Argmin for static arrays - c++

Question

Hello all I am trying to implement the Argmin function. following the example from one of the comments here ArgMin for vector<double> in C++?, I have written the following logic.
rev_tone = get_rev_tone(cam_model_path);
//reading the images
Mat img_rev = imread("C:/Users/20181217/Desktop/images/imgs/output_rev.png");
Mat ideal = imread("C:/Users/20181217/Desktop/images/imgs/output_fwd_v6.png");
int idx = 256;
int no_of_channels = 3;
float rev_tone_s[256][3];
///dynamic to static conversion
for (int i = 0; i < rev_tone.size(); i++) {
for (int j = 0; j < rev_tone[i].size(); j++) {
rev_tone_s[i][j] = rev_tone[i][j];
}
}
int dist = distance(rev_tone.begin(), std::min_element(rev_tone.begin(), rev_tone.end()));//rev_tone is of type vector<vector<float>>
cout << dist << endl;
int dist_s = distance(rev_tone_s.begin(), std::min_element(rev_tone_s.begin(), rev_tone_s.end())); //rev_tone_s is a static 2d float of type float[256][3]
cout << dist_s << endl;
when i am executing the program, this line is working without any problem.
int dist = distance(rev_tone.begin(), std::min_element(rev_tone.begin(), rev_tone.end()));//rev_tone is of type vector<vector<float>>
But the nature of my project doesn't allow any dynamic memory allocations. so I have converted the Vectors into arrays (float_rev_tone_s). When I try to perform the Argmin() on the converted array in the following line
int dist_s = distance(rev_tone_s.begin(), std::min_element(rev_tone_s.begin(), rev_tone_s.end())); //rev_tone_s is a static 2d float of type float[256][3]
it gives me an error saying expression must have a class type. I have looked at this error and it says it was a pointer issue. My question is why is it only a problem for an array but not for vector??
According to the cppreference, .begin() of an array also gives the initial iterator.
Is there a way to work around this for defining Argmin() over an array?
Any suggestions will be highly appreciated.
Thanks in advance

Answer 1

It seems like you're confusing C++'s std::array with C-style arrays (which is what you're using in the code you posted):
float rev_tone_s[256][3];
This is a C-style array, not a std::array. C-style arrays are basically raw pointers to objects of the underlying type and a pointer does not have the begin method. std::array on the other hand is an STL container which has the begin method.
This means you could just change the type of your rev_tone_s variable if you want to have stuff like STL-iterators:
std::array<std::array<float, 256>, 3> rev_tone_s;

Answer 2

My question is why is it only a problem for an array but not for vector?
vector is an object, so it has member functions like size() and begin() and end() for iterators. While array is just a raw type, just like and int or a float and not an object, so you don't have any member functions for it.
According to the cppreference, .begin() of an array also gives the initial iterator
That is correct, but that array is std::array which is a wrapper over raw arrays. It doesn't allocate any memory at runtime so you can and should use that. Advantage of using the std::array over raw array is that
You get member functions like size() and iterators.
It won't decay to a pointer whenever you pass it as an argument to a function.
Here's how you declare a 2D std::array:
std::array<std::array<float, 256>, 3> arr ;
Note that you need to #include <array> for this to work.

Answer 3

The problem is that you are trying to invoke begin() on a pointer
Here is an example of using distance with pointer:
#include <iostream>
#include <algorithm>
const int arr_size = 100;
float myarray[arr_size];
int main()
{
for(int i = 0; i < arr_size; ++i) {
myarray[i] = static_cast<float>(100 - i);
}
std::cout << std::distance(myarray, std::min_element(myarray, myarray + arr_size)) << '\n';
return 0;
}