Problem 81: Path sum: two ways

(see projecteuler.net/problem=81)

In the 5 by 5 matrix below, the minimal path sum from the top left to the bottom right,
by only moving to the right and down, is indicated in bold and is equal to 2427.

131 673 234 103 18
201 96 342 965 150
630 803 746 422 111
537 699 497 121 956
805 732 524 37 331

Find the minimal path sum, in matrix.txt (right click and "Save Link/Target As..."), a 31K text file containing a 80 by 80 matrix, from the top left to the bottom right by only moving right and down.

Algorithm

That's a classic example for breadth-first search (see )
Initialization: create a priority-queue (descendingly ordered by the path sum / called weight in my program) and insert the upper left corner.

As long as we haven't reached the destination:
- pick the position with the lowest weight (=partial path sum)
- discard it if we have already processed that position
- mark it as processed
- add its right and bottom neighbor to the priority queue, with their numbers added to the current weight

C++'s STL comes with a priority_queue. Unfortunately its top() returns the largest value but I need the smallest.
Therefore my struct Cell has operator<() implemented "the wrong way" on purpose.

Note

Project Euler's file has all values stored in a CSV (comma-separated values) format.
The C++ code looks a bit ugly because C++ recognizes only whitespaces by default.

My code

… was written in C++11 and can be compiled with G++, Clang++, Visual C++. You can download it, as well as the input data, too.

The code contains #ifdefs to switch between the original problem and the Hackerrank version.
Enable #ifdef ORIGINAL to produce the result for the original problem (default setting for most problems).

#include <queue>
#include <vector>
#include <iostream>
 
// 2D matrix: unfortunately x and y are swapped, so we need to write matrix[y][x]
// instead of the more common matrix[x][y]
typedef std::vector<std::vector<unsigned int>> Matrix;
 
// use a priority queue to find the next cell to process
struct Cell
{
// position
unsigned int x, y;
// sum of shortest route so far
unsigned long long weight;
 
Cell(unsigned int x_, unsigned int y_, unsigned long long weight_)
: x(x_), y(y_), weight(weight_) {}
 
// std::priority_queue returns the LARGEST element, therefore I implement this function "the other way 'round"
bool operator<(const Cell& cell) const
{
return weight > cell.weight; // ">" is not a typo !
}
};
 
// breadth-search
unsigned long long search(const Matrix& matrix)
{
// matrix height/width
const auto size = matrix.size();
 
// store already processed cells
std::vector<std::vector<bool>> processed(matrix.size());
for (auto& row : processed)
row.resize(matrix.size(), false);
 
// process cells in increasing distance from starting point
std::priority_queue<Cell> next;
// add starting point (upper left corner)
next.push(Cell(0, 0, matrix[0][0]));
 
while (!next.empty())
{
// get cell with the smallest weight
Cell cell = next.top();
// and remove it from the queue
next.pop();
 
// we have been here before ?
// must have been on a shorter route, hence discard current cell
if (processed[cell.y][cell.x])
continue;
 
processed[cell.y][cell.x] = true;
 
// finished ?
if (cell.x == size - 1 && cell.y == size - 1)
return cell.weight;
 
// one step right
if (cell.x + 1 < size)
next.push(Cell(cell.x + 1, cell.y, cell.weight + matrix[cell.y][cell.x + 1]));
// one step down
if (cell.y + 1 < size)
next.push(Cell(cell.x, cell.y + 1, cell.weight + matrix[cell.y + 1][cell.x]));
}
 
return -1; // failed
}
 
int main()
{
unsigned int size = 80;
//#define ORIGINAL
#ifndef ORIGINAL
std::cin >> size;
#endif
 
Matrix matrix(size);
for (auto& row : matrix)
{
row.resize(size);
for (auto& cell : row)
{
#ifdef ORIGINAL
// unfortunately, Project Euler used a CSV format which is a bit tricky to parse in C++
cell = 0;
// read until the number is complete or we run out of input
while (std::cin)
{
char c;
std::cin.get(c);
// number complete ?
if (c < '0' || c > '9')
break;
 
// add digit to current number
cell *= 10;
cell += c - '0';
}
#else
std::cin >> cell;
#endif
}
}
 
// go !
std::cout << search(matrix) << std::endl;
 
return 0;
}

This solution contains 18 empty lines, 24 comments and 8 preprocessor commands.

Interactive test

You can submit your own input to my program and it will be instantly processed at my server:

This live test is based on the Hackerrank problem.

Input data (separated by spaces or newlines):
Note: The first number is the width/height of the quadratic matrix, then follow all elements, separated by a space

This is equivalent to
echo "" | ./81

Output:

(please click 'Go !')

(this interactive test is still under development, computations will be aborted after one second)

Benchmark

The correct solution to the original Project Euler problem was found in less than 0.01 seconds on a Intel® Core™ i7-2600K CPU @ 3.40GHz.
(compiled for x86_64 / Linux, GCC flags: -O3 -march=native -fno-exceptions -fno-rtti -std=c++11 -DORIGINAL)

See here for a comparison of all solutions.

Note: interactive tests run on a weaker (=slower) computer. Some interactive tests are compiled without -DORIGINAL.

Changelog

March 12, 2017 submitted solution
May 4, 2017 added comments

Hackerrank

see https://www.hackerrank.com/contests/projecteuler/challenges/euler081

My code solves 7 out of 7 test cases (score: 100%)

Difficulty

Project Euler ranks this problem at 10% (out of 100%).

Hackerrank describes this problem as easy.

Note:
Hackerrank has strict execution time limits (typically 2 seconds for C++ code) and often a much wider input range than the original problem.
In my opinion, Hackerrank's modified problems are usually a lot harder to solve. As a rule thumb: brute-force is never an option.

Links

projecteuler.net/thread=81 - the best forum on the subject (note: you have to submit the correct solution first)

Code in various languages:

Python: www.mathblog.dk/project-euler-81-find-the-minimal-path-sum-from-the-top-left-to-the-bottom-right-by-moving-right-and-down/ (written by Kristian Edlund)
Java: github.com/nayuki/Project-Euler-solutions/blob/master/java/p081.java (written by Nayuki)
Scala: github.com/samskivert/euler-scala/blob/master/Euler081.scala (written by Michael Bayne)

Heatmap

green problems solve the original Project Euler problem and have a perfect score of 100% at Hackerrank, too.
yellow problems score less than 100% at Hackerrank (but still solve the original problem).
gray problems are already solved but I haven't published my solution yet.
blue problems are solved and there wasn't a Hackerrank version of it at the time I solved it or I didn't care about it because it differed too much.

Please click on a problem's number to open my solution to that problem:

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50
51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75
76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100
101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125
126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150
151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175
176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200
201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225
226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250
The 160 solved problems had an average difficulty of 21.8% at Project Euler and I scored 11,807 points (out of 13100) at Hackerrank's Project Euler+.
My username at Project Euler is stephanbrumme while it's stbrumme at Hackerrank.
more about me can be found on my homepage.
some names mentioned on this site may be trademarks of their respective owners.
thanks to the KaTeX team for their great typesetting library !