the krw conjecture
play

The KRW conjecture Results and Open problems Or Meir Introduction - PowerPoint PPT Presentation

The KRW conjecture Results and Open problems Or Meir Introduction 1 Known results 2 Proof strategy 3 Future directions 4 Depth complexity Let f : { 0 , 1 } n { 0 , 1 } . The depth complexity D ( f ) is the depth of the shallowest


  1. The KRW conjecture Results and Open problems Or Meir

  2. Introduction 1 Known results 2 Proof strategy 3 Future directions 4

  3. Depth complexity Let f : { 0 , 1 } n → { 0 , 1 } . The depth complexity D ( f ) is the depth of the shallowest circuit for f . Captures the complexity of parallel computation.

  4. Depth complexity Let f : { 0 , 1 } n → { 0 , 1 } . The depth complexity D ( f ) is the depth of the shallowest circuit for f . Captures the complexity of parallel computation. We only consider circuits with fan-in 2 .

  5. Depth complexity Let f : { 0 , 1 } n → { 0 , 1 } . The depth complexity D ( f ) is the depth of the shallowest circuit for f . Captures the complexity of parallel computation. We only consider circuits with fan-in 2 . Major frontier: Explicit f with D ( f ) = ω (log n ) . a.k.a. P � = NC 1 .

  6. Composition [Karchmer-Raz-Wigderson-91]: We need to understand composition.

  7. Composition [Karchmer-Raz-Wigderson-91]: We need to understand composition. Let f : { 0 , 1 } m → { 0 , 1 } , g : { 0 , 1 } n → { 0 , 1 } . The composition f ⋄ g : { 0 , 1 } m × n → { 0 , 1 } is n g f . m a ( f ⋄ g )( X ) . X .

  8. Composition [Karchmer-Raz-Wigderson-91]: We need to understand composition. Let f : { 0 , 1 } m → { 0 , 1 } , g : { 0 , 1 } n → { 0 , 1 } . The composition f ⋄ g : { 0 , 1 } m × n → { 0 , 1 } is n g f . m a ( f ⋄ g )( X ) . X .

  9. Composition [Karchmer-Raz-Wigderson-91]: We need to understand composition. Let f : { 0 , 1 } m → { 0 , 1 } , g : { 0 , 1 } n → { 0 , 1 } . The composition f ⋄ g : { 0 , 1 } m × n → { 0 , 1 } is n g f . m a ( f ⋄ g )( X ) . X .

  10. Composition [Karchmer-Raz-Wigderson-91]: We need to understand composition. Let f : { 0 , 1 } m → { 0 , 1 } , g : { 0 , 1 } n → { 0 , 1 } . The composition f ⋄ g : { 0 , 1 } m × n → { 0 , 1 } is n g f . m a ( f ⋄ g )( X ) . X .

  11. The KRW conjecture g D ( f ) f f . a ( f ⋄ g )( X ) . X . . . . D ( g ) g g . . . . . . X 1 X m Clearly, D ( f ⋄ g ) ≤ D ( f ) + D ( g ) .

  12. The KRW conjecture g D ( f ) f f . a ( f ⋄ g )( X ) . X . . . . D ( g ) g g . . . . . . X 1 X m Clearly, D ( f ⋄ g ) ≤ D ( f ) + D ( g ) . KRW conjecture: ∀ f, g : D ( f ⋄ g ) ≈ D ( f ) + D ( g ) .

  13. The KRW conjecture g D ( f ) f f . a ( f ⋄ g )( X ) . X . . . . D ( g ) g g . . . . . . X 1 X m Clearly, D ( f ⋄ g ) ≤ D ( f ) + D ( g ) . KRW conjecture: ∀ f, g : D ( f ⋄ g ) ≈ D ( f ) + D ( g ) . Theorem [KRW91]: the conjecture implies that P � = NC 1 .

  14. Outline Introduction 1 Known results 2 Proof strategy 3 Future directions 4

  15. Karchmer-Wigderson relations Relate D ( f ) to complexity of a communication problem KW f .

  16. Karchmer-Wigderson relations Relate D ( f ) to complexity of a communication problem KW f . The KW relation KW f is defined as follows: Alice gets x ∈ f − 1 (0) . Bob gets y ∈ f − 1 (1) . Clearly, x � = y , so ∃ i s.t. x i � = y i . Want to find such i .

  17. Karchmer-Wigderson relations Relate D ( f ) to complexity of a communication problem KW f . The KW relation KW f is defined as follows: Alice gets x ∈ f − 1 (0) . Bob gets y ∈ f − 1 (1) . Clearly, x � = y , so ∃ i s.t. x i � = y i . Want to find such i . Theorem [KW88]: D ( f ) = C ( KW f ) .

  18. Karchmer-Wigderson relations Relate D ( f ) to complexity of a communication problem KW f . The KW relation KW f is defined as follows: Alice gets x ∈ f − 1 (0) . Bob gets y ∈ f − 1 (1) . Clearly, x � = y , so ∃ i s.t. x i � = y i . Want to find such i . Theorem [KW88]: D ( f ) = C ( KW f ) . Only deterministic protocols!

  19. Karchmer-Wigderson relations Relate D ( f ) to complexity of a communication problem KW f . The KW relation KW f is defined as follows: Alice gets x ∈ f − 1 (0) . Bob gets y ∈ f − 1 (1) . Clearly, x � = y , so ∃ i s.t. x i � = y i . Want to find such i . Theorem [KW88]: D ( f ) = C ( KW f ) . Only deterministic protocols! KRW conjecture: C ( KW f ⋄ g ) ≈ C ( KW f ) + C ( KW g )

  20. KRW and KW Can we use KW games to attack the KRW conjecture? What does KW f ⋄ g look like? Recall: f ⋄ g maps { 0 , 1 } m × n to { 0 , 1 } . Alice Bob g g f f . . a . 0 1 . X b Y . .

