Sorting a 2D vector of doubles

Question

I have a list of data (in 4 columns) that I would like to sort by a certain column. It was read in from a file to a 2D vector. I the used the std::sort method and wrote my comparator functor. The program compiles and runs, but when I print the first 10 elements it is not sorted, but is certainly different from the order it was added to the 2D vector.

Here is the code:

#include <iostream>
#include <fstream>
#include <stdio.h>
#include <string>
#include <vector>
#include <algorithm>
#include <iomanip>

using namespace std;

typedef vector<double> Row;
typedef vector<Row> Matrix;

bool sortByFourthColumn(const Row& row1, const Row& row2){
    return (double) row1[3] < (double) row2[3];
}

int main(){
    std::ifstream infile;
    infile.open("Test8_Output.txt");

    double x,y,z,E;
    char line[200];
    int line_count=0; 

    ofstream outfile;
    outfile.open("WO_crp.txt");

    if (infile.is_open()){

        while (!infile.eof()){
            infile.getline(line,170);

            if (line[0] != '%'){
            outfile<<line<<"\n";         
            line_count++;
            }
            else{
            }
    }

    Matrix data(line_count,Row(4));

    outfile.close();

    std::ifstream myfile;
    myfile.open("WO_crp.txt");

    int i = 0;
    while(myfile >> x >> y >> z >> E){
        data[0][i] = x;
        data[1][i] = y;
        data[2][i] = z;
        data[3][i] = E;
        i++;
    }
    myfile.close();

    std::sort(data.begin(), data.end(), sortByFourthColumn);

    for (int u = 0; u <20; u++){
        cout << setprecision(5) << data[0][u] << "\t" << setprecision(5)<< data[1][u] << "\t" << setprecision(5)<< data[2][u] << "\t" << setprecision(5)<< data[3][u] << endl;
    }

    }
    else{
        cout << "Error: File is invalid.\n";

    }
    return(0);
}

EDIT - Sample of what the input file looks like:
EDIT 2 - swapped 4 and line_count in Matrix data(4,Row(line_count));

% Model:              CDS_Test8.mph
% Version:            COMSOL 5.2.0.220
% Date:               Jul 13 2016, 14:33
% Dimension:          3
% Nodes:              86183
% Expressions:        1
% Description:        Electric field norm
% Length unit:        m
% x                       y                        z                        es.normE (V/m)
0.13774675805195374       0.05012986567931247      0.20735                  67.35120820901535
0.13870000000000005       0.04957489750396299      0.20735000000000003      102.8772500513651
0.13870000000000002       0.050800000000000005     0.20735                  87.56008679032011
0.13792733849817027       0.050131465727838186     0.20801419247484804      73.55192534768238
0.13674627634411463       0.04992349737428063      0.20735                  63.23018910026428
0.13750191177019236       0.0508                   0.20735000000000003      67.26176884022838
0.13827743496772454       0.05193409099097887      0.20734999999999998      73.35474409597487
0.13803618792088135       0.05134931748395268      0.20841988134890965      75.3712126982815
0.13905949760011943       0.05141879754884912      0.20734999999999998      83.70739713476813
0.13896970815034013       0.05092428105421264      0.208142746399683        84.73571510992372
0.1390220807917094        0.0501245422629353       0.20817502757007986      85.57119242707628
0.13944867847480893       0.05161480113017738      0.2081969878426443       89.65643851109644

And so on it goes for another 87k lines or so.

