Count Inversions with merge sort in C++ - c++

I'm working on my first few algorithms to build my C++ skills and am currently coding up a method of counting inversions with merge sort. I've managed to get a working merge sort together but I'm having a bit of trouble keeping track of the number of inversions. Any ideas of where to go from here? How can I keep track of the number of inversions on a recursive algorithm like this? Additionally I've seen a couple different implementations of this in my internet travels and have found most people stray away from the std::vector method, any idea why? Thanks for any help, my code is below!
#include <iostream>
#include <math.h>
#include <vector>
using namespace std;
vector<int> print(vector<int> input){
for(int i=0; i<input.size(); i++){
cout<<input[i]<<",";
}
cout<<endl;
return input;
}
vector<int> merge(vector<int> left,vector<int> right){
//set up some varibles
vector<int> output;
int i=0;
int j=0;
//loop through the lists and merge
while(i<left.size() && j<right.size()){
//push the smallest of the two to the vector output
if(left[i]<=right[j]){
output.push_back(left[i]);
i+=1;
}
if(left[i]>right[i]){
output.push_back(right[j]);
j+=1;
}
}
//push the remnants of the vectors to output
for(i; i<left.size(); i++){
output.push_back(left[i]);
}
for(j; j<right.size(); j++){
output.push_back(right[j]);
}
return output;
}//end merge
vector<int> merge_sort(vector<int> input){
//check the size of the vector
if(input.size()<2){
return input;
}
else{
//int new vectors
vector<int> left;
vector<int> right;
vector<int> output;
//find the middle of the input vector
int middle=(input.size())/2;
//build the left vector
for(int i=0; i<middle; i++){
left.push_back(input[i]);
}
//build the right vector
for(int i=middle; i<input.size(); i++){
right.push_back(input[i]);
}
//make recursive calls
left=merge_sort(left);
right=merge_sort(right);
//call merge
output=merge(left,right);
return output;
}
}
int main()
{
vector<int> output;
vector<int> input;
input.push_back(2);
input.push_back(1);
input.push_back(10);
input.push_back(4);
output=merge_sort(input);
print(output);
}

Good news: counting inversions is pretty easy from here.
Think about your "merge" method. Every time you put an element from the left vector into output, you are not changing its position relative to elements from the right. On the other hand, every time you add an element from the right vector, you are putting it "before" all elements still to be processed in the left vector, when it was prevously "after" them, i.e. creating (left.size - i) "inversions".
You can prove this easily by induction if needed.
So the answer is simply : pass an int* to your merge method, and increment it by (left.size - i) every time you push an element from the right vector.
EDIT: Working code sample
#include <iostream>
#include <vector>
// removed useless dependency math.h
using namespace std;
// void type -> does not return anything
void print (vector<int> input) {
// range-based for loop (since C++ 11)
// no brackets -> only one instruction in for loop
for(int i : input)
cout << i << ",";
}
vector<int> merge (vector<int> left, vector<int> right, int * inv_count) {
vector<int> output;
// multiple variable definition of the same type
int i=0, j=0;
// spaces around "<", after "while", before "{" for readability
while (i < left.size() && j < right.size()) {
// one-instruction trick again
if (left[i] <= right[j])
// i++ is evaluated to <previous value of i> and then increments i
// this is strictly equivalent to your code, but shorter
// check the difference with ++i
output.push_back(left[i++]);
// else because the two conditions were complementary
else {
output.push_back(right[j++]);
// pointer incrementation
*inv_count += (left.size() - i);
}
}
// first field of for ommited because there is no need to initialize i
for(; i < left.size(); i++)
output.push_back(left[i]);
for(; j < right.size(); j++)
output.push_back(right[j]);
return output;
}
vector<int> merge_sort (vector<int> input, int * inv_count) {
// no-braces-idiom again
// spaces around "<" and after "if" for readability
if (input.size() < 2)
return input;
// no need for else keyword because of the return
// multiple variable definition
vector<int> left, right;
int middle = input.size() / 2;
// one-instruction for loop
for(int i=0; i < middle; i++)
left.push_back(input[i]);
for(int i=middle; i < input.size(); i++)
right.push_back(input[i]);
// no need for intermediate variable
return merge( merge_sort(left, inv_count),
merge_sort(right, inv_count),
inv_count);
}
// consistent convention : brace on the same line as function name with a space
int main () {
// vector initialization (valid only since C++ 11)
vector<int> input = {2, 1, 10, 4, 42, 3, 21, 7};
int inv_count = 0;
// No need for intermediate variables again, you can chain functions
print( merge_sort(input, &inv_count) );
// The value inv_count was modified although not returned
cout << "-> " << inv_count << " inversions" << endl;
}
I modified your code to include a few usual C++ idioms. Because you used the C++14 tag, I also used tricks available only since C++11. I do not recommend using all of these tricks everywhere, they are included here because it is a good learning experience.
I suggest you read about pointers before diving deeper into C++.
Also note that this code is in no way optimal : too many intermediate vectors are created, and vectors are not useful here, arrays would be enough. But I'll leave this for another time.

