advances in possible orders of circulant hadamard
play

Advances in Possible Orders of Circulant Hadamard Matrices, and - PowerPoint PPT Presentation

Advances in Possible Orders of Circulant Hadamard Matrices, and Sequences with Large Merit Factor Jason Hu 1 Brooke Logan 2 1 Department of Mathematics University of California, Berkeley 2 Department of Mathematics Rowan University August 7,


  1. Advances in Possible Orders of Circulant Hadamard Matrices, and Sequences with Large Merit Factor Jason Hu 1 Brooke Logan 2 1 Department of Mathematics University of California, Berkeley 2 Department of Mathematics Rowan University August 7, 2014

  2. Outline Introduction Circulant Hadamard Matrices The Search for Circulant Hadamard Matrices Restrictions on n Ascending Wieferich Prime Pairs Descending Wieferich Prime Pairs Comparing results New Family of Binary Sequences Note: A generalization of Galois sequences New Family of Polyphase Sequences L 4 norms of polynomials Merit Factor Comparisons

  3. Introduction Circulant Hadamard Matrices The Search for Circulant Hadamard Matrices Restrictions on n Ascending Wieferich Prime Pairs Descending Wieferich Prime Pairs Comparing results New Family of Binary Sequences Note: A generalization of Galois sequences New Family of Polyphase Sequences L 4 norms of polynomials Merit Factor Comparisons

  4. Autocorrelations Definition (Aperiodic Autocorrelation) of a sequence of length n at shift k, 0 ≤ k < n is n − 1 − k � c k = a i ¯ a i + k i = 0 Definition (Periodic Autocorrelation) of a sequence of length n at shift k, 0 ≤ k < n is n − 1 � γ k = a i ¯ a i + kmod ( n ) i = 0

  5. Barker Sequences Definition A barker sequence is a binary sequence { a 0 , a 1 , ... a n − 1 } of length n such that when calculating the sequence’s aperiodic autocorrelation at shift k = 0, c 0 = n and for shift ranging from 1 ≤ k < n the aperiodic autocorrelation is | c k | ≤ 1

  6. Example ( { 1 , 1 , − 1 } ) 3 − 1 − 0 � c 0 = a i a i + 0 = 1 ( 1 ) + 1 ( 1 ) + ( − 1 )( − 1 ) = 3 i = 0 3 − 1 − 1 � c 1 = a i a i + 1 = 1 ( 1 ) + 1 ( − 1 ) = 0 i = 0 3 − 1 − 2 � c 2 = a i a i + 2 = 1 ( − 1 ) = − 1 i = 0 Barker Sequence!

  7. Barker Conjecture There exists no Barker sequence of n > 13 Proven for � n of odd length � even n = 3979201339721749133016171583224100 or n > 4 ∗ 10 33 (P. Borwein & M. Mossinghoff, 2014)

  8. Introduction Circulant Hadamard Matrices The Search for Circulant Hadamard Matrices Restrictions on n Ascending Wieferich Prime Pairs Descending Wieferich Prime Pairs Comparing results New Family of Binary Sequences Note: A generalization of Galois sequences New Family of Polyphase Sequences L 4 norms of polynomials Merit Factor Comparisons

  9. Circulant Hadamard Matrices Definition (Hadamard Matrix) An n × n matrix H of ± 1 where HH T = nI n Definition (Circulant Matrix) A matrix where each row after the first row is one cyclic shift to the right of the previous row. Example (Circulant Hadamard Matrix) + + + - − + + + + − + + + + − +

  10. Circulant Hadamard Matrices Definition (Hadamard Matrix) An n × n matrix H of ± 1 where HH T = nI n Definition (Circulant Matrix) A matrix where each row after the first row is one cyclic shift to the right of the previous row. Example (Circulant Hadamard Matrix)       + + + - + − + + 4 0 0 0 − + + + + + − + 0 4 0 0        ·  =       + − + + + + + − 0 0 4 0     + + − + - + + + 0 0 0 4

  11. Relationship between Barker sequences and Circulant Hadamard matrices Definition (Circulant Hadamard Conjecture) There exists no Circulant Hadamard Matrix with n > 4 Barker Sequence ⇒ small aperiodic autocorrelations ⇒ small periodic autocorrelations ⇒ Circulant Hadamard Matrix

  12. Introduction Circulant Hadamard Matrices The Search for Circulant Hadamard Matrices Restrictions on n Ascending Wieferich Prime Pairs Descending Wieferich Prime Pairs Comparing results New Family of Binary Sequences Note: A generalization of Galois sequences New Family of Polyphase Sequences L 4 norms of polynomials Merit Factor Comparisons

  13. Introduction Circulant Hadamard Matrices The Search for Circulant Hadamard Matrices Restrictions on n Ascending Wieferich Prime Pairs Descending Wieferich Prime Pairs Comparing results New Family of Binary Sequences Note: A generalization of Galois sequences New Family of Polyphase Sequences L 4 norms of polynomials Merit Factor Comparisons

  14. Restrictions � By Definition of a Hadamard Matrix � Must be a multiple of 4 (or n = 1 , 2) � Turyn, 1965 � assuming n>2 then � n = 4 m 2 � m is odd � m cannot be a prime power � more to come

  15. n = 4 m 2 When searching in a given bound M: m = p 1 p 2 ... p u ≤ M (1) Theorem (Turyn) 1 p ≤ ( 2 M 2 ) 3

