如何分配一个当前在分配过程中抛出 std::bad_alloc 错误的大量布尔数组

How can I allocate a massive boolean array that currently throws std::bad_alloc errors during allocation?

本文关键字:分配 alloc bad 错误 数组 布尔 std 过程中 何分配 一个      更新时间:2023-10-16

我已经看到了几个与此相关的问题,但我想验证我是否遇到了类似的问题。我的代码分配了一个包含大量元素的布尔数组。这是我的代码,在x86_64 Linux 机器上编译:

#include <iostream>
#include <math.h>
using std::cout;
using std::endl;
using std::nothrow;
long problem3()
{
    long upper_bound = 600851475143;
    long max_prime_factor = 1;
    long max_possible_prime = (long) sqrt(upper_bound) + 1;
    bool * primes;
    primes = new (nothrow) bool[upper_bound];
    primes[0] = false; //segmentation fault occurs here
    primes[1] = false;
    for (long i = 2; i < upper_bound; i++)
       primes[i] = true;
    for (long number = 2; number < max_possible_prime; number++)
    {
        if (primes[number] == true)
        {
            if (upper_bound % number == 0)
            {
                max_prime_factor = number;
            }
            for (long j = number + number; j < upper_bound; j += number)
                primes[j] = false;
        }
        else { continue; }
    }
    return max_prime_factor;
}
int main ( int argc, char *argv[] )
{
    cout<<"Problem 3: "<<problem3()<<endl;
}

构建此代码并按原样运行它会在以下行上给出分段错误:

primes[0] = false

如果我删除nothrow指令以更改此行:

primes = new (nothrow) bool[upper_bound];

对此:

primes = new bool[upper_bound];

我收到一条错误消息,指出:

terminate called after throwing an instance of 'std::bad_alloc'

我认为这意味着分配失败,可能是因为大小(基于类似的问题和其他参考链接。

CodeBlock 中的调试器显示,即使在应该分配primes之后,仍设置为0x0。瓦尔格林德证实了这一点:

==15436== Command: ./main
==15436== 
==15436== Invalid write of size 1
==15436==    at 0x400A81: problem3() (main.cpp:54)
==15436==    by 0x400B59: main (main.cpp:77)
==15436==  Address 0x0 is not stack'd, malloc'd or (recently) free'd
==15436== 
==15436== 
==15436== Process terminating with default action of signal 11 (SIGSEGV)
==15436==  Access not within mapped region at address 0x0
==15436==    at 0x400A81: problem3() (main.cpp:54)
==15436==    by 0x400B59: main (main.cpp:77)
==15436==  If you believe this happened as a result of a stack
==15436==  overflow in your program's main thread (unlikely but
==15436==  possible), you can try to increase the size of the
==15436==  main thread stack using the --main-stacksize= flag.
==15436==  The main thread stack size used in this run was 8388608.
==15436== 
==15436== HEAP SUMMARY:
==15436==     in use at exit: 0 bytes in 0 blocks
==15436==   total heap usage: 1 allocs, 0 frees, 0 bytes allocated
==15436== 
==15436== All heap blocks were freed -- no leaks are possible
==15436== 
==15436== For counts of detected and suppressed errors, rerun with: -v
==15436== ERROR SUMMARY: 1 errors from 1 contexts (suppressed: 3 from 3)
Segmentation fault 

问:我知道std::vector,所以我应该使用它来分配这个数组吗?我也愿意尝试不同的算法,但我想知道我是否缺少C++细微差别,允许我分配这样的数组(即使它绝对很大,我理解这一点)。我也尝试尽可能多地调试这个问题,但是如果我应该提供其他内容,请告诉我,以便下次遇到麻烦时可以使用这些工具。

一个非常简单的算法,可以用来分解你正在使用的(大)数字是Pollard的Rho算法。

为了简洁起见,我将省略数学解释,但您可以在维基百科文章中找到详细信息。

unisgned long long GCD(unisgned long long x, unisgned long long y)
{
    while (y != 0)
    {
        unsigned long long t = b;
        b = a % b;
        a = t;
    }
    return a;
}
unsigned long long f(unsigned long long x, unsigned long long n)
{
    return (x * x + 1) % n;
}
unsigned long long PollardRho(unsigned long long n)
{
    unsigned long long x = 2, y = 2, d = 1;
    while (d == 1)
    {
        x = f(x);
        y = f(f(y));
        d = GCD(std::abs(x - y), n);
    }
    if (d == n)
        throw "Failure";
    return d;
}
unsigned long long MaxFactor(unsigned long long n)
{
    unsigned long long largest = 1;
    while (n != 1)
    {
        unsigned long long factor = PollardRho(n);
        largest = std::max(largest, factor);
        n /= factor;
    }
    return largest;
}

注意:我实际上并没有测试C++代码。我在 Mathematica 中对其进行了原型设计,它正确地返回了最大素因数6857

使用专门

为此目的而设计的std::vector<bool>std::bitset,同时考虑到数据密度和快速操作。在常规的布尔数组中,每个元素至少分配一个字节,而不是一个位。

假设600851475143是一个合数,其最大素因数小于或等于sqrt(600851475143)或大约775146。您的筛子不需要比这大。

这个问题(欧拉项目#3)也可以通过简单的暴力分解来解决。在我的台式电脑上,这种方法只需要大约 .002 秒。