Related

Error when using std::vector::size to create another vector

I am learning DSA and while practising my LeetCode questions I came across a question-( https://leetcode.com/problems/find-pivot-index/).
Whenever I use vector prefix(size), I am greeted with errors, but when I do not add the size, the program runs fine.
Below is the code with the size:
class Solution {
public:
int pivotIndex(vector<int>& nums) {
//prefix[] stores the prefix sum of nums[]
vector<int> prefix(nums.size());
int sum2=0;
int l=nums.size();
//Prefix sum of nums in prefix:
for(int i=0;i<l;i++){
sum2=sum2+nums[i];
prefix.push_back(sum2);
}
//Total stores the total sum of the vector given
int total=prefix[l-1];
for(int i=0; i<l;i++)
{
if((prefix[i]-nums[i])==(total-prefix[i]))
{
return i;
}
}
return -1;
}
};
I would really appreciate if someone could explain this to me.
Thanks!
You create prefix to be the same size as nums and then you push_back the same number of elments. prefix will therefore be twice the size of nums after the first loop. You never access the elements you've push_backed in the second loop so the algorithm is broken.
I suggest that you simplify your algorithm. Keep a running sum for the left and the right side. Add to the left and remove from the right as you loop.
Example:
#include <numeric>
#include <vector>
int pivotIndex(const std::vector<int>& nums) {
int lsum = 0;
int rsum = std::accumulate(nums.begin(), nums.end(), 0);
for(int idx = 0; idx < nums.size(); ++idx) {
rsum -= nums[idx]; // remove from the right
if(lsum == rsum) return idx;
lsum += nums[idx]; // add to the left
}
return -1;
}
If you use vector constructor with the integer parameter, you get vector with nums.size() elements initialized by default value. You should use indexing to set the elements:
...
for(int i = 0; i < l; ++i){
sum2 = sum2 + nums[i];
prefix[i] = sum2;
}
...
If you want to use push_back method, you should create a zero size vector. Use the constructor without parameters. You can use reserve method to allocate memory before adding new elements to the vector.

