Considere las matrices Toeplitz de 30 por 30, todas cuyas entradas son 0 o 1. Este desafío es un simple desafío de optimización para encontrar la matriz con el mayor determinante posible.

Entrada Ninguno

Salida de una matriz de Toeplitz de 30 por 30, todas cuyas entradas son 0 o 1 junto con su determinante.

Puntuación El determinante de la matriz que genera. Si dos personas obtienen el mismo puntaje, gana la primera respuesta.

Entradas principales hasta ahora

• 65,455,857,159,975 en Matlab por Nick Alger (aproximadamente (10 ^ 13.8)
• 65,455,857,159,975 en Python por isaacg (aproximadamente 10 ^ 13.8)
• 39,994,961,721,988 en Mathematica para 2012 Arcampion (aproximadamente 10 ^ 13.6)
• 39,788,537,400,052 en R por Flounderer (aproximadamente 10 ^ 13.6)
• 8.363.855.075.832 en Python por Vioz- (aproximadamente 10 ^ 12.9)
• 6,984,314,690,903 en Julia de Alex A. (aproximadamente 10 ^ 12.8)

Restricciones adicionales molestas 16 de julio de 2015

Si es posible, utilice una aritmética arbitraria o de alta precisión para calcular el determinante de salida final para que podamos estar seguros de lo que realmente es (siempre debe ser un número entero). En python esto debería ser útil.

Matlab, 65,455,857,159,975 (10 ^ 13.8159)

El método es ascenso en gradiente en el interior del cubo [0,1] ^ 59, con muchas conjeturas iniciales aleatorias, y redondeando al final para hacer que todo sea ceros y unos.

Matriz:

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

Código:

% 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);

Las matemáticas detrás de calcular el gradiente:

En el producto interno por elementos (es decir, producto interno de Hilbert-Schmidt), el gradiente del determinante tiene un representante de Riesz G dado por

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

El mapa, J, desde las variables de optimización (valores diagonales) hasta las matrices de toeplitz es lineal, por lo que el gradiente general g es la composición de estos dos mapas lineales,

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

donde vec () es el operador de vectorización que toma una matriz y la despliega en un vector.

Ascenso de gradiente interior:

Después de esto, todo lo que tiene que hacer es elegir un vector inicial de valores diagonales w_0, y para algunos pasos pequeños, alfa iterar:

1. w_proposed = w_k + alpha * g_k

2. para obtener w_ (k + 1), tome w_proposed y trunque valores fuera de [0,1] a 0 o 1

3. repite hasta que estés satisfecho, luego redondea todo a 0 o 1.

Mi resultado logró este determinante después de hacer aproximadamente 80,000 ensayos con conjeturas iniciales aleatorias uniformes.

Python 2 con Numpy, 65,455,857,159,975 ~ = 10 ^ 13.8

Esta es la escalada, tan simple como puede ser. Cálculo determinante final realizado utilizando SymPy para obtener un resultado exacto. Todas las matrices encontradas con este determinante son circulantes.

Matrices encontradas con este determinante, dado como valor de diagonal desde la parte inferior izquierda a la superior derecha:

01000100101101000011100111011101000100101101000011100111011
01011101110011100001011010010001011101110011100001011010010
01100001000111011101001110100101100001000111011101001110100
01110100111010010110000100011101110100111010010110000100011
01011101110001000011010010111001011101110001000011010010111
01000101100010110100111101110001000101100010110100111101110
01000100101101000011100111011101000100101101000011100111011

El primero, como matriz:

[[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]]

Código:

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

Aquí está mi intento de hacer un algoritmo genético pero solo con reproducción asexual. Espero haber entendido el desafío correctamente. Editar: aceleré un poco, probé una semilla aleatoria diferente y restringí a 100 generaciones.

    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

}

Salida:

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

Esto solo construye 1,000,000 matrices de Toeplitz aleatorias y verifica sus determinantes, registrando el máximo encontrado. Esperemos que alguien presente una solución analítica elegante, pero mientras tanto ...

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]))

Puede ver la salida aquí .

Mathematica, 39,994,961,721,988 (10 ^ 13.60)

Un método de optimización de recocido simulado simple; Sin optimización ni ajuste todavía.

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

Salida de muestra:

{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

)

Pitón 2, 8 363 855 075 832

Esto implica una estrategia muy básica, casi inexistente.

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

Aquí está la mejor matriz que encontró después de ~ 5,580,000 intentos:

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

Sigue corriendo...