previous Approach 3. Approach with Alternating Two-Way Automata for a subclass 4. Approach with Partial Observation Games Guillermo A. P´ erez 10/27

  28. PAD & Extensions ◮ Presburger arithmetic (PA) := � N , 0 , 1 , + , < � Guillermo A. P´ erez 11/27

  29. PAD & Extensions ◮ Presburger arithmetic (PA) := � N , 0 , 1 , + , < � ◮ Presburger arithmetic with Divisibility (PAD) := PA + | ( a | b ⇐ ⇒ ∃ c ∈ Z : b = ac ) Guillermo A. P´ erez 11/27

  30. PAD & Extensions ◮ Presburger arithmetic (PA) := � N , 0 , 1 , + , < � ◮ Presburger arithmetic with Divisibility (PAD) := PA + | ( a | b ⇐ ⇒ ∃ c ∈ Z : b = ac ) Theorem (Robinson’49, Lipshitz’81) Full PAD is undecidable; one alternation suffices for undecidability. Guillermo A. P´ erez 11/27

  31. PAD & Extensions ◮ Presburger arithmetic (PA) := � N , 0 , 1 , + , < � ◮ Presburger arithmetic with Divisibility (PAD) := PA + | ( a | b ⇐ ⇒ ∃ c ∈ Z : b = ac ) Theorem (Robinson’49, Lipshitz’81) Full PAD is undecidable; one alternation suffices for undecidability. Theorem (Lipshitz’78, Lechner-Ouaknine-Worrell’15) The existential fragment of PAD (EPAD) is decidable in NEXP. Guillermo A. P´ erez 11/27

  32. ∀∃ R PAD & Undecidability ◮ ∀∃ R PAD := ∀ z 1 . . . ∀ z n ∃ x 1 . . . ∃ x m . ϕ ( x , z ) ◮ divisibilities of the form f ( z ) | g ( x , z ) Guillermo A. P´ erez 12/27

  33. ∀∃ R PAD & Undecidability ◮ ∀∃ R PAD := ∀ z 1 . . . ∀ z n ∃ x 1 . . . ∃ x m . ϕ ( x , z ) ◮ divisibilities of the form f ( z ) | g ( x , z ) ◮ ∀∃ R PAD + := ∀∃ R PAD with ¬ not allowed before divisibility. Guillermo A. P´ erez 12/27

  34. ∀∃ R PAD & Undecidability ◮ ∀∃ R PAD := ∀ z 1 . . . ∀ z n ∃ x 1 . . . ∃ x m . ϕ ( x , z ) ◮ divisibilities of the form f ( z ) | g ( x , z ) ◮ ∀∃ R PAD + := ∀∃ R PAD with ¬ not allowed before divisibility. Claim (Bozga-Iosif’05, Lechner’15) SynthReach for SOCAP is decidable by a reduction to ∀∃ R PAD + . Guillermo A. P´ erez 12/27

  35. ∀∃ R PAD & Undecidability ◮ ∀∃ R PAD := ∀ z 1 . . . ∀ z n ∃ x 1 . . . ∃ x m . ϕ ( x , z ) ◮ divisibilities of the form f ( z ) | g ( x , z ) ◮ ∀∃ R PAD + := ∀∃ R PAD with ¬ not allowed before divisibility. Claim (Bozga-Iosif’05, Lechner’15) SynthReach for SOCAP is decidable by a reduction to ∀∃ R PAD + . ∀∃ R PAD ≡ ∀∃ R PAD + Guillermo A. P´ erez 12/27

Download Presentation
Download Policy: The content available on the website is offered to you 'AS IS' for your personal information and use only. It cannot be commercialized, licensed, or distributed on other websites without prior consent from the author. To download a presentation, simply click this link. If you encounter any difficulties during the download process, it's possible that the publisher has removed the file from their server.

Recommend


More recommend