Why is my solution to the 3n+1 puzzle being rejected? [closed] - c++

Question

Closed. This question needs debugging details. It is not currently accepting answers.
Edit the question to include desired behavior, a specific problem or error, and the shortest code necessary to reproduce the problem. This will help others answer the question.
Closed 7 years ago.
Improve this question
I am trying to solve algorithmic problems on the UVa online judge but I am stuck on the 3n+1 problem. I see the right output every time, but the judge says it is a wrong answer. Why?
Also, how can I optimize this code so that it does not take a long time for 1000000?
#include <iostream>
#include <string>
#include <stdio.h>
using namespace std;
int main(){
int a;
int b;
int cyclelength(int i);
while (scanf("%d %d\n",&a,&b)==2){
int max = 0;
if (b < a){
for (int i = b; i < a; i++){
if (cyclelength(i) > max)
max = cyclelength(i);
}
}
else
{
for (int i = a; i < b; i++){
if (cyclelength(i) > max)
max = cyclelength(i);
}
}
cout << a << " " << b << " " << max << endl;
}
}
int cyclelength(int i){
if(i==1)
return 1;
if(!(i%2))
return cyclelength(i/2)+1;
else
return cyclelength(3*i+1)+1;
}

Answer 1

Your submission is incorrect because you aren't iterating over the range from a to b, inclusive.
The problem statement reads:
For each pair of input integers i and j you should output i, j, and the maximum cycle length for integers between and including i and j.
Your code loops over
for (int i = b; i < a; i++)
and
for (int i = a; i < b; i++)
thus omitting a in the former case and b in the latter.
You should change the loops to
for (int i = b; i <= a; i++)
and
for (int i = a; i <= b; i++)
You can speed up your code somewhat by memoizing the results of cyclelength. This lets you look up previously computed values instead of spending time on computing them again.
Here is a memoized version of cyclelength:
#include <map>
using namespace std;
map<long long, int> memo;
int cyclelength(long long i) {
if (i == 1) {
return 1;
}
if (memo.count(i) == 1) {
return memo[i];
}
if (i%2 == 0) {
return memo[i] = 1 + cyclelength(i/2);
} else {
return memo[i] = 1 + cyclelength(3*i + 1);
}
}
An unordered_map would be faster, but the UVA judge returned a compilation error for #include <unordered_map>.
Note that my version of cyclelength takes a long long argument. That lets me calculate cyclelength(999999), which eventually causes an integer overflow due to repeated 3n+1 operations if you limit yourself to 32-bit integer values. The problem statement promises that "no operation overflows a 32-bit integer" and you can get your submission accepted if you only use int values, but you have to use a larger integer type if you want to compute cyclelength for values in the range [1, 1000000).