I'm confused. Doesn't std::sort take care of the sorting (and therefore swapping)? Why should there be a swap function? — Chami Sangeeth Amarasinghe, Jul 19 '16 at 16:03
With `Matrix data(4,Row(line_count))` you are declaring a vector of 4 elements of type `Row(line_count)` (a vector of line_count doubles). You should swap 4 with line_count and so on in the rest of the program. — Bob__, Jul 19 '16 at 16:08
Thank you for spotting this @Bob__. I did not understand the reason, nevertheless I implemented the change. The output looks 'more' sorted now. However it is not. Could it be a precision issue? — Chami Sangeeth Amarasinghe, Jul 19 '16 at 16:21
Can you please, provide an example of the input file? Are you copying that file only to get rid of comments lines (the ones starting with '%')? — Bob__, Jul 19 '16 at 16:24
Done. The lines starting with '%' do not strictly need to be gotten rid of; however, I did it that way so that filling the vector would be easier. — Chami Sangeeth Amarasinghe, Jul 19 '16 at 16:32
You still have the dimensions reversed when indexing. Now that `line_count` is the first index, your loop index `i` or `u` must also be the first index. — Peter, Jul 19 '16 at 16:45
Thanks very much @Peter! I did not follow through @Bob__'s suggestions fully. It works now. — Chami Sangeeth Amarasinghe, Jul 19 '16 at 16:52

score 2 · Accepted Answer · edited May 23 '17 at 12:08

I have a list of data (in 4 columns) that I would like to sort by a certain column.

The problem is that the dimensions of the vector of vectors used to store the data in OP program is not consistent between declaration and use.

A minor problem is the use of while(!infile.eof()){... which should be avoided.

A fixed version is like this:

#include <iostream>
#include <fstream>
#include <sstream>
#include <string>
#include <vector>
#include <array>
#include <algorithm>
#include <iomanip>

using Row = std::array<double,4>;     // instead of typedefs
using Matrix = std::vector<Row>;
using std::cout;

bool sortByFourthColumn(const Row& row1, const Row& row2){
    return row1[3] < row2[3];
    //     ^  The cast is unnecessary
}

int main(){

    std::string file_name{"Test8_Output.txt"};
    std::ifstream infile{file_name, std::ios_base::in};
    if ( !infile ) {
        cout << "Error: unable to open file " << file_name << '\n';
        return EXIT_FAILURE;
    }

    Matrix data;
    data.reserve(90000);  // if you are afraid of the reallocations
    int count = 0;
    std::string line;

    // instead of two loops you can use one loop and read the file once
    // I'll use std::getline to extract a row in a std::string
    while ( std::getline(infile, line) ) {
        // skip comments and empty lines
        if ( line.empty() || line[0] == '%' )
            continue;
        ++count;

        // extract data from the string using a stringstream
        std::stringstream ss{line};
        Row r;
        ss >> r[0] >> r[1] >> r[2] >> r[3];
        if ( !ss ) {
            cout << "Format error in line " << count << " of file.\n";
            break;
        }
        data.push_back(std::move(r));
    }

    std::sort(data.begin(), data.end(), sortByFourthColumn);
    cout << std::setprecision(5) << std::fixed;

    for ( const auto & r : data ) {
        for ( auto const &x : r ) {
            cout << std::setw(10) << x;
        }
        cout << '\n';
    }
    return EXIT_SUCCESS;
}

The output, given the example data is:

   0.13675   0.04992   0.20735  63.23019
   0.13750   0.05080   0.20735  67.26177
   0.13775   0.05013   0.20735  67.35121
   0.13828   0.05193   0.20735  73.35474
   0.13793   0.05013   0.20801  73.55193
   0.13804   0.05135   0.20842  75.37121
   0.13906   0.05142   0.20735  83.70740
   0.13897   0.05092   0.20814  84.73572
   0.13902   0.05012   0.20818  85.57119
   0.13870   0.05080   0.20735  87.56009
   0.13945   0.05161   0.20820  89.65644
   0.13870   0.04957   0.20735 102.87725

Thank you for pointing out the key mistake, and also for the suggestions. I got the program working. I will implement most of your version - it's much more streamlined. — Chami Sangeeth Amarasinghe, Jul 19 '16 at 23:28
@ChamiSangeethAmarasinghe You are welcome. Glad to be helpful. — Bob__, Jul 20 '16 at 06:05

Sorting a 2D vector of doubles

1 Answers1