Find the number of pairwise coprime triples of positive integers ( a , b , c ) with a < b < c such that
a ∣ b c − 3 1 , b ∣ c a − 3 1 and c ∣ a b − 3 1 .
Details and assumptions
The notation n ∣ m means that n is a divisor of m .
Clarification: ( a b − 3 1 ) , ( a c − 3 1 ) , and ( b c − 3 1 ) may be zero or negative.
This section requires Javascript.
You are seeing this because something didn't load right. We suggest you, (a) try
refreshing the page, (b) enabling javascript if it is disabled on your browser and,
finally, (c)
loading the
non-javascript version of this page
. We're sorry about the hassle.
Nicely done!
Nice solution. I had no idea why your first statement should hold, so for anyone else not seeing it at once: we just multiply the original equations and observe that if we don't pick 31 from at least two factors we get a multiple of a b c and finally we can clear 31 squared. You could have mentioned this, it's definitely not obvious.
Log in to reply
There is a faster way to see this. Note that a ∣ b c − 3 1 is equivalent to a ∣ a b + b c + c a − 3 1 , and same for the other two conditions. Then these three conditions are equivalent to a b c ∣ a b + b c + c a − 3 1 because a , b , c are pairwise coprime.
Log in to reply
Thanks, that's very neat! Using this approach I can derive the key formula in my answer in one line since b c ∣ a b + a c − 3 1 , so that a ( b + c ) − 3 1 = n b c . The rest of the solution is similar to this answer (i.e. consider possible values of n ) but I believe it's a bit simpler since we have less terms&factors to consider than here.
Log in to reply
@Marek Bernat – Yes, your way may be slightly shorter. But it is hard to say, unless all the cases are carefully written out.
First, we can see that a ∣ ( b c − 3 1 ) and a ∣ a ( b + c ) so a ∣ ( a b + b c + c a − 3 1 ) . It is similarly with b , c . In the other hand, because g cd ( a , b ) = g cd ( b , c ) = g cd ( c , a ) = 1 so a b c ∣ ( a b + b c + c a − 3 1 ) . (*)
We consider some cases: (1) If a = 1 then b ∣ c − 3 1 and c ∣ b − 3 1 . If b + c = 3 1 then two conditions satisfy and from a = 1 < b < c , we have 2 b < 3 1 and 2 ≤ b ≤ 1 5 , so there are 14 values of b and 14 values of c corresponding. And if b + c = 3 1 , we have 1 < b < c < 3 1 . If not, there are 3 cases: if b > 3 1 then c > b > b − 3 1 > 0 a contradiction because c|(b-31); if b = 3 1 then 3 1 ∣ c − 3 1 leads to g c d ( b , c ) > 1 a contradiction; if b < 3 1 < c then c > 3 1 − b , − c < b − 3 1 so ∣ c ∣ > ∣ b − 3 1 ∣ a contradiction. Hence, b + c < 3 1 , note that b c ∣ 3 1 − ( b + c ) . If b ≥ 5 then c ≥ 6 , so b c ≥ 3 0 , 3 1 − ( b + c ) ≤ 1 1 , a contradiction. So b = 3 or b = 4 . If b = 3 then 3 ∣ ( c − 3 1 ) and c ∣ 2 8 , it is easy to see that c = 4 , c = 7 . If b = 4 then 4 ∣ ( c − 3 1 ) and c ∣ 2 7 , leads to c = 2 7 , which is consider in the previous case. Hence, in this case, there are 16 triples ( a , b , c ) satisfy the condition. (2) If a = 2 then b ∣ 2 c − 3 1 and c ∣ 2 b − 3 1 , we also have b c ∣ ( 2 b + 2 c − 3 1 ) . With similar argument, we have b = 3 , c = 5 . (3) If a ≥ 3 , then a b c ≥ 3 b c > a b + b c + c a > ∣ a b + b c + c a − 3 1 ∣ ≥ ∣ a b c ∣ which a contradiction. Therefore, in total, we have 17 triples of ( a , b , c ) .
