Considera 30 per 30 matrici di Toeplitz tutte le cui voci sono 0 o 1. Questa sfida è una semplice sfida di ottimizzazione per trovare la matrice con il determinante più grande possibile.

Input Nessuno

Output A 30 per 30 matrice di Toeplitz tutte le cui voci sono 0 o 1 insieme al suo determinante.

Punteggio Il determinante della matrice prodotta. Se due persone ottengono lo stesso punteggio, vince la prima risposta.

Le voci principali finora

• 65.455.857.159.975 in Matlab di Nick Alger (circa (10 ^ 13,8)
• 65.455.857.159.975 in Python di isaacg (circa 10 ^ 13,8)
• 39.994.961.721.988 in Mathematica entro il 2012rcampion (circa 10 ^ 13.6)
• 39.788.537.400.052 pollici R di Flounderer (circa 10 ^ 13.6)
• 8.363.855.075.832 in Python di Vioz- (circa 10 ^ 12.9)
• 6.984.314.690.903 a Julia di Alex A. (circa 10 ^ 12,8)

Vincoli aggiuntivi fastidiosi 16 luglio 2015

Se è possibile, si prega di utilizzare l'aritmetica arbitraria o di alta precisione per calcolare il determinante di output finale in modo da poter essere sicuri di cosa sia realmente (dovrebbe essere sempre un numero intero). In Python questo dovrebbe essere utile.

Matlab, 65.455.857.159.975 (10 ^ 13.8159)

Il metodo è l' ascesa a gradiente all'interno del cubo [0,1] ^ 59, con molte ipotesi iniziali casuali, e l'arrotondamento alla fine per rendere tutto zero e uno.

Matrice:

0   1   1   1   0   0   0   0   1   0   1   1   0   1   0   0   1   0   0   0   1   0   1   1   1   0   1   1   1   0
0   0   1   1   1   0   0   0   0   1   0   1   1   0   1   0   0   1   0   0   0   1   0   1   1   1   0   1   1   1
1   0   0   1   1   1   0   0   0   0   1   0   1   1   0   1   0   0   1   0   0   0   1   0   1   1   1   0   1   1
1   1   0   0   1   1   1   0   0   0   0   1   0   1   1   0   1   0   0   1   0   0   0   1   0   1   1   1   0   1
1   1   1   0   0   1   1   1   0   0   0   0   1   0   1   1   0   1   0   0   1   0   0   0   1   0   1   1   1   0
0   1   1   1   0   0   1   1   1   0   0   0   0   1   0   1   1   0   1   0   0   1   0   0   0   1   0   1   1   1
1   0   1   1   1   0   0   1   1   1   0   0   0   0   1   0   1   1   0   1   0   0   1   0   0   0   1   0   1   1
1   1   0   1   1   1   0   0   1   1   1   0   0   0   0   1   0   1   1   0   1   0   0   1   0   0   0   1   0   1
1   1   1   0   1   1   1   0   0   1   1   1   0   0   0   0   1   0   1   1   0   1   0   0   1   0   0   0   1   0
0   1   1   1   0   1   1   1   0   0   1   1   1   0   0   0   0   1   0   1   1   0   1   0   0   1   0   0   0   1
1   0   1   1   1   0   1   1   1   0   0   1   1   1   0   0   0   0   1   0   1   1   0   1   0   0   1   0   0   0
0   1   0   1   1   1   0   1   1   1   0   0   1   1   1   0   0   0   0   1   0   1   1   0   1   0   0   1   0   0
0   0   1   0   1   1   1   0   1   1   1   0   0   1   1   1   0   0   0   0   1   0   1   1   0   1   0   0   1   0
0   0   0   1   0   1   1   1   0   1   1   1   0   0   1   1   1   0   0   0   0   1   0   1   1   0   1   0   0   1
1   0   0   0   1   0   1   1   1   0   1   1   1   0   0   1   1   1   0   0   0   0   1   0   1   1   0   1   0   0
0   1   0   0   0   1   0   1   1   1   0   1   1   1   0   0   1   1   1   0   0   0   0   1   0   1   1   0   1   0
0   0   1   0   0   0   1   0   1   1   1   0   1   1   1   0   0   1   1   1   0   0   0   0   1   0   1   1   0   1
1   0   0   1   0   0   0   1   0   1   1   1   0   1   1   1   0   0   1   1   1   0   0   0   0   1   0   1   1   0
0   1   0   0   1   0   0   0   1   0   1   1   1   0   1   1   1   0   0   1   1   1   0   0   0   0   1   0   1   1
1   0   1   0   0   1   0   0   0   1   0   1   1   1   0   1   1   1   0   0   1   1   1   0   0   0   0   1   0   1
1   1   0   1   0   0   1   0   0   0   1   0   1   1   1   0   1   1   1   0   0   1   1   1   0   0   0   0   1   0
0   1   1   0   1   0   0   1   0   0   0   1   0   1   1   1   0   1   1   1   0   0   1   1   1   0   0   0   0   1
1   0   1   1   0   1   0   0   1   0   0   0   1   0   1   1   1   0   1   1   1   0   0   1   1   1   0   0   0   0
0   1   0   1   1   0   1   0   0   1   0   0   0   1   0   1   1   1   0   1   1   1   0   0   1   1   1   0   0   0
0   0   1   0   1   1   0   1   0   0   1   0   0   0   1   0   1   1   1   0   1   1   1   0   0   1   1   1   0   0
0   0   0   1   0   1   1   0   1   0   0   1   0   0   0   1   0   1   1   1   0   1   1   1   0   0   1   1   1   0
0   0   0   0   1   0   1   1   0   1   0   0   1   0   0   0   1   0   1   1   1   0   1   1   1   0   0   1   1   1
1   0   0   0   0   1   0   1   1   0   1   0   0   1   0   0   0   1   0   1   1   1   0   1   1   1   0   0   1   1
1   1   0   0   0   0   1   0   1   1   0   1   0   0   1   0   0   0   1   0   1   1   1   0   1   1   1   0   0   1
1   1   1   0   0   0   0   1   0   1   1   0   1   0   0   1   0   0   0   1   0   1   1   1   0   1   1   1   0   0