Using recursion in bubble sort

Hello everyone I am starting to learn Data structures and Algorithms and implemented bubble sort myself after learning the concept. Following is the code I have written with my understanding but the problem is it runs for only one cycle and does not sort recursively.
For example:
{ 5,1,4,2,8} is sorted one time -> {1,4,2,5,8,}
What can be the problem?
vector<int> bubble_sort(vector<int> vec){
int temp = 0;
for(int i = 0; i < vec.size()-1; i++){
temp = vec.at(i+1); // holds the adjacent element.
// the following loop swaps the adjacent elements if the previous one is big
if(vec.at(i) > vec.at(i+1)){
vec.at(i+1) = vec.at(i);
vec.at(i) = temp;
}
temp = 0;
}
for(int i = 0; i < vec.size()-1; i++){
if(vec.at(i) > vec.at(i+1)){
bubble_sort(vec);
}
}
return vec;
}
Your function takes a vector<int> vector by copy, hence after first swaps only this copy is send to recursively sort.
Just add & to your function parameter: vector<int> bubble_sort(vector<int> &vec) and it should work
If you want to implement recursion fully and do not want to use for loop in the code, then follow this example. It will be helpful.
#include <iostream>
using namespace std;
/* Function to print an array */
void printArray(int arr[], int n)
{
for (int i=0; i <= n; i++)
cout<<arr[i];
}
void bubble_sort_recursive(int arr[], int j, int n) {
// base case
if (n==0 || j>n){
return;
}
// single pass has been completed and the higher element moved to right for that subarray
// now call the recursive function by keeping aside the already sorted positioned element
//i.e next pass wil start from this call
if (j == n){
bubble_sort_recursive(arr,0,n-1);
}
// swap consecutive 2 elements - main basic of bubble sort
if (arr[j]>arr[j+1]){
int t = arr[j];
arr[j] = arr[j+1];
arr[j+1] =t;
}
// go forward for next element of a single pass
bubble_sort_recursive(arr,j+1,n);
}
int main() {
int arr[] = {5,4,3,2,1};
// get the length of array
int n = sizeof(arr)/sizeof(arr[0]);
// call from 0 to len-1 as index starts from 0
bubble_sort_recursive(arr,0,n-1);
// print the sorted array
cout<<"Sorted array:"<<endl;
printArray(arr, n-1);
}

C++ Merge Sort Visualizer

I am trying to make a c++ console application that tries to show you how merge sort looks like. I understand merge sort, and I created a program that organizes a vector of strings called sort_visualize, and each string in it is filled with many #. This is completely randomized for every string. The merge sort will organize them depending on length, instead of the traditional number organizing people do with it. Every time I make a change to the vector, I also clear the screen and print out the entire vector through a draw function, to give the effect of it actively visualizing the sort every frame. The problem is that when I use the draw function to print out the entire sort_visualize string, it does not print out any changes that I have made to it, and prints out the same thing over and over again until the end, when it finally prints the sorted order. What is going on? I Don't understand. I even tried changing the draw(sort_visualize) to draw(sort_visualize_), and that shows small areas of the vector it is working on. Makes no sense. Please try this code and tell me any solutions. Thank you.
Here's the code:
#include <vector>
#include <iostream>
#include <ctime>
#include "stdlib.h"
#include "windows.h"
using namespace std;
void merge_sort(vector<string> &sort_visual_);
void merge_halves(vector<string>&left, vector<string>& right, vector<string>& sort_visual_);
void draw(vector <string> &sort_visual_);
vector <string> sort_visual;
int main()
{
srand(time(NULL));
//vector
vector<int> num_list;
//fill vector with random integers
for (int i = 0; i < 40; i++)
num_list.push_back(rand() % 40);
//Fill the visualizer strings which will be bars with #'s
for (int i = 0; i < num_list.size(); i++)
{
sort_visual.push_back("");
string temp;
for (int j = 0; j < num_list.at(i); j++)
{
temp.push_back('#');
}
sort_visual.at(i) = temp;
}
draw(sort_visual);
system("pause");
//sort function
merge_sort(sort_visual);
}
void merge_sort(vector<string> &sort_visual_)
{
//dont do anything if the size of vector is 0 or 1.
if (sort_visual_.size() <= 1) return;
//middle of vector is size/2
int mid = sort_visual_.size() / 2;
//2 vectors created for left half and right half
vector<string> left;
vector<string> right;
//divided vectors
for (int j = 0; j < mid; j++)
{
left.push_back(sort_visual_[j]); //add all the elements from left side of original vector into the left vector
}
for (int j = 0; j < (sort_visual_.size()) - mid; j++)
{
right.push_back(sort_visual_[mid + j]);//add all the elements from right side of original vector into the right vector
}
//recursive function for dividing the left and right vectors until they are length of 1
merge_sort(left);
merge_sort(right);
//do the actual merging function
merge_halves(left, right, sort_visual_);
}
void merge_halves(vector<string>&left, vector<string>&right, vector<string>& sort_visual_) //pass in 3 vectors
{
// sizes of each vector (left and right)
int nL = left.size();
int nR = right.size();
//declaring variables pointint to elements for each vector. i will represent finished produce vector
int i = 0, j = 0, k = 0;
//as long as j and k are less than the left and right sizes
while (j < nL && k < nR)
{
if (left[j].length() < right[k].length()) //if the string in the left vector is smaller than string in right vector
{
sort_visual_[i] = left[j];//ad the string from left vector in the sort_visuals vector(which is the final product)
j++;//increment j to move on
}
else
{
sort_visual_[i] = right[k];//otherwise add the string from right vector in the sort_visual vector
k++; //increment k to move on
}
i++; //i is the final vector, and we have to increment it to set it up to take in the next number
system("CLS");
draw(sort_visual);
Sleep(15);
}
while (j < nL)
{
sort_visual_[i] = left[j];
j++; i++;
system("CLS");
draw(sort_visual);
Sleep(15);
}
while (k < nR)
{
sort_visual_[i] = right[k];
k++; i++;
system("CLS");
draw(sort_visual);
Sleep(15);
}
}
void draw(vector <string> &sort_visual)
{
for (int i = 0; i < sort_visual.size(); i++)
{
cout << sort_visual.at(i) << endl;
}
}
In merge_halves you work on sort_visual_ but draw sort_visual which is a global that does not seem to be changed. Make sure there are no globals and it will be harder to make mistakes.