This was the best of the submitted solutions. A slight mistake at the end: " a b + b c + c a > ∣ a b + b c + c a − 3 1 ∣ ≥ ∣ a b c ∣ " is not well justified. It follows from the fact that for a ≥ 3 a b + b c + a c > 3 1 , but it is not mentioned. Without that, the first inequality might be false if a b + b c + a c < 1 5 , and the second might be false if a b + b c + c a − 3 1 = 0 . Otherwise, the solution is correct and very nice.
Observe that (bc - 31)(ab - 31)(ac-31) is a multiple of abc since a,b,c are coprime. Expanding, we get abc = 31 (impossible) or bc + ca + ab - 31 is a multiple of abc. Let bc + ca + ab - 31 = kabc
If k = 0, bc + ca + ab = 31 checking a = 1, we have (b,c) = bc + c + b = 31. Simple checking shows only (1,3,7) as the only solution. checking a = 2 it is clear 2,3,5 is the only solution.
If k >= 1, bc + ca + ab >= 31 + abc bc + ca + ab > abc 1/a + 1/b + 1/c > 1 The only possible solutions are (1,b,c), (2,3,4), (2,3,5). We have already checked and rejected (2,3,4), and accepted (2,3,5) as a solution. Considering bc + c + b = 31 + bc, we get (a,b,c) = (1,b,31-b) of which there are 14 solutions.
The last case, where k is negative:
ab + bc + ca - 31 < 0 ab + bc + ca < 31 Checking this condition: (1,2,3) (1,2,5) (1,2,7) (1,2,9), (2,3,4) are the only possibilities. We can easily check and reject all of them
Thus there are 17 possibilities: (2,3,5), (1,3,7), (1,3,4) and (1,b,31-b) for b from 2 to 15.
"Observe that (bc - 31)(ab - 31)(ac-31) is a multiple of abc since a,b,c are coprime." Coprimality is irrelevant here.
"Expanding, we get abc = 31 (impossible) or bc + ca + ab - 31 is a multiple of abc" Why?
k=0 case is not finished: what if a\geq 3?
(1,3,4) was never found, yet appeared in the final answer. Cheating or bad writing?
From the problem, it's already obvious that a|ab+bc+ac-31,b|ab+bc+ac-31, and c|ab+bc+ac-31. As a, b, and c are coprime, it's easy to prove that abc|ab+bc+ac-31 however, the only way to make ab+bc+ac-31 divisible by abc is to have it larger or equal abc, equal 0, or less or equal -abc, because abc obviously can't divide anything between 0 and abc or between 0 and -abc.
If ab+bc+ac-31 <= -abc ab+ac+bc+abc <=31 we know that the smallest combination of three coprime positive numbers are (1,2,3),(1,2,5),(1,3,5), and (2,3,5). Even among them, only (1,2,5) and (1,3,5) can result in ab+bc+ac+abc less or equal 31. Larger combination will have larger result as well. However, neither of those two combination works for the original problem. This option doesn't have suitable number.
If ab+bc+ac-31 = 0 again, we pick the number from the smallest coprime numbers combinations possible. The only combination fitting is (2,3,5) and it does fit the original problem after checking. If ab+bc+ac-31>=abc note that ab+bc+ac <3bc , so abc<3bc the only possible a are 1 and 2 if a=1 abc|ab+bc+ac-31 bc|b+c-31 again, b+c-31 may be 0, >=bc or <=-bc we can easily check that all combination of (1,b,31-b) works for the problem. If b+c-31>=bc while b+c-31<2c 2c>bc b<2 is a contradiction. If b+c-31<=-bc b+c+bc<=31 we can see by listing the possibilities that the largest b fulfilling this criteria is 4. As the range is already narrowed down, the remaining is trial and error. From here, we get working combinations of (1,3,4) and (1,3,7) if a=2 abc|ab+bc+ac-31 bc|2b+2c-31 again, we try all three possibilities 2b+2c-31=0 is already imposible. If 2b+2c-31>=bc while 2b+2c-31<4c b<4 the only b possible is 3, and the equation becomes abc|ab+bc+ac-31 6c |5c-25 it's impossible to have 5c-25>=6c or <=-6c as c is obviously >3. The only way to make this divisibility work is to have 5c-25=0 again, we get the combination of (2,3,5) If 2b+2c-31<=-2bc 2b+2c+2bc<=31 b+c+bc<=15 it's easy to see that any b larger than 2 and c larger than b will not work for this inequality.
