Find all pairs of natural numbers $ (a, b)$ such that $ 7^a - 3^b$ divides $ a^4 + b^2$. Author: Stephan Wagner, Austria
Problem
Source: ISL 2007, N1, VAIMO 2008, P4
Tags: Divisibility, IMO Shortlist, number theory
22.04.2008 14:51
The left hand side is even, so $ a^4+b^2$ is even, and $ a,b$ has the same parity. However, if $ a,b$ are odd, we have that $ 4$ divides $ 7^a-3^b$, but $ 4$ doesn't divide $ a^4+b^2$. So $ a,b$ are even, and let's say $ a=2a_1,b=2b_1$. You'll get $ (7^{a_1}-3^{b_1})(7^{a_1}+3^{b_1})$ divides $ 4(4a_1^4+b_1^2)$ If $ b_1$ is odd, then $ 8$ divides $ (7^{a_1}-3^{b_1})(7^{a_1}+3^{b_1})$ but not $ 4(4a_1^4+b_1^2)$, so $ b_1$ is even. Say $ b_1=2c_1$. You'll get $ (7^{a_1}-3^{2c_1})(7^{a_1}+3^{2c_1})$ divides $ 16(a_1^4+c_1^2)$ So $ |(7^{a_1}-3^{2c_1})(7^{a_1}+3^{2c_1})|\leq |16(a_1^4+c_1^2)|$ But $ |(7^{a_1}-3^{2c_1})\geq 2$ (because both terms are even), and so $ 2(7^{a_1}+3^{2c_1})\leq 16(a_1^4+c_1^2)$. But for $ a_1=1,2,3$ we easily get $ c_1=1,2$, and for $ a_1\geq 4$, we have that $ 2(7^{a_1})>16a_1^4$. Also, $ 2(3^{2c_1})>16c_1^2$ for all $ c_1\in \mathbb{N}$. So it is enough to check $ (a_1,c_1)=(1,1),(2,1),(3,1),(1,2),(2,2),(3,2)$. We'll get $ (a_1,c_1)=1,1$, or $ (a,b)=(2,4)$ as the only solution. By the way, Dida Drogbier, are you from Chelsea or AC Milan?
31.03.2011 04:59
dig, dig, dig alternative solution anybody?
31.03.2011 05:11
This is my alternative solution. The left hand side is even, so $ a^4+b^2$ is even, and $ a,b$ has the same parity. However, if $ a,b$ are odd, we have that $ 4$ divides $ 7^a-3^b$, but $ 4$ doesn't divide $ a^4+b^2$. So $ a,b$ are even, and let's say $ a=2x,b=2y$. You'll get $ (7^{x}-3^{y})(7^{x}+3^{y})$ divides $ 4(4x^4+y^2)$ If $ y$ is odd, then $ 8$ divides $ (7^{x}-3^{y})(7^{x}+3^{y})$ but not $ 4(4x^4+y^2)$, so $ y$ is even. Say $ y=2z$. You'll get $ (7^{x}-3^{2z})(7^{x}+3^{2z})$ divides $ 16(x^4+z^2)$ So $ |(7^{x}-3^{2z})(7^{x}+3^{2z})|\leq |16(x^4+z^2)|$ But $ |(7^{x}-3^{2z})\geq 2$ (because both terms are even), and so $ 2(7^{x}+3^{2z})\leq 16(x^4+z^2)$. But for $ x=1,2,3$ we easily get $ z=1,2$, and for $ x\geq 4$, we have that $ 2(7^{x})>16x^4$. Also, $ 2(3^{2z})>16z^2$ for all $ z\in \mathbb{N}$. So it is enough to check $ (x,z)=(1,1),(2,1),(3,1),(1,2),(2,2),(3,2)$. We'll get $ (x,z)=1,1$, or $ (a,b)=(2,4)$ as the only solution.
31.03.2011 05:22
tom: you shouldn't give solution to shortlist problem before IMO. you should talk to me (phone number below)
23.05.2016 20:49
I believe that proving $b \equiv 0 \pmod{4}$ is unnecessary, the solution in that case is a bit more messy, but similar. The motivation for this solution is using size issues in divisibility, similar to the above posts.
15.01.2017 15:26
AMN300 wrote: $4d^2-3^d >0 \implies d=1$ why doesn't $4d^2-3^d >0$ imply $d=1,2,or 3 $ ?