Codice:

% Toeplitz 0-1 determinant optimization

n = 30;
m = n + n-1;

toeplitz_map = @(w) toeplitz(w(n:-1:1), w(n:end));

objective = @(w) det(toeplitz_map(w));

detgrad = @(A) det(A)*inv(A)';

toeplitz_map_matrix = zeros(n^2,m);
for k=1:m
    ek = zeros(m,1);
    ek(k) = 1;
    M = toeplitz_map(ek);
    toeplitz_map_matrix(:,k) = M(:);
end

gradient = @(w) (reshape(detgrad(toeplitz_map(w)),1,n^2)*...
                 toeplitz_map_matrix)';

%check gradient with finite differences
w = randn(m,1);
dw = randn(m,1);
s = 1e-6;
g_diff = (objective(w+s*dw) - objective(w))/s;
g = gradient(w)'*dw;
grad_err = (g - g_diff)/g_diff

warning('off')
disp('multiple gradient ascent:')
w_best = zeros(m,1);
f_best = 0;
for trial=1:100000
    w0 = rand(m,1);
    w = w0;
    alpha0 = 1e-5; %step size
    for k=1:20
        f = objective(w);
        g = gradient(w);
        alpha = alpha0;
        for hh=1:100
            w2 = w + alpha*g;
            f2 = objective(w2);
            if f2 > f
                w = w2;
                break;
            else
                alpha = alpha/2;
            end
        end

        buffer = 1e-4;
        for jj=1:m
            if (w(jj) > 1)
                w(jj) = 1 - buffer;
            elseif (w(jj) < 0)
                w(jj) = 0 + buffer;
            end
        end
    end

    w = round(w);
    f = objective(w);
    if f > f_best
        w_best = w;
        f_best = f;
    end
    disp(trial)
    disp(f_best)
    disp(f)
end

M = toeplitz_map(w_best);

La matematica dietro il calcolo del gradiente:

Nel prodotto interno elementwise (cioè, prodotto interno di Hilbert-Schmidt), il gradiente del determinante ha il rappresentante G di Riesz dato da

G = det (A) A ^ (- *).

La mappa, J, dalle variabili di ottimizzazione (valori diagonali) alle matrici di toeplitz è lineare, quindi il gradiente generale g è la composizione di queste due mappe lineari,

g = (vec (G) * J) ',

dove vec () è l' operatore di vettorializzazione che prende una matrice e la sviluppa in un vettore.