Deleting elements from a vector that meet a condition

I am trying to program the Sieve of Eratosthenes, but I am not sure how to delete elements from the vector I made given a specific condition. Does anyone know how to achieve this? Here is my code:
#include <iostream>
#include <vector>
using namespace std;
int prime(int n);
int prime(int n)
{
vector<int> primes;
for(int i = 2; i <= n; i++)
{
primes.push_back(i);
int t = i % (i + 1);
if(t == 0)
{
delete t; // is there a way of deleting the elements from
// the primes vector that follow this condition t?
}
cout << primes[i] << endl;
}
}
int main()
{
int n;
cout << "Enter a maximum numbers of primes you wish to find: " << endl;
cin >> n;
prime(n);
return 0;
}
Your algorithm is wrong:
t = i % (i + 1);
is
i
which is always != 0 because i is larger than 1.
By the way if you absolutely want to remove the t-th element you have to be sure that the vector is not empty and then you do:
primes.erase(primes.begin()+t);
Even if you fix the algorithm your approach is inefficient: erasing an element in the middle of a vector means copying back of one position all the ones following the erased element.
You don't usually want to delete elements in the middle of a Sieve of Eratosthenes, but when you do want to, you usually want to use the remove/erase idiom:
x.erase(std::remove_if(x.begin(), x.end(), condition), x.end());
std::remove basically just partitions the collection into those that don't meet the specified condition, followed by objects that may have been used as the source of either a copy or a move, so you can't count on their value, but they are in some stable state so erasing them will work fine.
The condition can be either a function or a functor. It receives (a reference to a const) object that it examines and determines whether it lives or dies (so to speak).
Find here a c++ pseudocode for the sieve algorithm. Once you've understood the algorithm you can start working on this.
primes(vector& primes, size_t max){
vector primesFlag(1,max);
i=1
while(i*i<max){
++i;
for(j=i*i; j < max; j+= i){
primesFlag[j] = 0;
}
}
primes.clear()
primes.reserve(...);
for(j >= 2;
if primesFlag[j] = 1
primes.push_back(j);
}

