How do I sort a std::vector by the values of a dif

Question

I have several std::vector, all of the same length. I want to sort one of these vectors, and apply the same transformation to all of the other vectors. Is there a neat way of doing this? (preferably using the STL or Boost)? Some of the vectors hold ints and some of them std::strings.

Pseudo code:

std::vector<int> Index = { 3, 1, 2 };
std::vector<std::string> Values = { "Third", "First", "Second" };

Transformation = sort(Index);
Index is now { 1, 2, 3};

... magic happens as Transformation is applied to Values ...
Values are now { "First", "Second", "Third" };

Answer 1

Only one rough solution comes to my mind: create a vector that is the sum of all other vectors (a vector of structures, like {3,Third,...},{1,First,...}) then sort this vector by the first field, and then split the structures again.

Probably there is a better solution inside Boost or using the standard library.

Answer 2

ltjax's answer is a great approach - which is actually implemented in boost's zip_iterator http://www.boost.org/doc/libs/1_43_0/libs/iterator/doc/zip_iterator.html

It packages together into a tuple whatever iterators you provide it.

You can then create your own comparison function for a sort based on any combination of iterator values in your tuple. For this question, it would just be the first iterator in your tuple.

A nice feature of this approach is that it allows you to keep the memory of each individual vector contiguous (if you're using vectors and that's what you want). You also don't need to store a separate index vector of ints.

Answer 3

This would have been an addendum to Konrad's answer as it an approach for a in-place variant of applying the sort order to a vector. Anyhow since the edit won't go through I will put it here

Here is a in-place variant with a slightly higher time complexity that is due to a primitive operation of checking a boolean. The additional space complexity is of a vector which can be a space efficient compiler dependent implementation. The complexity of a vector can be eliminated if the given order itself can be modified.

Here is a in-place variant with a slightly higher time complexity that is due to a primitive operation of checking a boolean. The additional space complexity is of a vector which can be a space efficient compiler dependent implementation. The complexity of a vector can be eliminated if the given order itself can be modified. This is a example of what the algorithm is doing. If the order is 3 0 4 1 2, the movement of the elements as indicated by the position indices would be 3--->0; 0--->1; 1--->3; 2--->4; 4--->2.

template<typename T>
struct applyOrderinPlace
{
void operator()(const vector<size_t>& order, vector<T>& vectoOrder)
{
vector<bool> indicator(order.size(),0);
size_t start = 0, cur = 0, next = order[cur];
size_t indx = 0;
T tmp; 

while(indx < order.size())
{
//find unprocessed index
if(indicator[indx])
{   
++indx;
continue;
}

start = indx;
cur = start;
next = order[cur];
tmp = vectoOrder[start];

while(next != start)
{
vectoOrder[cur] = vectoOrder[next];
indicator[cur] = true; 
cur = next;
next = order[next];
}
vectoOrder[cur] = tmp;
indicator[cur] = true;
}
}
};

Answer 4

friol's approach is good when coupled with yours. First, build a vector consisting of the numbers 1…n, along with the elements from the vector dictating the sorting order:

typedef vector<int>::const_iterator myiter;

vector<pair<size_t, myiter> > order(Index.size());

size_t n = 0;
for (myiter it = Index.begin(); it != Index.end(); ++it, ++n)
    order[n] = make_pair(n, it);

Now you can sort this array using a custom sorter:

struct ordering {
    bool operator ()(pair<size_t, myiter> const& a, pair<size_t, myiter> const& b) {
        return *(a.second) < *(b.second);
    }
};

sort(order.begin(), order.end(), ordering());

Now you've captured the order of rearrangement inside order (more precisely, in the first component of the items). You can now use this ordering to sort your other vectors. There's probably a very clever in-place variant running in the same time, but until someone else comes up with it, here's one variant that isn't in-place. It uses order as a look-up table for the new index of each element.

template <typename T>
vector<T> sort_from_ref(
    vector<T> const& in,
    vector<pair<size_t, myiter> > const& reference
) {
    vector<T> ret(in.size());

    size_t const size = in.size();
    for (size_t i = 0; i < size; ++i)
        ret[i] = in[reference[i].first];

    return ret;
}

Answer 5

You can probably define a custom "facade" iterator that does what you need here. It would store iterators to all your vectors or alternatively derive the iterators for all but the first vector from the offset of the first. The tricky part is what that iterator dereferences to: think of something like boost::tuple and make clever use of boost::tie. (If you wanna extend on this idea, you can build these iterator types recursively using templates but you probably never want to write down the type of that - so you either need c++0x auto or a wrapper function for sort that takes ranges)

Answer 6

A slightly more compact variant of xtofl's answer for if you are just looking to iterate through all your vectors based on the of a single keys vector. Create a permutation vector and use this to index into your other vectors.

#include <boost/iterator/counting_iterator.hpp>
#include <vector>
#include <algorithm>

std::vector<double> keys = ...
std::vector<double> values = ...

std::vector<size_t> indices(boost::counting_iterator<size_t>(0u), boost::counting_iterator<size_t>(keys.size()));
std::sort(begin(indices), end(indices), [&](size_t lhs, size_t rhs) {
    return keys[lhs] < keys[rhs];
});

// Now to iterate through the values array.
for (size_t i: indices)
{
    std::cout << values[i] << std::endl;
}