Salita interna al gradiente:

Dopodiché tutto ciò che devi fare è scegliere un vettore iniziale di valori diagonali w_0 e per alcuni piccoli passi alfa iterare:

1. w_proposed = w_k + alpha * g_k

2. per ottenere w_ (k + 1), prendere w_proposed e troncare i valori al di fuori di [0,1] a 0 o 1

3. ripetere fino a quando non è soddisfatto, quindi arrotondare tutto a 0 o 1.

Il mio risultato ha raggiunto questo determinante dopo aver fatto circa 80.000 prove con ipotesi iniziali casuali uniformi.

Python 2 con Numpy, 65.455.857.159.975 ~ = 10 ^ 13,8

Questa è l'arrampicata in collina, il più semplice possibile. Calcolo del determinante finale eseguito utilizzando SymPy per ottenere un risultato esatto. Tutte le matrici trovate con questo determinante sono circolanti.

Matrici trovate con questo determinante, dato come valore della diagonale dal basso a sinistra in alto a destra:

01000100101101000011100111011101000100101101000011100111011
01011101110011100001011010010001011101110011100001011010010
01100001000111011101001110100101100001000111011101001110100
01110100111010010110000100011101110100111010010110000100011
01011101110001000011010010111001011101110001000011010010111
01000101100010110100111101110001000101100010110100111101110
01000100101101000011100111011101000100101101000011100111011

Il primo, come matrice:

[[1 0 1 0 0 0 1 0 0 1 0 1 1 0 1 0 0 0 0 1 1 1 0 0 1 1 1 0 1 1]
 [1 1 0 1 0 0 0 1 0 0 1 0 1 1 0 1 0 0 0 0 1 1 1 0 0 1 1 1 0 1]
 [1 1 1 0 1 0 0 0 1 0 0 1 0 1 1 0 1 0 0 0 0 1 1 1 0 0 1 1 1 0]
 [0 1 1 1 0 1 0 0 0 1 0 0 1 0 1 1 0 1 0 0 0 0 1 1 1 0 0 1 1 1]
 [1 0 1 1 1 0 1 0 0 0 1 0 0 1 0 1 1 0 1 0 0 0 0 1 1 1 0 0 1 1]
 [1 1 0 1 1 1 0 1 0 0 0 1 0 0 1 0 1 1 0 1 0 0 0 0 1 1 1 0 0 1]
 [1 1 1 0 1 1 1 0 1 0 0 0 1 0 0 1 0 1 1 0 1 0 0 0 0 1 1 1 0 0]
 [0 1 1 1 0 1 1 1 0 1 0 0 0 1 0 0 1 0 1 1 0 1 0 0 0 0 1 1 1 0]
 [0 0 1 1 1 0 1 1 1 0 1 0 0 0 1 0 0 1 0 1 1 0 1 0 0 0 0 1 1 1]
 [1 0 0 1 1 1 0 1 1 1 0 1 0 0 0 1 0 0 1 0 1 1 0 1 0 0 0 0 1 1]
 [1 1 0 0 1 1 1 0 1 1 1 0 1 0 0 0 1 0 0 1 0 1 1 0 1 0 0 0 0 1]
 [1 1 1 0 0 1 1 1 0 1 1 1 0 1 0 0 0 1 0 0 1 0 1 1 0 1 0 0 0 0]
 [0 1 1 1 0 0 1 1 1 0 1 1 1 0 1 0 0 0 1 0 0 1 0 1 1 0 1 0 0 0]
 [0 0 1 1 1 0 0 1 1 1 0 1 1 1 0 1 0 0 0 1 0 0 1 0 1 1 0 1 0 0]
 [0 0 0 1 1 1 0 0 1 1 1 0 1 1 1 0 1 0 0 0 1 0 0 1 0 1 1 0 1 0]
 [0 0 0 0 1 1 1 0 0 1 1 1 0 1 1 1 0 1 0 0 0 1 0 0 1 0 1 1 0 1]
 [1 0 0 0 0 1 1 1 0 0 1 1 1 0 1 1 1 0 1 0 0 0 1 0 0 1 0 1 1 0]
 [0 1 0 0 0 0 1 1 1 0 0 1 1 1 0 1 1 1 0 1 0 0 0 1 0 0 1 0 1 1]
 [1 0 1 0 0 0 0 1 1 1 0 0 1 1 1 0 1 1 1 0 1 0 0 0 1 0 0 1 0 1]
 [1 1 0 1 0 0 0 0 1 1 1 0 0 1 1 1 0 1 1 1 0 1 0 0 0 1 0 0 1 0]
 [0 1 1 0 1 0 0 0 0 1 1 1 0 0 1 1 1 0 1 1 1 0 1 0 0 0 1 0 0 1]
 [1 0 1 1 0 1 0 0 0 0 1 1 1 0 0 1 1 1 0 1 1 1 0 1 0 0 0 1 0 0]
 [0 1 0 1 1 0 1 0 0 0 0 1 1 1 0 0 1 1 1 0 1 1 1 0 1 0 0 0 1 0]
 [0 0 1 0 1 1 0 1 0 0 0 0 1 1 1 0 0 1 1 1 0 1 1 1 0 1 0 0 0 1]
 [1 0 0 1 0 1 1 0 1 0 0 0 0 1 1 1 0 0 1 1 1 0 1 1 1 0 1 0 0 0]
 [0 1 0 0 1 0 1 1 0 1 0 0 0 0 1 1 1 0 0 1 1 1 0 1 1 1 0 1 0 0]
 [0 0 1 0 0 1 0 1 1 0 1 0 0 0 0 1 1 1 0 0 1 1 1 0 1 1 1 0 1 0]
 [0 0 0 1 0 0 1 0 1 1 0 1 0 0 0 0 1 1 1 0 0 1 1 1 0 1 1 1 0 1]
 [1 0 0 0 1 0 0 1 0 1 1 0 1 0 0 0 0 1 1 1 0 0 1 1 1 0 1 1 1 0]
 [0 1 0 0 0 1 0 0 1 0 1 1 0 1 0 0 0 0 1 1 1 0 0 1 1 1 0 1 1 1]]