  16. n = 4 M 2 When searching in a given bound M: p 1 → p 2 → ... → p 1

  17. Introduction Circulant Hadamard Matrices The Search for Circulant Hadamard Matrices Restrictions on n Ascending Wieferich Prime Pairs Descending Wieferich Prime Pairs Comparing results New Family of Binary Sequences Note: A generalization of Galois sequences New Family of Polyphase Sequences L 4 norms of polynomials Merit Factor Comparisons

  18. Links q → p � Ascending: q p − 1 ≡ 1 mod p 2 q < p and p → q � Descending: p q − 1 ≡ 1 mod q 2 q < p and p � q � Flimsy: q | ( p − 1 )

  19. Types of Ascending Pairs Different Cases q ⇆ p � Worst Case Scenario � Previous Search M = 10 13 1 � q < p ≤ min ( M q , ( 2 M 2 ) 3 ) M = 5 ∗ 10 14

  20. Double Wieferich Prime Pair q ⇆ p q < p ≤ min ( M 1 q , ( 2 M 2 ) 3 ) Theorem (W. Keller & J. Richstein ) Let p 1 be a primitive root of the prime q and define p 2 = p q 1 mod q 2 . mod q 2 : m = 0 , 1 , .., q − 2 } represents a complete set of Then { p m 2 incongruent solutions of p q − 1 ≡ 1 ( mod q 2 ) , each of which generates an infinite sequence of solutions in arithmetic progression with difference q 2 p q − 1 ≡ 1 mod q 2

  21. Example � q = 83 � p 1 = PrimitiveRoot ( q ) = 2 � Primitive root generator of the multiplicative group mod p 1 mod q 2 = 1081 � p 2 = p q mod q 2 , m = 0 , 1 , .., q − 3 � p m 2 2 � Even case: a = ( p 2 ) m + q 2 and b = q 2 − a � Odd Case: a = ( p 2 ) m and b = 2 q 2 − a � a , a + 2 q 2 , ... , � m = 37 ⇒ a = 4871

  22. Ascending and Flimsy q → p and p � q q < p ≤ min ( M 1 3 q , ( 2 M 2 ) 3 ) q | ( p − 1 ) q p − 1 ≡ 1 ( mod p 2 )

  23. Special Ascending Double Wieferich Prime Pairs Ascending and Flimsy 3 ⇆ 1006003 3 → 1006003 5 ⇆ 1645333507 5 → 20771 83 ⇆ 4871 5 → 53471161 911 ⇆ 318917 13 → 1747591 2903 ⇆ 18787 44963 → 5395561

  24. Strictly Ascending ( 3 → 11 → 71 → 3 ) → ... → ( q → p ) r → q → p → r

  25. Strictly Ascending M q < p ≤ 3 ∗ 11 ∗ 71 ∗ q r → q → p → r

  26. Strictly Ascending M q < p ≤ 3 ∗ 11 ∗ 71 ∗ q q < p ≤ M r ∗ q q < p ≤ M q 2

  27. Strictly Ascending 1 q 2 ) , ( 2 M 2 ) 3 ∗ 11 ∗ 71 ∗ q , M M r ∗ q , M q < p ≤ min ( max ( 3 )

  28. Introduction Circulant Hadamard Matrices The Search for Circulant Hadamard Matrices Restrictions on n Ascending Wieferich Prime Pairs Descending Wieferich Prime Pairs Comparing results New Family of Binary Sequences Note: A generalization of Galois sequences New Family of Polyphase Sequences L 4 norms of polynomials Merit Factor Comparisons

  29. 1: List A and List B = All primes in Ascending Pairs 2: while Length B > 0 do for p ∈ B do 3: for All Primes, q, such that 3 ≤ q < p do 4: if p q − 1 ≡ 1 mod q 2 then 5: Add ( p , q ) to solid link list and add q to T 6: else if q | ( p − 1 ) then 7: Add ( p , q ) to flimsy link list and add q to T 8: end if 9: end for 10: end for 11: B = T / A , A = A ∪ B , and Clear T 12: 13: end while

  30. Introduction Circulant Hadamard Matrices The Search for Circulant Hadamard Matrices Restrictions on n Ascending Wieferich Prime Pairs Descending Wieferich Prime Pairs Comparing results New Family of Binary Sequences Note: A generalization of Galois sequences New Family of Polyphase Sequences L 4 norms of polynomials Merit Factor Comparisons

  31. Comparing results 10 13 5 ∗ 10 14 M Bound Vertices 643931 15342 Ascending 59837 6616 Descending 1673025 33935 Flimsy 1729116 33264

  32. Creating Circuits � Johnson’s Circuit Finding Algorithm and Augmenter= 501630 � F Test � M = 10 13 cycles 2064 � M = 5 ∗ 10 14 cycles 6683 � Turyn Test � Leung Schmidt Test Theorem 1,5,10

  33. F-Test Leung and Schmidt 2005 � v p ( m ) = multiplicity of p in factorization of m � m q = q-free and squarefree part of m: m q = � p | m , p � = q p � b ( p , m ) = max q | m , q ≤ p { v p ( q p − 1 − 1 ) + v p ( ord m q ( q )) } � F ( m ) = gcd ( m 2 , � p | m p b ( p , m ) ) Theorem If n = 4 m 2 is the order of a circulant Hadamard matrix, then F ( m ) ≥ m φ ( m )

  34. Turyn Test Definition a is semi-primitive mod b: a j ≡ − 1 mod b for some j Definition r is self-conjugate mod s: For each p | r , p is semi-primitive mod the p-free part of s. Theorem If n = 4 m 2 is the order of a Circulant Hadamard Matrix, r | m , s | n , gcd ( r , s ) has k ≥ 1 distinct prime divisors, and r is self-conjugate mod s, then rs ≤ 2 k − 1 n

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