How to sort an array in C++ in a specific way

I want somehow sort an array, so that it looks like -
a[0]>=a[1]<=a[2]>=a[3]<=a[4]
I don't know where to start.
Any suggestion would be appreciated!
Sort the entire array (Choose any sort algorithm you wish to). Then take each pair from the beginning and swap the elements in the pair
2,4,1,5,6,3,7,9,8,10
Sorted to : 1,2,3,4,5,6,7,8,9,10
Pair and swap : (2,1),(4,3),(6,5),(8,7),(10,9)
result : 2,1,4,3,6,5,8,7,10,9
Here's the code, obviously you can alter the array length and numbers to meet your specifications.
#include <iostream>
#include <algorithm>
using namespace std;
void special_Sort(int *array, int size){
//doesn't return a value, changes the values inside the array
int temp;
//for swapping purposes
sort(array, array+size);
//sorts the array in ascending order
for(int i=0; i<size; i=i+2){
temp=array[i];
array[i]=array[i+1];
array[i+1]=temp;
}
//array is now sorted
}
int main(){
// array declaration, call the function, etc...
int array[10]={2,4,1,5,6,3,7,9,8,10};
int *pointer;
pointer=&array[0];
special_Sort(pointer, 10);
// if you want to print the result
// for(int i =0; i<10; i++)
// cout<<array[i]<<" ";
return 0;
}
I'm assuming here that the relations are inclusive (in the sense that they continue to the end of the line - a[0]>=max(a[1],a[2],...), and a[1]<=min(a[2],a[3],..) and so on). Otherwise this isn't uniquely defined, as {5,4,3,2,1} can get sorted for example into {5,1,4,3,2} or {3,2,5,1,4}.
So, assuming this is the case, it's easily solved by sorting the entire array in descending order, then just interleave them -
a[0], a[n-1], a[1], a[n-2], ...
and so on. Just loop with two indices, one starting from the beginning and one from the end, or use something like this -
for (i=0; i<n/2; i++) {
result[i*2] = sorted[i];
result[i*2+1] = sorted[n-i];
}
if (n%2)
result[n-1] = sorted[n/2]
If you are only sorting it in a way that you want values to rise and fall arbitrarily, you can achieve this by checking values in your array and swapping elements if they do not satisfy the constraints of your sort.
Don't have a compiler on me at the moment and you'd have to implement the swap but something like this could work:
for(i=0; i < a.length(); i++){
//If index is even
if(i%2 == 0){
if(a[i] < a[i+1]){
swap(a[i], a[i+1]);
}
} else { ///If index is odd
if(a[i]>a[i+1]){
swap(a[i], a[i+1];
}
}
}
I don't disagree with the other answers posted here so you will have to find what you need depending on the relation of the even and odd indexed elements.
Steps taken:
1) generate some random array
2) sort array
3) switch elements as needed with alternate <=, >= comparisons
Here's the code that does that: (disregard the random generator, its just an easy way to generate an array)
#define sizeArr 50
int main(void)
{
int array[sizeArr];
int i, temp;
for(i=0;i<sizeArr;i++)
{
array[i]=randomGenerator(1, 1000);
Sleep(2);//force clock tick for new srand() to be effective in rand() generator
}
//sort array
qsort(array, sizeArr, sizeof(int), cmpfunc);
//pick the first non repeat 90th percent and print
for(i=0;i<sizeArr-1;i++)
{
if(i%2==0)//alternate between >= && <=
{
if(array[i+1] >= array[i])
{
temp = array[i+1];
array[i+1]=array[i];
array[i]=temp;
}
}
else
{
if(array[i+1] <= array[i])
{
temp = array[i+1];
array[i+1]=array[i];
array[i]=temp;
}
}
}
getchar();
return 0;
}
int cmpfunc (const void * a, const void * b)
{
return ( *(int*)a - *(int*)b );
}
int randomGenerator(int min, int max)
{
int random=0, trying=0;
trying = 1;
srand(clock());
while(trying)
{
random = (rand()/32767.0)*(max+1);
(random >= min) ? (trying = 0) : (trying = 1);
}
return random;
}