Codice:

import numpy as np
import sympy as sp
import random
import time
SIZE = 30

random.seed(0)

def gen_diag():
    return [random.randint(0, 1) for i in range(SIZE*2 - 1)]

def diag_to_mat(diag):
    return [diag[a:a+SIZE] for a in range(SIZE-1, -1, -1)]

def diag_to_det(diag):
    matrix = diag_to_mat(diag)
    return np.linalg.det(matrix)

def improve(diag):
    old_diag = diag
    really_old_diag = []
    while really_old_diag != old_diag:
        really_old_diag = old_diag
        for flip_at in range(SIZE * 2 - 1):
            new_diag = old_diag[:]
            new_diag[flip_at] ^= 1
            old_diag = max(old_diag, new_diag, key=diag_to_det)
    return old_diag

overall_best_score = 0
time.clock()
while time.clock() < 500:
    best = improve(gen_diag())
    best_score = diag_to_det(best)
    if best_score > overall_best_score:
        overall_best_score = best_score
        overall_best = best
        print(time.clock(), sp.Matrix(diag_to_mat(overall_best)).det(), ''.join(map(str,overall_best)))


mat = diag_to_mat(overall_best)

sym_mat = sp.Matrix(mat)

print(overall_best)
print(sym_mat.det())

R, 39 788 537 400 052

Ecco il mio tentativo di fare un algoritmo genetico ma solo con riproduzione asessuata. Spero di aver capito bene la sfida. Modifica: accelerato un po ', provato un seme casuale diverso e limitato a 100 generazioni.

    options(scipen=999)

toeplitz <- function(x){
# make toeplitz matrix with first row
# x[1:a] and first col x[(a+1):n]
# where n is the length of x and a= n/2
# Requires x to have even length
#
# [1,1] entry is x[a+1]

N <- length(x)/2
out <- matrix(0, N, N)
out[1,] <- x[1:N]
out[,1] <- x[(N+1):length(x)]
for (i in 2:N){
  for (j in 2:N){
    out[i,j] <- out[i-1, j-1]
  }
} 

out
}

set.seed(1002)

generations <- 100
popsize <- 25
cols <- 60
population <- matrix(sample(0:1, cols*popsize, replace=T), nc=cols)
numfresh <- 5 # number of totally random choices added to population

for (i in 1:generations){

fitness <- apply(population, 1, function(x) det(toeplitz(x)) )
mother <- which(fitness==max(fitness))[1]

population <- matrix(rep(population[mother,], popsize), nc=cols, byrow=T)
for (i in 2:(popsize-numfresh)){
  x <- sample(cols, 1)
  population[i,x] <- 1-population[i,x]
}
for (i in (popsize-numfresh +1):popsize){
  population[i,] <- sample(0:1, cols, replace=T)
}


print(population[1,])
print(fitness[mother])
print(det(toeplitz(population[1,]))) # to check correct

