<< problem 72 - Counting fractions | Digit factorial chains - problem 74 >> |

# Problem 73: Counting fractions in a range

(see projecteuler.net/problem=73)

Consider the fraction, dfrac{n}{d}, where n and d are positive integers. If n<d and HCF(n,d)=1, it is called a reduced proper fraction.

If we list the set of reduced proper fractions for d <= 8 in ascending order of size, we get:

dfrac{1}{8}, dfrac{1}{7}, dfrac{1}{6}, dfrac{1}{5}, dfrac{1}{4}, dfrac{2}{7}, dfrac{1}{3}, dfrac{3}{8}, dfrac{2}{5}, dfrac{3}{7}, dfrac{1}{2}, dfrac{4}{7}, dfrac{3}{5}, dfrac{5}{8}, dfrac{2}{3}, dfrac{5}{7}, dfrac{3}{4}, dfrac{4}{5}, dfrac{5}{6}, dfrac{6}{7}, dfrac{7}{8}

It can be seen that there are 3 fractions between 1/3 and 1/2.

How many fractions lie between 1/3 and 1/2 in the sorted set of reduced proper fractions for d <= 12,000 ?

# Algorithm

This is a problem where I had to try three different approaches until I found one that is fast enough to solve all Hackerrank test cases.

(all algorithms almost instantly solve the original problem, though).

All algorithms are related to the Farey sequence (https://en.wikipedia.org/wiki/Farey_sequence).

They ignore the numerator because it is actually not needed to solve the problem.

The most simple algorithm is based on recursion (look at my function `recursion`

).

Starting with `recursion(3, 2)`

(which means 1/3 and 1/2) the mediant m of 1/3 and 1/2 is found

and then the function calls itself with 1/3 and m and a second second with m a 1/2.

This continues until the denominator exceeds the limit 12000.

Each call returns the number of fractions which is `0`

when the denominator turns out to be too big or `1 + leftSide + rightSide`

else.

The second algorithm (see `iterative`

) computes the adjacent fraction of 1/3 (its "right neighbor").

Then, the denominator of the next fraction is nextD = maxD - \lfloor dfrac{maxD + prevD}{currentD} \rfloor.

We are done when nextD = toD.

The third and by far fastest algorithm computes the "rank" of fraction. Thereby rank(n, d) means: how many fractions are between 0 and dfrac{n}{d} ?

I found the idea/concept online: people.csail.mit.edu/mip/papers/farey/talk.pdf

Then the number of fractions between dfrac{1}{fromD} and dfrac{1}{toD} is rank(1, toD) - rank(1, fromD) - 1.

The algorithm is similar to a prime sieve.

# My code

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

The code contains `#ifdef`

s 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 <iostream>
#include <vector>
// maximum denominator

unsigned int maxD = 12000;
// algorithm I:
// count mediants between 1/fromD and 1/toD using recursion

unsigned int recursion(unsigned int fromD, unsigned int toD)
{
auto mediantD = fromD + toD;
// denominator too big ?
if (mediantD > maxD)
return 0;
// recursion
return 1 + recursion(fromD, mediantD) + recursion(mediantD, toD);
}
// algorithm II:
// iteratively enumerate all denominators

unsigned int iteration(unsigned int fromD, unsigned int toD)
{
// find denominator of closest mediant of "from"
// initial mediant
auto d = fromD + toD;
// is there a mediant closer to fromD ?
while (d + fromD <= maxD)
d += fromD;
// if prevD and d are denominators of adjacent fractions prevN/prevD and n/d
// then the next denominator is nextD = maxD - (maxD + prevD) % d
auto prevD = fromD;
unsigned int count = 0;
// until we reach the final denominator
while (d != toD)
{
// find next denominator
auto nextD = maxD - (maxD + prevD) % d;
// shift denominators, the current becomes the previous
prevD = d;
d = nextD;
count++;
}
return count;
}
// algorithm III:
// return numbers of irreducible fractions a/b < n/d where b is less than maxD

