Matrix: Priebel/145bit

Description: Quadratic sieve; factoring a 145bit number. D. Priebel, Tenn. Tech Univ

Priebel/145bit graph
(bipartite graph drawing)


Priebel/145bit dmperm of Priebel/145bit
scc of Priebel/145bit

  • Home page of the UF Sparse Matrix Collection
  • Matrix group: Priebel
  • Click here for a description of the Priebel group.
  • Click here for a list of all matrices
  • Click here for a list of all matrix groups
  • download as a MATLAB mat-file, file size: 31 KB. Use UFget(2251) or UFget('Priebel/145bit') in MATLAB.
  • download in Matrix Market format, file size: 42 KB.
  • download in Rutherford/Boeing format, file size: 33 KB.

    Matrix properties
    number of rows1,002
    number of columns993
    nonzeros11,315
    structural full rank?no
    structural rank964
    # of blocks from dmperm119
    # strongly connected comp.28
    explicit zero entries0
    nonzero pattern symmetry 0%
    numeric value symmetry 0%
    typebinary
    structurerectangular
    Cholesky candidate?no
    positive definite?no

    authorD. Priebel
    editorT. Davis
    date2009
    kindcombinatorial problem
    2D/3D problem?no

    Additional fieldssize and type
    factor_basefull 993-by-1
    smooth_numberfull 1002-by-22
    solutionfull 385-by-1

    Notes:

    Each column in the matrix corresponds to a number in the factor base     
    less than some bound B.  Each row corresponds to a smooth number (able   
    to be completely factored over the factor base).  Each value in a row    
    binary vector corresponds to the exponent of the factor base mod 2.      
    For example:                                                             
                                                                             
        factor base: 2 7 23                                                  
        smooth numbers: 46, 28, 322                                          
        2^1       * 23^1 = 46                                                
        2^2 * 7^1        = 28                                                
        2^1 * 7^1 * 23^1 = 322                                               
        Matrix:                                                              
            101                                                              
            010                                                              
            111                                                              
                                                                             
    A solution to the matrix is considered to be a set of rows which when    
    combined in GF2 produce a null vector. Thus, if you multiply each of     
    the smooth numbers which correspond to that particular set of rows you   
    will get a number with only even exponents, making it a perfect          
    square. In the above example you can see that combining the 3 vectors    
    results in a null vector and, indeed, it is a perfect square: 644^2.     
                                                                             
    Problem.A: A GF(2) matrix constructed from the exponents of the          
    factorization of the smooth numbers over the factor base. A solution of  
    this matrix is a kernel (nullspace). Such a solution has a 1/2 chance of 
    being a factorization of N.                                              
                                                                             
    Problem.aux.factor_base: The factor base used. factor_base(j) corresponds
    to column j of the matrix. Note that a given column may or may not have  
    nonzero elements in the matrix.                                          
                                                                             
    Problem.aux.smooth_number: The smooth numbers, smooth over the factor    
    base.  smooth_number(i) corresponds to row i of the matrix.              
                                                                             
    Problem.aux.solution: A sample solution to the matrix. Combine, in GF(2) 
    the rows with these indicies to produce a solution to the matrix with the
    additional property that it factors N (a matrix solution only has 1/2    
    probability of factoring N).                                             
                                                                             
    Problem specific information:                                            
                                                                             
    n = 27393004579711727757848513391018843988362569 (145-bits)              
    passes primality test, n is composite, continuing...                     
    1) Initial bound: 20000, pi(20000) estimate: 2019,                       
        largest found: 17569 (actual bound)                                  
    2) Number of quadratic residues estimate: 1347, actual number found: 992 
    3) Modular square roots found: 1984(2x residues)                         
    4) Constructing smooth number list [sieving] (can take a while)...       
    Sieving for: 1002                                                        
    5. Constructing a matrix of size: 1002x993                               
    Set a total of 11315 exponents, with 503 negatives                       
    Matrix solution found with: 385 combinations                             
    Divisor: 4762476283061573160587 (probably prime)                         
    Divisor: 5751840629031342254587 (probably prime)                         
    

    Ordering statistics:result
    nnz(V) for QR, upper bound nnz(L) for LU, with COLAMD139,428
    nnz(R) for QR, upper bound nnz(U) for LU, with COLAMD150,622

    SVD-based statistics:
    norm(A)39.2964
    min(svd(A))2.03061e-32
    cond(A)1.9352e+33
    rank(A)964
    sprank(A)-rank(A)0
    null space dimension29
    full numerical rank?no
    singular value gap4.50578e+13

    singular values (MAT file):click here
    SVD method used:s = svd (full (A)) ;
    status:ok

    Priebel/145bit svd

    For a description of the statistics displayed above, click here.

    Maintained by Tim Davis, last updated 12-Mar-2014.
    Matrix pictures by cspy, a MATLAB function in the CSparse package.
    Matrix graphs by Yifan Hu, AT&T Labs Visualization Group.