Thus, the only answers are (1,3,4),(1,3,7),(1,b,31-b), and (2,3,5) while there are 14 solutions in the form of (1,b,31-b)
This one was a lot of fun, took me a few hours to get it right.
coprimetriples.c
#include <stdio.h>
#define LIMIT 10000
int gcd(int a, int b) {
int c;
while (a!=0) {
c=a;
a=b%a;
b=c;
}
return b;
}
int coprime3(int a, int b, int c) {
return gcd(a,b) == 1 && gcd(a,c) == 1 && gcd(b,c) == 1;
}
int main(int count, char** args) {
int a,b,c = 0;
int results = 0;
for (a=1; a<LIMIT; a++) {
for (b=1; b<LIMIT; b++) {
for (c=1; c<LIMIT; c++) {
if (a < b && b < c &&
((b*c - 31) % a == 0) &&
((c*a - 31) % b == 0) &&
((a*b - 31) % c == 0) &&
coprime3(a,b,c)==1) {
results++;
printf("(%d,%d,%d) found:%d\n\n",
a,
b,
c,
results);
}
}
}
}
return 0;
}
Output
(1,2,29) found:1
(1,3,4) found:2
(1,3,7) found:3
(1,3,28) found:4
(1,4,27) found:5
(1,5,26) found:6
(1,6,25) found:7
(1,7,24) found:8
(1,8,23) found:9
(1,9,22) found:10
(1,10,21) found:11
(1,11,20) found:12
(1,12,19) found:13
(1,13,18) found:14
(1,14,17) found:15
(1,15,16) found:16
(2,3,5) found:17
Are you able to justify why "LIMIT 10000" works? That would close up a gap.
Log in to reply
hope you guys aren't mad on my problem solving style not being purely mathematical. I've been able to connect with young users who are really good at math and they seem very interested in learning how to code, hopefully Brilliant.org problems can serve as small projects for them. This is why I keep posting them in different languages (python, javascript, C, Java)
I'll see if I can refresh my dusty old Haskell, and show some solutions in a more functional form.
Log in to reply
We are grateful that you are taking an interest in teaching youngsters how to code, especially since it will become important for them when they start working. We are looking at providing various avenues for our members to share their knowledge and interests. A possible suggestion is for you to post interesting computing problems that you see in the discussion section, and see how others would approach it.
for that I'd have to do some math. Just a hunch, when I ran it, I actually only did it for 1,000.
The algorithm changed quite a bit before this, as I had a wrong implementation of the corprime3() check, and I was having over 999 resuts.
So to answer, nope, not able to justify it any more than a hunch and using clues available to me, like the 2 other attempts, and the awesome answering constraints from 0 to 999.
Perhaps (as a suggestion to hinder lazy hackers like me) I'd make the range of possible answers larger than [0,999]
Log in to reply
I am pointing out the limitation of your program, namely that you had to assume that c < 1 0 0 0 0 (or 1000). This is not immediately obvious, and your solution is otherwise incomplete. If you were asked to find the number of prime numbers, can you say that you searched up to 1 0 1 0 0 and conclude that there are much less than 1 0 1 0 0 prime numbers?
There are, of course, many ways to hack around the problem, esp making use of the assumption that the answer is a small integer.
Go to projecteuler.net if you want to do computer science. I personally don't find programming approaches to mathematical problems helpful to be posted into the solutions.
Log in to reply
been there, done that. don't read the solution if you don't care.