  21. KRW and KW Can we use KW games to attack the KRW conjecture? What does KW f ⋄ g look like? Recall: f ⋄ g maps { 0 , 1 } m × n to { 0 , 1 } . Alice Bob g g f f . . a . 0 1 . X b Y . .

  22. KRW and KW Can we use KW games to attack the KRW conjecture? What does KW f ⋄ g look like? Recall: f ⋄ g maps { 0 , 1 } m × n to { 0 , 1 } . Alice Bob g g f f . . a . 0 1 . X b Y . .

  23. KRW and KW Can we use KW games to attack the KRW conjecture? What does KW f ⋄ g look like? Recall: f ⋄ g maps { 0 , 1 } m × n to { 0 , 1 } . Alice Bob g g f f . . a . 0 1 . X b Y . .

  24. KRW and KW Can we use KW games to attack the KRW conjecture? What does KW f ⋄ g look like? Recall: f ⋄ g maps { 0 , 1 } m × n to { 0 , 1 } . Alice Bob g g f f . . a . 0 1 . X b Y . .

  25. KRW and KW Can we use KW games to attack the KRW conjecture? What does KW f ⋄ g look like? Recall: f ⋄ g maps { 0 , 1 } m × n to { 0 , 1 } . Alice Bob g g f f . . a . 0 1 . X b Y . .

  26. KRW and KW Can we use KW games to attack the KRW conjecture? What does KW f ⋄ g look like? Recall: f ⋄ g maps { 0 , 1 } m × n to { 0 , 1 } . Alice Bob g g f f . . a . 0 1 . X b Y . . Hence, C ( KW f ⋄ g ) ≤ C ( KW f ) + C ( KW g ) . KRW conjecture: the obvious protocol is essentially optimal.

  27. The universal relation The KRW conjecture is hard. [KRW91] suggested a starting point.

  28. The universal relation The KRW conjecture is hard. [KRW91] suggested a starting point. The universal relation U n is: Alice gets x ∈ { 0 , 1 } n . Bob gets y ∈ { 0 , 1 } n . x � = y . Wish to find i s.t. x i � = y i .

  29. The universal relation The KRW conjecture is hard. [KRW91] suggested a starting point. The universal relation U n is: Alice gets x ∈ { 0 , 1 } n . Bob gets y ∈ { 0 , 1 } n . x � = y . Wish to find i s.t. x i � = y i . Easy to prove: C (U n ) ≥ n .

  30. The universal relation The KRW conjecture is hard. [KRW91] suggested a starting point. The universal relation U n is: Alice gets x ∈ { 0 , 1 } n . Bob gets y ∈ { 0 , 1 } n . x � = y . Wish to find i s.t. x i � = y i . Easy to prove: C (U n ) ≥ n . [KRW91] suggested to study U m ⋄ U n .

  31. The composition of the universal relation [KRW91] suggested to study the composition U m ⋄ U n . Alice Bob a X b Y a � = b . If a i � = b i then X i � = Y i .

  32. The composition of the universal relation [KRW91] suggested to study the composition U m ⋄ U n . Alice Bob a X b Y a � = b . If a i � = b i then X i � = Y i .

  33. The composition of the universal relation [KRW91] suggested to study the composition U m ⋄ U n . Alice Bob a X b Y a � = b . If a i � = b i then X i � = Y i .

  34. The composition of the universal relation [KRW91] suggested to study the composition U m ⋄ U n . Alice Bob a X b Y a � = b . If a i � = b i then X i � = Y i .

  35. The composition of the universal relation Goal: C (U m ⋄ U n ) = C (U m ) + C (U n ) ≥ m + n . Alice Bob a X b Y

  36. The composition of the universal relation Goal: C (U m ⋄ U n ) = C (U m ) + C (U n ) ≥ m + n . Challenge was met by [Edmonds-Impagliazzo-Rudich-S’gall-91]. Alice Bob a X b Y

  37. The composition of the universal relation Goal: C (U m ⋄ U n ) = C (U m ) + C (U n ) ≥ m + n . Challenge was met by [Edmonds-Impagliazzo-Rudich-S’gall-91]. Alternative proof obtained by [H˚ astad-Wigderson-93]. Alice Bob a X b Y

  38. Composing a function and the universal relation An analog of KRW conjecture for KW f ⋄ U n for any f . [Gavinsky-M-Weinstein-Wigderson-14] Alice Bob f f a 0 1 X b Y If a i � = b i then X j � = Y j . The obvious protocol works.

  39. Composing a function and the universal relation An analog of KRW conjecture for KW f ⋄ U n for any f . [Gavinsky-M-Weinstein-Wigderson-14] Quantative improvement by [Koroth-M-18]. Alice Bob f f a 0 1 X b Y If a i � = b i then X j � = Y j . The obvious protocol works.

  40. Composing any function and parity [Dinur-M-16]: (Re-)proved KRW conjecture for f ⋄ � n

  41. Composing any function and parity [Dinur-M-16]: (Re-)proved KRW conjecture for f ⋄ � n Actually, this case was already implicit in [H˚ astad 98].

  42. Composing any function and parity [Dinur-M-16]: (Re-)proved KRW conjecture for f ⋄ � n Actually, this case was already implicit in [H˚ astad 98]. However, our proof was very different, and more in line with the other works on the KRW conjecture.

  43. Outline Introduction 1 Known results 2 Proof strategy 3 Future directions 4

  44. Why should the obvious protocol be optimal? Alice Bob g g f f . . a . 0 1 . X b Y . .

  45. Why should the obvious protocol be optimal? Alice Bob g g f f . . a . 0 1 . X b Y . . The players must solve KW g on some row.

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