}

Produzione:

print(population[1, 1:(cols/2)]) # first row
print(population[1, (cols/2+1):(cols)]) # first column (overwrites 1st row)

to <- toeplitz(population[1,])

for (i in 1:(cols/2)) cat(to[i,], "\n")

1 1 1 0 1 0 0 0 0 1 0 0 1 0 0 0 0 1 1 1 0 1 1 1 1 0 0 0 1 0 
0 1 1 1 0 1 0 0 0 0 1 0 0 1 0 0 0 0 1 1 1 0 1 1 1 1 0 0 0 1 
1 0 1 1 1 0 1 0 0 0 0 1 0 0 1 0 0 0 0 1 1 1 0 1 1 1 1 0 0 0 
0 1 0 1 1 1 0 1 0 0 0 0 1 0 0 1 0 0 0 0 1 1 1 0 1 1 1 1 0 0 
0 0 1 0 1 1 1 0 1 0 0 0 0 1 0 0 1 0 0 0 0 1 1 1 0 1 1 1 1 0 
0 0 0 1 0 1 1 1 0 1 0 0 0 0 1 0 0 1 0 0 0 0 1 1 1 0 1 1 1 1 
1 0 0 0 1 0 1 1 1 0 1 0 0 0 0 1 0 0 1 0 0 0 0 1 1 1 0 1 1 1 
1 1 0 0 0 1 0 1 1 1 0 1 0 0 0 0 1 0 0 1 0 0 0 0 1 1 1 0 1 1 
1 1 1 0 0 0 1 0 1 1 1 0 1 0 0 0 0 1 0 0 1 0 0 0 0 1 1 1 0 1 
1 1 1 1 0 0 0 1 0 1 1 1 0 1 0 0 0 0 1 0 0 1 0 0 0 0 1 1 1 0 
0 1 1 1 1 0 0 0 1 0 1 1 1 0 1 0 0 0 0 1 0 0 1 0 0 0 0 1 1 1 
1 0 1 1 1 1 0 0 0 1 0 1 1 1 0 1 0 0 0 0 1 0 0 1 0 0 0 0 1 1 
1 1 0 1 1 1 1 0 0 0 1 0 1 1 1 0 1 0 0 0 0 1 0 0 1 0 0 0 0 1 
1 1 1 0 1 1 1 1 0 0 0 1 0 1 1 1 0 1 0 0 0 0 1 0 0 1 0 0 0 0 
0 1 1 1 0 1 1 1 1 0 0 0 1 0 1 1 1 0 1 0 0 0 0 1 0 0 1 0 0 0 
0 0 1 1 1 0 1 1 1 1 0 0 0 1 0 1 1 1 0 1 0 0 0 0 1 0 0 1 0 0 
0 0 0 1 1 1 0 1 1 1 1 0 0 0 1 0 1 1 1 0 1 0 0 0 0 1 0 0 1 0 
0 0 0 0 1 1 1 0 1 1 1 1 0 0 0 1 0 1 1 1 0 1 0 0 0 0 1 0 0 1 
1 0 0 0 0 1 1 1 0 1 1 1 1 0 0 0 1 0 1 1 1 0 1 0 0 0 0 1 0 0 
0 1 0 0 0 0 1 1 1 0 1 1 1 1 0 0 0 1 0 1 1 1 0 1 0 0 0 0 1 0 
0 0 1 0 0 0 0 1 1 1 0 1 1 1 1 0 0 0 1 0 1 1 1 0 1 0 0 0 0 1 
1 0 0 1 0 0 0 0 1 1 1 0 1 1 1 1 0 0 0 1 0 1 1 1 0 1 0 0 0 0 
0 1 0 0 1 0 0 0 0 1 1 1 0 1 1 1 1 0 0 0 1 0 1 1 1 0 1 0 0 0 
0 0 1 0 0 1 0 0 0 0 1 1 1 0 1 1 1 1 0 0 0 1 0 1 1 1 0 1 0 0 
0 0 0 1 0 0 1 0 0 0 0 1 1 1 0 1 1 1 1 0 0 0 1 0 1 1 1 0 1 0 
0 0 0 0 1 0 0 1 0 0 0 0 1 1 1 0 1 1 1 1 0 0 0 1 0 1 1 1 0 1 
1 0 0 0 0 1 0 0 1 0 0 0 0 1 1 1 0 1 1 1 1 0 0 0 1 0 1 1 1 0 
0 1 0 0 0 0 1 0 0 1 0 0 0 0 1 1 1 0 1 1 1 1 0 0 0 1 0 1 1 1 
1 0 1 0 0 0 0 1 0 0 1 0 0 0 0 1 1 1 0 1 1 1 1 0 0 0 1 0 1 1 
1 1 0 1 0 0 0 0 1 0 0 1 0 0 0 0 1 1 1 0 1 1 1 1 0 0 0 1 0 1 