Instead of "if (a < b && b < c &&" you could have replaced "for (b = 1" with "for (b = a+1" etc. . Good on you for using standard-compliant language
Let us write b k = a c − 3 1 and c l = a b − 3 1 . Multiplying the first equation by l we get b k l = a ( a b − 3 1 ) − 3 1 l so that b ( a 2 − k l ) = 3 1 ( a + l ) .
3 1 can't divide b because that would also imply that 3 1 divides a c in contradiction with coprimalilty assumption. Therefore 3 1 divides a 2 − k l and we can write a 2 − k l = 3 1 n to obtain b n = ( a + l ) .
Now, let us return to the very first two equations. Subtracting them, we get b k − c l = a ( c − b ) and adding 0 = b l − b l we get b ( k − l ) − l ( c − b ) = a ( c − b ) b ( k − l ) = ( a + l ) ( c − b ) so that, using b n = a + l we derived previously, k − l = n ( c − b ) . Using the very first equations again to express k and l and multiplying the last equation by b c we get c ( a c − 3 1 ) − b ( a b − 3 1 ) = n b c ( c − b ) . Finally, dividing by c − b ) we obtain the key formula a ( b + c ) − 3 1 = n b c . Now, because a < b < c we get that a ( b + c ) / b c < 2 . Suppose now that b c > 3 1 . Then the only possible values of n are 0 and 1 . For n = 0 we have a ( b + c ) = 3 1 which is non-sense. When n = 1 we have a ∣ 3 1 + b c which, combined with the fact that a ∣ b c − 3 1 implies that a = 1 .
So, we are left with two possibilities, a = 1 or b c < 3 1 . Since this answer is already too long, I'll be brief, but with bit more work we can show that for a = 1 2 ≤ b ≤ 1 5 , b + c = 3 1 are all solutions (e.g., 1 , 4 , 2 7 is one of them). All of these have n = 1 . For n = − 1 we get ( 1 , 3 , 7 ) and ( 2 , 3 , 5 ) and for n = − 2 we have ( 1 , 3 , 4 ). No other n can work (proof left for the reader), so this gives us all of the 1 7 solutions.
A minor correction: a ( b + c ) = 3 1 is not non-sense, but implies a = 1 , b + c = 3 1 , which are actually the 1 4 answers mentioned lower in the post.
The ending part is very sketchy with minor mistakes. Sorry about that. Nevertheless, I believe that 90% of the work lies in obtaining the key formula. The rest is pretty straightforward or, for a low-brow approach, can be actually checked by hand, since there are not that many cases left. Cheers :)
a,b,c all divide ab+ac+bc−31, so since they are pairwise coprime, abc∣(ab+ac+bc−31) Note that the above is equivalent to the given conditions. Consider 3 cases, where ab+ac+bc−31 is =0,<0,>0.
Case 1: ab+ac+bc−31=0, then 31=ab+ac+bc≥a(a+1)+a(a+2)+(a+1)(a+2)=3a2+6a+2, so a≤2.
If a=1, then c>b>1, and (b+1)(c+1)=32. We have 3≤b+1<32−−√ so b+1=4,c+1=8 and (a,b,c)=(1,3,7).
If a=2, then c>b>2 and (b+2)(c+2)=35. We have 5≤b+2<35−−√ so b+2=5,c+2=7 and (a,b,c)=(2,3,5).
Case 2: ab+ac+bc−31<0, then abc≤|ab+ac+bc−31|=31−ab−ac−bc.
If a≥2, then b≥3,c≥4, so 31≥abc+ab+ac+bc≥2(3)(4)+2(3)+2(4)+3(4)>31, a contradiction. Thus a=1, so c>b>1 and 2bc+b+c≤31.
If b≥3, then c≥4, so 31≥2bc+b+c≥2(3)(4)+(3)+(4)=31, so equality holds and we have (a,b,c)=(1,3,4). Checking, this works.
