Find the maximum value in vector<vector<int>> without a 'for' loop

Question

We know that for vector<int> A, we can use *max_element(A.begin(), A.end()) to find the maximum value in A. However, I would like to know whether there is clean way to find the maximum value in a vector<vector<int>> B, avoiding using for loop?

If we use a for loop, the code could be trivial like:

int maxvalue = INT_MIN;
for (int i = 0; i < m; i++)
    for (int j = 0; j < n; j++)
        maxvalue = max(maxvalue, B[i][j]);

or

int maxvalue = INT_MIN;
for (int i = 0; i < m; i++)
{
    int temp = *max_element(B[i].begin(), B[i].end());
    maxvalue = max(maxvalue, temp);
}

But I still feel it not clean enough. And I don't like the for loop.

Finally, I chose the following code to do it:

auto itv = max_element(A.begin(), A.end(), [](vector<int>& a, vector<int>& b)
        { return *max_element(a.begin(), a.end()) < *max_element(b.begin(), b.end()); });
int ans = *max_element(itv->begin(), itv->end());

For loops are great. Combine them with iterators and you are gold. — Quirk
– Quirk, Commented Mar 25, 2016 at 21:25
@c-wang: Have you tried using a custom comparator function which compares between *max_element() of two vector<int>s ? — Quirk
– Quirk, Commented Mar 25, 2016 at 21:26
@c-wang: I haven't tested it. I am not even sure if something like that would work. Just a passing thought. — Quirk
– Quirk, Commented Mar 25, 2016 at 21:31
Loops are fine. I wouldn't worry about using a more complicated method. — David
– David, Commented Mar 25, 2016 at 21:38

Edward Strange · Accepted Answer · 2016-03-25 21:40:33Z

5

auto max_value = std::accumulate(std::begin(B), std::end(B),
      std::numeric_limits<int>::min(), 
      [] (int cur_max, auto && vec) 
      { 
        return std::max(cur_max, *std::max_element(std::begin(v), std::end(v));
      });

edited Mar 25, 2016 at 21:40

answered Mar 25, 2016 at 21:35

Edward Strange

41k9 gold badges80 silver badges128 bronze badges

Sign up to request clarification or add additional context in comments.

Comments

Quirk · Accepted Answer · 2016-03-25 21:58:28Z

0

I used a custom comparison operator in max_element() to get the desired effect. No loops whatsoever, except the implied accumulation that max_element() runs.

bool mycomp(vector<int> a, vector<int> b) {
  return *max_element(a.begin(), a.end()) < *max_element(b.begin(), b.end());
}

vector<vector<int>> vv; // our vector of vectors

auto itv = max_element(vv.begin(), vv.end(), mycomp); // find the vector 
                                                     // with the max element

int answer = *max_element((*itv).begin(), (*itv).end()); // finds the max element 
                                                        // in the desired vector

This is by no means clean. But it does what it says it does.

answered Mar 25, 2016 at 21:58

Quirk

1,3474 gold badges17 silver badges29 bronze badges

6 Comments

C. Wang Over a year ago

is it possible to use lambda function to replace mycomp?

Quirk Over a year ago

With reference to the comment: stackoverflow.com/questions/36228231/…

Quirk Over a year ago

@c-wang: I see no reason why not.

Quirk Over a year ago

Thank you for accepting the answer. It is worthy to note that @crazy-eddie code does the same thing, essentially (in perhaps a more efficient way).

Edward Strange Over a year ago

This version makes WAY more comparisons than mine.

|

Rakete1111 · Accepted Answer · 2016-03-25 21:38:55Z

0

Instead of the for loop you could use std::for_each. Maybe something like:

int maxvalue = std::numeric_limits<int>::min();
std::for_each(std::begin(B), std::end(B), [&maxvalue](const auto& v)
{
    maxvalue = std::max(maxvalue, *std::max_element(std::begin(v), std::end(b)));
});

answered Mar 25, 2016 at 21:38

Rakete1111

49.5k19 gold badges133 silver badges173 bronze badges

Comments

Peter Mortensen · Accepted Answer · 2021-07-14 00:08:49Z

0

If the thing you want to avoid using loops are long structures in your program, using C++11 you can find the maximum in a single line with loops:

std::vector< std::vector<int> > w;


int max = 0;
for (auto i : w) for (auto j : i) max = j > max ? j : max;

Or

int max = 0;
for (auto i : w) for (auto j : i) if (j > max) max = j;

Anyway, I don't think this is a good practice. This option would be better:

int max = 0;
for (auto i : w)
    for (auto j : i)
        max = j > max ? j : max;

edited Jul 14, 2021 at 0:08

Peter Mortensen

31.4k22 gold badges110 silver badges134 bronze badges

answered Mar 25, 2016 at 21:47

Ediolot

5312 gold badges7 silver badges21 bronze badges

4 Comments

Yam Marcovic Over a year ago

"without for loop"

Ediolot Over a year ago

True, it was just something to make it cleaner, I will update the answer

Yam Marcovic Over a year ago

Worth nothing that putting things in a single line was possible even prior to C++11.

Ediolot Over a year ago

The point is using those smaller for loops. Of course you can even write all your c++ code in one single line.

Collectives™ on Stack Overflow

Find the maximum value in vector<vector<int>> without a 'for' loop

4 Answers 4

Comments

6 Comments

Comments

4 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

4 Answers 4

Comments

6 Comments

Comments

4 Comments

Your Answer

Sign up or log in

Post as a guest

Related