Julia, 6.984.314.690.902.998

Questo costruisce appena 1.000.000 di matrici Toeplitz casuali e controlla i loro determinanti, registrando il massimo riscontrato. Eventualmente qualcuno troverà un'elegante soluzione analitica, ma nel frattempo ...

function toeplitz(a, b)
    n = length(a)
    T = Array(Int, n, n)
    T[1,:] = b
    T[:,1] = a
    for i = 2:n
        T[i,2:n] = T[i-1,1:n-1]
    end
    T
end

d = 0
A = Any[]

for i = 1:1000000
    # Construct two random 0,1 arrays
    r1 = rand(0:1, 30)
    r2 = rand(0:1, 30)

    # Compute the determinant of a toeplitz matrix constructed
    # from the two random arrays
    D = det(toeplitz(r1, r2))

    # If the computed determinant is larger than anything we've
    # encountered so far, add it to A so we can access it later
    D > d && begin
        push!(A, (D, r1, r2))
        d = D
    end
end

M,N = findmax([i[1] for i in A])

println("Maximum determinant: ", M, "\n")
println(toeplitz(A[N][2], A[N][3]))

È possibile visualizzare l'output qui .

Mathematica, 39.994.961.721.988 (10 ^ 13,60)

Un semplice metodo di ottimizzazione della ricottura simulato; nessuna ottimizzazione o ottimizzazione ancora.

n = 30;
current = -\[Infinity];
best = -\[Infinity];
saved = ConstantArray[0, {2 n - 1}];
m := Array[a[[n + #1 - #2]] &, {n, n}];
improved = True;
iters = 1000;
pmax = 0.1;
AbsoluteTiming[
 While[improved || RandomReal[] < pmax,
   improved = False;
   a = saved;
   Do[
    Do[
      a[[i]] = 1 - a[[i]];
      With[{d = Det[m]},
       If[d > best,
          best = d;
          current = d;
          saved = a;
          improved = True;
          Break[];,
          If[d > current || RandomReal[] < pmax (1 - p/iters),
           current = d;
           Break[];,
           a[[i]] = 1 - a[[i]];
           ]
          ];
        ;
       ],
      {i, 2 n - 1}];,
    {p, iters}];
   ];
 ]
best
Log10[best // N]
a = saved;
m // MatrixForm

Uscita campione:

{20.714876,Null}
39994961721988
13.602
(1  1   1   0   1   0   0   0   0   1   1   0   1   1   1   0   1   1   0   1   1   0   0   0   0   1   0   0   0   0
0   1   1   1   0   1   0   0   0   0   1   1   0   1   1   1   0   1   1   0   1   1   0   0   0   0   1   0   0   0
0   0   1   1   1   0   1   0   0   0   0   1   1   0   1   1   1   0   1   1   0   1   1   0   0   0   0   1   0   0
0   0   0   1   1   1   0   1   0   0   0   0   1   1   0   1   1   1   0   1   1   0   1   1   0   0   0   0   1   0
0   0   0   0   1   1   1   0   1   0   0   0   0   1   1   0   1   1   1   0   1   1   0   1   1   0   0   0   0   1
1   0   0   0   0   1   1   1   0   1   0   0   0   0   1   1   0   1   1   1   0   1   1   0   1   1   0   0   0   0
0   1   0   0   0   0   1   1   1   0   1   0   0   0   0   1   1   0   1   1   1   0   1   1   0   1   1   0   0   0
0   0   1   0   0   0   0   1   1   1   0   1   0   0   0   0   1   1   0   1   1   1   0   1   1   0   1   1   0   0
0   0   0   1   0   0   0   0   1   1   1   0   1   0   0   0   0   1   1   0   1   1   1   0   1   1   0   1   1   0
0   0   0   0   1   0   0   0   0   1   1   1   0   1   0   0   0   0   1   1   0   1   1   1   0   1   1   0   1   1
1   0   0   0   0   1   0   0   0   0   1   1   1   0   1   0   0   0   0   1   1   0   1   1   1   0   1   1   0   1
1   1   0   0   0   0   1   0   0   0   0   1   1   1   0   1   0   0   0   0   1   1   0   1   1   1   0   1   1   0
0   1   1   0   0   0   0   1   0   0   0   0   1   1   1   0   1   0   0   0   0   1   1   0   1   1   1   0   1   1
1   0   1   1   0   0   0   0   1   0   0   0   0   1   1   1   0   1   0   0   0   0   1   1   0   1   1   1   0   1
1   1   0   1   1   0   0   0   0   1   0   0   0   0   1   1   1   0   1   0   0   0   0   1   1   0   1   1   1   0
0   1   1   0   1   1   0   0   0   0   1   0   0   0   0   1   1   1   0   1   0   0   0   0   1   1   0   1   1   1
1   0   1   1   0   1   1   0   0   0   0   1   0   0   0   0   1   1   1   0   1   0   0   0   0   1   1   0   1   1
1   1   0   1   1   0   1   1   0   0   0   0   1   0   0   0   0   1   1   1   0   1   0   0   0   0   1   1   0   1
1   1   1   0   1   1   0   1   1   0   0   0   0   1   0   0   0   0   1   1   1   0   1   0   0   0   0   1   1   0
0   1   1   1   0   1   1   0   1   1   0   0   0   0   1   0   0   0   0   1   1   1   0   1   0   0   0   0   1   1
1   0   1   1   1   0   1   1   0   1   1   0   0   0   0   1   0   0   0   0   1   1   1   0   1   0   0   0   0   1
1   1   0   1   1   1   0   1   1   0   1   1   0   0   0   0   1   0   0   0   0   1   1   1   0   1   0   0   0   0
0   1   1   0   1   1   1   0   1   1   0   1   1   0   0   0   0   1   0   0   0   0   1   1   1   0   1   0   0   0
0   0   1   1   0   1   1   1   0   1   1   0   1   1   0   0   0   0   1   0   0   0   0   1   1   1   0   1   0   0
0   0   0   1   1   0   1   1   1   0   1   1   0   1   1   0   0   0   0   1   0   0   0   0   1   1   1   0   1   0
0   0   0   0   1   1   0   1   1   1   0   1   1   0   1   1   0   0   0   0   1   0   0   0   0   1   1   1   0   1
1   0   0   0   0   1   1   0   1   1   1   0   1   1   0   1   1   0   0   0   0   1   0   0   0   0   1   1   1   0
0   1   0   0   0   0   1   1   0   1   1   1   0   1   1   0   1   1   0   0   0   0   1   0   0   0   0   1   1   1
1   0   1   0   0   0   0   1   1   0   1   1   1   0   1   1   0   1   1   0   0   0   0   1   0   0   0   0   1   1
1   1   0   1   0   0   0   0   1   1   0   1   1   1   0   1   1   0   1   1   0   0   0   0   1   0   0   0   0   1

)

Python 2, 8363 855 075 832

Questo ha una strategia molto semplice, quasi inesistente.

from scipy import linalg

start = 2**28
mdet  = 0
mmat  = []
count = 0
powr  = 1
while 1:
 count += 1
 v = map(int,bin(start)[2:].zfill(59))
 m = [v[29:]]
 for i in xrange(1,30):
     m += [v[29-i:59-i]]
 d = 0
 try: d = linalg.det(m, check_finite=False)
 except: print start
 if d > mdet:
     print d
     print m
     mdet = d
     mmat = m
     start += 1
     powr = 1
 else:
     start += 2**powr
     powr += 1
     if start>(2**59-1):
         start-=2**59-1
         powr = 1
 if count%10000==0: print 'Tried',count

Ecco la matrice migliore che ha trovato dopo ~ 5.580.000 tentativi:

1 1 0 0 1 1 0 1 1 1 0 1 1 1 0 0 0 0 0 0 1 0 0 1 0 1 1 0 1 0
1 1 1 0 0 1 1 0 1 1 1 0 1 1 1 0 0 0 0 0 0 1 0 0 1 0 1 1 0 1
1 1 1 1 0 0 1 1 0 1 1 1 0 1 1 1 0 0 0 0 0 0 1 0 0 1 0 1 1 0
0 1 1 1 1 0 0 1 1 0 1 1 1 0 1 1 1 0 0 0 0 0 0 1 0 0 1 0 1 1
1 0 1 1 1 1 0 0 1 1 0 1 1 1 0 1 1 1 0 0 0 0 0 0 1 0 0 1 0 1
0 1 0 1 1 1 1 0 0 1 1 0 1 1 1 0 1 1 1 0 0 0 0 0 0 1 0 0 1 0
1 0 1 0 1 1 1 1 0 0 1 1 0 1 1 1 0 1 1 1 0 0 0 0 0 0 1 0 0 1
0 1 0 1 0 1 1 1 1 0 0 1 1 0 1 1 1 0 1 1 1 0 0 0 0 0 0 1 0 0
0 0 1 0 1 0 1 1 1 1 0 0 1 1 0 1 1 1 0 1 1 1 0 0 0 0 0 0 1 0
1 0 0 1 0 1 0 1 1 1 1 0 0 1 1 0 1 1 1 0 1 1 1 0 0 0 0 0 0 1
1 1 0 0 1 0 1 0 1 1 1 1 0 0 1 1 0 1 1 1 0 1 1 1 0 0 0 0 0 0
0 1 1 0 0 1 0 1 0 1 1 1 1 0 0 1 1 0 1 1 1 0 1 1 1 0 0 0 0 0
0 0 1 1 0 0 1 0 1 0 1 1 1 1 0 0 1 1 0 1 1 1 0 1 1 1 0 0 0 0
0 0 0 1 1 0 0 1 0 1 0 1 1 1 1 0 0 1 1 0 1 1 1 0 1 1 1 0 0 0
0 0 0 0 1 1 0 0 1 0 1 0 1 1 1 1 0 0 1 1 0 1 1 1 0 1 1 1 0 0
0 0 0 0 0 1 1 0 0 1 0 1 0 1 1 1 1 0 0 1 1 0 1 1 1 0 1 1 1 0
1 0 0 0 0 0 1 1 0 0 1 0 1 0 1 1 1 1 0 0 1 1 0 1 1 1 0 1 1 1
0 1 0 0 0 0 0 1 1 0 0 1 0 1 0 1 1 1 1 0 0 1 1 0 1 1 1 0 1 1
1 0 1 0 0 0 0 0 1 1 0 0 1 0 1 0 1 1 1 1 0 0 1 1 0 1 1 1 0 1
1 1 0 1 0 0 0 0 0 1 1 0 0 1 0 1 0 1 1 1 1 0 0 1 1 0 1 1 1 0
1 1 1 0 1 0 0 0 0 0 1 1 0 0 1 0 1 0 1 1 1 1 0 0 1 1 0 1 1 1
1 1 1 1 0 1 0 0 0 0 0 1 1 0 0 1 0 1 0 1 1 1 1 0 0 1 1 0 1 1
1 1 1 1 1 0 1 0 0 0 0 0 1 1 0 0 1 0 1 0 1 1 1 1 0 0 1 1 0 1
1 1 1 1 1 1 0 1 0 0 0 0 0 1 1 0 0 1 0 1 0 1 1 1 1 0 0 1 1 0
1 1 1 1 1 1 1 0 1 0 0 0 0 0 1 1 0 0 1 0 1 0 1 1 1 1 0 0 1 1
1 1 1 1 1 1 1 1 0 1 0 0 0 0 0 1 1 0 0 1 0 1 0 1 1 1 1 0 0 1
0 1 1 1 1 1 1 1 1 0 1 0 0 0 0 0 1 1 0 0 1 0 1 0 1 1 1 1 0 0
0 0 1 1 1 1 1 1 1 1 0 1 0 0 0 0 0 1 1 0 0 1 0 1 0 1 1 1 1 0
1 0 0 1 1 1 1 1 1 1 1 0 1 0 0 0 0 0 1 1 0 0 1 0 1 0 1 1 1 1
0 1 0 0 1 1 1 1 1 1 1 1 0 1 0 0 0 0 0 1 1 0 0 1 0 1 0 1 1 1

Ancora correndo...