Otherwise b=2, so 5c≤29, giving c=3,4,5. However we also have c∣ab−31=−29, so this gives no solution.
Case 3: ab+ac+bc−31>0, then abc≤ab+ac+bc−31.
If a≥3, then abc≥3bc≥ab+ac+bc>ab+ac+bc−31≥abc, a contradiction.
Thus a=1,2. If a=2, then c>b>2 and 2bc≤2b+2c+bc−31, so bc+31≤2b+2c, so (b−2)(c−2)+27≤0, a contradiction.
Thus a=1, so bc≤b+c+bc−31, so b+c≥31. Also bc∣(b+c+bc−31), so bc∣(b+c−31). If b+c=31, we have (a,b,c)=(1,2,29),(1,3,28),…,(1,15,16). Otherwise bc≤b+c−31, so (b−1)(c−1)+30≤0, a contradiction.
To conclude, the solutions are (1,3,7),(2,3,5),(1,3,4),(1,2,29),(1,3,28),…,(1,15,16) i.e. number of solutions are 3+14=17
a,b,c all divide ab+ac+bc−31, so since they are pairwise coprime, abc∣(ab+ac+bc−31)
Note that the above is equivalent to the given conditions. Consider 3 cases, where ab+ac+bc−31 is =0,<0,>0.
Case 1: ab+ac+bc−31=0, then 31=ab+ac+bc≥a(a+1)+a(a+2)+(a+1)(a+2)=3a2+6a+2, so a≤2.
If a=1, then c>b>1, and (b+1)(c+1)=32. We have 3≤b+1<32−−√ so b+1=4,c+1=8 and (a,b,c)=(1,3,7).
If a=2, then c>b>2 and (b+2)(c+2)=35. We have 5≤b+2<35−−√ so b+2=5,c+2=7 and (a,b,c)=(2,3,5).
Case 2: ab+ac+bc−31<0, then abc≤|ab+ac+bc−31|=31−ab−ac−bc.
If a≥2, then b≥3,c≥4, so 31≥abc+ab+ac+bc≥2(3)(4)+2(3)+2(4)+3(4)>31, a contradiction. Thus a=1, so c>b>1 and 2bc+b+c≤31.
If b≥3, then c≥4, so 31≥2bc+b+c≥2(3)(4)+(3)+(4)=31, so equality holds and we have (a,b,c)=(1,3,4). Checking, this works.
Otherwise b=2, so 5c≤29, giving c=3,4,5. However we also have c∣ab−31=−29, so this gives no solution.
Case 3: ab+ac+bc−31>0, then abc≤ab+ac+bc−31.
If a≥3, then abc≥3bc≥ab+ac+bc>ab+ac+bc−31≥abc, a contradiction.
Thus a=1,2. If a=2, then c>b>2 and 2bc≤2b+2c+bc−31, so bc+31≤2b+2c, so (b−2)(c−2)+27≤0, a contradiction.
Thus a=1, so bc≤b+c+bc−31, so b+c≥31. Also bc∣(b+c+bc−31), so bc∣(b+c−31). If b+c=31, we have (a,b,c)=(1,2,29),(1,3,28),…,(1,15,16). Otherwise bc≤b+c−31, so (b−1)(c−1)+30≤0, a contradiction.
To conclude, the solutions are (1,3,7),(2,3,5),(1,3,4),(1,2,29),(1,3,28),…,(1,15,16)
hence solutions is 17
a,b,c all divide ab+ac+bc−31 , so since they are pairwise coprime, abc∣(ab+ac+bc−31)
Note that the question can be split into 3 cases where ab+ac+bc−31 is =0,<0,>0.
Case 1: ab+ac+bc−31=0, then 31=ab+ac+bc≥a(a+1)+a(a+2)+(a+1)(a+2)=3a^2+6a+2, so a≤2. If a=1, then c>b>1, and (b+1)(c+1)=32. We have 3≤b+1< √32 so b+1=4,c+1=8 and (a,b,c)=(1,3,7).