unsigned int rank(unsigned int n, unsigned int d)
{
// algorithm from "Computer Order Statistics in the Farey Sequence" by C. & M. Patrascu
// http://people.csail.mit.edu/mip/papers/farey/talk.pdf
std::vector<unsigned int> data(maxD + 1);
for (unsigned int i = 0; i < data.size(); i++)
data[i] = i * n / d; // n is always 1 but I wanted to keep the original algorithm
// remove all multiples of 2*i, 3*i, 4*i, ...
// similar to a prime sieve
for (unsigned int i = 1; i < data.size(); i++)
for (unsigned int j = 2*i; j < data.size(); j += i)
data[j] -= data[i];
// return sum of all elements
unsigned int sum = 0;
for (auto x : data)
sum += x;
return sum;
}
int main()
{
// denominators are abbreviated D
unsigned int toD = 2; // which means 1/2 (original problem)
//#define ORIGINAL

#ifndef ORIGINAL
std::cin >> toD >> maxD;
#endif
// the algorithm search from 1/fromD to 1/toD
auto fromD = toD + 1;
// algorithm 1
//std::cout << recursion(fromD, toD) << std::endl;
// algorithm 2
//std::cout << iteration(fromD, toD) << std::endl;
// algorithm 3
auto result = rank(1, toD) - rank(1, fromD) - 1;
std::cout << result << std::endl;
return 0;
}

This solution contains 17 empty lines, 30 comments and 4 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.

This is equivalent to`echo "2 8" | ./73`

Output:

*Note:* the original problem's input `2 12000`

__cannot__ be entered

because just copying results is a soft skill reserved for idiots.

*(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 15, 2017 submitted solution

May 3, 2017 added comments

# Hackerrank

see https://www.hackerrank.com/contests/projecteuler/challenges/euler073

My code solves **14** out of **14** test cases (score: **100%**)

# Difficulty

Project Euler ranks this problem at **15%** (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 rarely an option.

# Links

projecteuler.net/thread=73 - **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-73-sorted-reduced-proper-fractions/ (written by Kristian Edlund)

Java: github.com/nayuki/Project-Euler-solutions/blob/master/java/p073.java (written by Nayuki)

Go: github.com/frrad/project-euler/blob/master/golang/Problem073.go (written by Frederick Robinson)

Scala: github.com/samskivert/euler-scala/blob/master/Euler073.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.

red problems are solved but exceed the time limit of one minute.

*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 |

251 | 252 | 253 | 254 | 255 | 256 | 257 | 258 | 259 | 260 | 261 | 262 | 263 | 264 | 265 | 266 | 267 | 268 | 269 | 270 | 271 | 272 | 273 | 274 | 275 |

276 | 277 | 278 | 279 | 280 | 281 | 282 | 283 | 284 | 285 | 286 | 287 | 288 | 289 | 290 | 291 | 292 | 293 | 294 | 295 | 296 | 297 | 298 | 299 | 300 |

301 | 302 | 303 | 304 | 305 | 306 | 307 | 308 | 309 | 310 | 311 | 312 | 313 | 314 | 315 | 316 | 317 | 318 | 319 | 320 | 321 | 322 | 323 | 324 | 325 |

326 | 327 | 328 | 329 | 330 | 331 | 332 | 333 | 334 | 335 | 336 | 337 | 338 | 339 | 340 | 341 | 342 | 343 | 344 | 345 | 346 | 347 | 348 | 349 | 350 |

351 | 352 | 353 | 354 | 355 | 356 | 357 | 358 | 359 | 360 | 361 | 362 | 363 | 364 | 365 | 366 | 367 | 368 | 369 | 370 | 371 | 372 | 373 | 374 | 375 |

I scored 12,626 points (out of 14300 possible points, top rank was 20 out ouf ≈60000 in July 2017) at Hackerrank's Project Euler+.

Look at my progress and performance pages to get more details.

My username at Project Euler is

**stephanbrumme**while it's stbrumme at Hackerrank.

<< problem 72 - Counting fractions | Digit factorial chains - problem 74 >> |