If a=2, then c>b>2 and (b+2)(c+2)=35. We have 5≤b+2<√35 so b+2=5,c+2=7 and (a,b,c)=(2,3,5).
Case 2: ab+ac+bc−31<0, then abc≤|ab+ac+bc−31|=31−ab−ac−bc. If a≥2, then b≥3,c≥4, so 31≥abc+ab+ac+bc≥2(3)(4)+2(3)+2(4)+3(4)>31, a contradiction.
Thus a=1, so c>b>1and 2bc+b+c≤31.
If b≥3, then c≥4, so 31≥2bc+b+c≥2(3)(4)+(3)+(4)=31, so equality holds and we have (a,b,c)=(1,3,4).
Otherwise b=2, so 5c≤29, giving c=3,4,5. However we also have c∣ab−31=−29, so this gives no solution.
Case 3: ab+ac+bc−31>0 , then abc≤ab+ac+bc−31. If a≥3, then abc≥3bc≥ab+ac+bc>ab+ac+bc−31≥abc, a contradiction. Thus a=1,2.
If a=2, then c>b>2 and 2bc≤2b+2c+bc−31, so bc+31≤2b+2c, so (b−2)(c−2)+27≤0, a contradiction.
Thus a=1 , so bc≤b+c+bc−31 , so b+c≥31. Also bc∣(b+c+bc−31), so bc∣(b+c−31). If b+c=31, we have (a,b,c)=(1,2,29),(1,3,28),…,(1,15,16) .
Otherwise bc≤b+c−31 , so (b−1)(c−1)+30≤0, a contradiction.
To conclude, the solutions are (1,3,7),(2,3,5),(1,3,4),(1,2,29),(1,3,28),…,(1,15,16). There are a total of 17 solutions.
a|bc−31 means a divide bc-31 , also a divide ac and ab. therefor a divide ab+bc+ca-31 i.e. a|(ab+bc+ca-31) also b|(ab+bc+ca-31) and c|(ab+bc+ca-31)
now since a,b,c are coprime positive integers
we have (a b c)|(ab+bc+ca-31)
therefor abc<=|(ab+bc+ca-31)| where || is mode function
case 1) let ab+ac+bc−31>0 then abc≤ab+ac+bc−31
If a≥3, then abc≥3bc≥ab+ac+bc>ab+ac+bc−31≥abc, a contradiction.
Thus a=1,2. If a=2, then c>b>2 and 2bc≤2b+2c+bc−31, so bc+31≤2b+2c, so (b−2)(c−2)+27≤0, a contradiction.
Thus a=1, so bc≤b+c+bc−31, so b+c≥31. Also bc∣(b+c+bc−31), so bc∣(b+c−31).
If b+c=31, we have (a,b,c)=(1,2,29),(1,3,28),(1,4,27),(1,5,26),(1,6,25),(1,7,24),(1,8,23),(1,9,22),(1,10,21),(1,11,20),(1,12,19),(1,13,18),(1,14,17),(1,15,16)
Otherwise bc≤b+c−31, so (b−1)(c−1)+30≤0, a contradiction.
Case 2) ab+ac+bc−31=0, then 31=ab+ac+bc≥a(a+1)+a(a+2)+(a+1)(a+2)=3a2+6a+2, so a≤2.
If a=1, then c>b>1, and (b+1)(c+1)=32. We have 3≤b+1<√32 so b+1=4,c+1=8 and (a,b,c)=(1,3,7).
If a=2, then c>b>2 and (b+2)(c+2)=35. We have 5≤b+2<√35 so b+2=5,c+2=7 and (a,b,c)=(2,3,5).
Case 3) ab+ac+bc−31<0, then abc≤|ab+ac+bc−31|=31−ab−ac−bc.
If a≥2, then b≥3,c≥4, so 31≥abc+ab+ac+bc≥2(3)(4)+2(3)+2(4)+3(4)>31, a contradiction. Thus a=1, so c>b>1 and 2bc+b+c≤31.
If b≥3, then c≥4, so 31≥2bc+b+c≥2(3)(4)+(3)+(4)=31, so equality holds and we have (a,b,c)=(1,3,4). Checking, this works.
Otherwise b=2, so 5c≤29, giving c=3,4,5. However we also have c∣ab−31=−29, so this gives no solution.
the solutions are (1,3,7),(2,3,5),(1,3,4),(1,2,29),(1,3,28),(1,4,27),(1,5,26),(1,6,25),(1,7,24),(1,8,23),(1,9,22),(1,10,21),(1,11,20),(1,12,19),(1,13,18),(1,14,17),(1,15,16)
Problem Loading...
Note Loading...
Set Loading...
The given conditions are equivalent to a b c ∣ a b + b c + c a − 3 1 , since a , b , c are pairwise coprime.
→ a b + b c + c a = 3 1 + ( a b c ) k , k is a integer.
Case 1: k ≤ 0
→ 3 a 2 < a b + b c + c a ≤ 3 1 → a ≤ 3
Case 1.1: a = 3
→ 3 ( b + c ) + b c ≤ 3 1 → ( b + 3 ) 2 < ( b + 3 ) ( c + 3 ) ≤ 4 0 → b ≤ 3 contradicts a < b
Case 1.2: a = 2
→ 2 ( b + c ) + b c ≤ 3 1 → ( b + 2 ) 2 < ( b + 2 ) ( c + 2 ) < 3 5 → b ≤ 3 → b = 3
→ 5 c ≤ 2 5 and c ∣ ( 2 ) ( 3 ) − 3 1 → c = 4 or c = 5 and c ∣ 2 5 → c = 5
→ ( a , b , c ) = ( 2 , 3 , 5 )
Case 1.3: a = 1
→ ( b + c ) + b c ≤ 3 1 → ( b + 1 ) 2 < ( b + 1 ) ( c + 1 ) < 3 2 → b ≤ 4 → b = 2 , 3 or 4
Case 1.3.1: b = 2
→ 3 c ≤ 2 9 and c ∣ ( 1 ) ( 2 ) − 3 1 → 3 ≤ c ≤ 9 and c ∣ 2 9 → no solution. Case 1.3.2: b = 3
→ 4 c ≤ 2 8 and c ∣ ( 1 ) ( 3 ) − 3 1 → 3 ≤ c ≤ 7 and c ∣ 2 8 → c = 4 or c = 7
By checking, both sets of solutions of work, so we have ( a , b , c ) = ( 1 , 3 , 4 ) and ( 1 , 3 , 7 )
Case 1.3.3: b = 4
→ 5 c ≤ 2 7 and c ∣ ( 1 ) ( 4 ) − 3 1 → c = 5 and c ∣ 2 7 → no solution.
Case 2: k ≥ 1
Dividing both sides by a b c , we have
a 1 + b 1 + c 1 = a b c 3 1 + k > 1
1 < a 1 + b 1 + c 1 < a 3 → a = 1 or a = 2
Case 2.1: a = 2
→ 2 1 < b 1 + c 1 < b 2 → b = 3
→ c ∣ ( 2 ) ( 3 ) − 3 1 → c = 5 or c = 2 5
( 2 , 3 , 5 ) is already in our list of solution, while ( 2 , 3 , 2 5 ) does not satisfy 3 ∣ ( 2 ) ( 2 5 ) − 3 1 → no solution.
Case 2.2: a = 1
→ b 1 + c 1 − b c 3 1 = k − 1 ≥ 0
Case 2.2.1: k ≥ 1
1 < b 1 + c 1 < b 2 → no solution.
Case 2.2.2: k − 1 = 0
→ b + c = 3 1 . We have 14 solutions: a , b , c ) = ( 1 , 2 , 2 9 ) , ( 1 , 3 , 2 8 ) , . . . , ( 1 , 1 5 , 1 6 ) .
Hence, the answer is 1 + 2 + 1 4 = 1 7 .