03.04.2018 22:41
01.10.2019 22:47
18.03.2020 08:59
It feels nice to get back to NT. Here's my solution that uses the same idea of bounding. Proof: Notice that clearly $7^a-3^b \equiv 0 \pmod 2$. Since $7,3$ have the same parity so this forces that either $(a,b)$ are both odd or are both even. We'll firstly work with the case when $(a,b)$ are both odd. Let $(a,b)=(2k+1,2l+1)$. Notice that $$7^a-3^b \equiv 7^{2k+1}-3^{2l+1} \equiv 0 \pmod 4$$But $a^4+b^2 \equiv (2k+1)^4 +(2l+1)^2 \equiv 2 \pmod 4$ a contradiction. Now let $(a,b)=(2k,2l)$. Clearly we must have that $(7^k+3^l)(7^k-3^l) \mid 16k^4+4l^2$. But since $$2(7^k+3^l) \leq |(7^k+3^l)(7^k-3^l)| \leq 16k^4+4l^2$$so we can estabilish that $7^k+3^l \leq 8k^4+2l^2$. This is equivalent to $3^l-2l^2 \leq 8k^4-7^k$. But notice that from induction $\forall l \geq 1$ we have that $3^l>2l^2$ and hence we have $0 \leq 3^l-2l^2 \leq 8k^4-7^k$ . But notice that $7^k \geq 8k^4$ $\forall k \geq 4$ hence we have to check for $k \leq 3$. Now notice that if $k=3$ then we have that $343+3^l \leq 648+2l^2 \implies 3^l-2l^2 \leq 305$. But notice that $3^l -2l^2 > 305$ for all $l\geq 6$ hence we have to check till $l=6$. But clearly none of the solution sets work. $\square$. Now if $k=2$ then we have $3^l-2l^2 \leq 128-49=79$ . But again notice that we that $3^l-2l^2 > 78$ for $l\geq 5$. Hence we check all the way uptil $5$. But clearly none of the solution sets work again $\square$. Now if $k=1$ then notice that we have $3^l-2l^2 \leq 1$ but notice that this is true only when $l \leq 2$ . So we check till $l=2$ and we have a solution set that woks which is $(k,l)=(1,2)$ or that $\boxed{(a,b)=(2,4)}$ $\blacksquare$.
16.06.2020 11:45
tom_damrong wrote: This is my alternative solution. Ummm... You gave the exact solution (quite literally) as the one two posts above. How is yours an "alternative" solution?
29.08.2020 07:10
Surprisingly, if you push hard enough this problem just falls. We claim only $(2,4)$ works, and we can easily verify that it does. Note that $2\mid 7^a-3^b,$ so $2\mid a^4+b^2,$ implying $a\equiv b\pmod{2}.$ This implies that $7^a\equiv 3^b\pmod {4},$ so $4\mid a^4+b^2,$ implying that $2\mid a$ and $2\mid b.$ Say that $a=2x$ and $b=2k.$ Then note $8\mid 49^{x}-9^k,$ so we must have $8\mid 4(4x^2+k^2),$ implying that $2\mid k.$ Let $k=2y,$ then we want to solve $(7^x-9^y)(7^x+9^y)\mid 16(x^4+y^2).$ Now we just bound. Note that $2(7^x+9^y)\leq |(7^x-9^y)(7^x+9^y)|\leq 16(x^4+y^2),$ implying $7^x+9^y\leq 8(x^4+y^2).$ Note that for all $x\geq 4,$ we have $7^x> 8x^4,$ and for all positive integers $y,$ $9^y>8y^2.$ So the only possible solutions are natural numbers $x\leq 3.$ In the case of $x=1,$ note that we must have \[(7-9^y)(7+9^y)\mid 16(1+y^2),\]or $81^y-49\leq 16y^2+16.$ Note this fails for any $y\geq 2,$ and that $y=1$ yields equality, so $(1,1)$ is the only solution for $x=1.$ In the case of $x=2,$ note that we must have \[(49-9^y)(49+9^y)\mid 16(16+y^2).\]We can easily verify that $y=1$ fails. For $y\geq 2,$ we want to ensure $81^y-7^4\leq 16y^2+16\cdot 16,$ which fails for all $y\geq 2.$ In the case of $x=3,$ note that we must have \[(343-9^y)(343+9^y)\mid 16(81+y^2).\]We can easily verify that $y=1$ and $y=2$ fail by size reasons, noting $343-9^y<16$ and $343+9^y>81+y^2$ for $y=1,2.$ For $y\geq 3,$ we want to ensure that $81^y-7^6\leq 16y^2+16\cdot 81,$ which fails for all $y\geq 3.$ So the only solution in $(x,y)$ is $(1,1),$ corresponding to $(2,4).$
21.03.2021 17:54
If $a$ and $b$ don't have the same parity, then $7^a-3^b$ is even and $a^4+b^2$ is odd, absurd. Otherwise, $4\mid 7^a-3^b\mid a^4+b^2$ so $a$ and $b$ are both even. Let $a=2m,b=2n$. Then $(7^m-3^n)(7^m+3^n)\mid 16m^4+4n^2$. Now, if $m$ and $n$ are the same parity, the LHS is divisible by $8$ and otherwise it is still divisible by $8$, so either way $2\mid n$ and thus we can let $n=2k$ so $(7^m-9^k)(7^m+9^k)\mid 16(m^4+k^2)$. Now, we use the crowbar known as "size": it is clear that the LHS's absolute value is at least $2(7^m+9^k)$, so at least one of $8k^2\ge 9^k$ and $8m^4\ge 7^m$ occurs. The former can never occur and the latter can only occur for $m\le 3$. Now, at $m=3$ we get $(343-9^k)(343+9^k)\mid 16(81+k^2)$. But the value of $k$ that minimizes $|343-9^k|$ is $262$, which is bad because $16\cdot81/262<343$. If we have $m=2$, we get $(49-9^k)(49+9^k)\mid 16(16+k^2)$. Similarly to before, note $|49-9^k|$ is minimized at $32$, so since $16\cdot16/32=8<49$, this is bad. So finally $m=1$ and $49-81^k=(7-9^k)(7+9^k)\mid 16(1+k^2)$. Equivalently, $0<81^k-49\mid 16(1+k^2)$. Equality holds exactly at $k=1$. Moreover, $\log(81)\cdot 81>32$ and $\log(81)^2\cdot81>32$, so by differentiation the result follows: the divisibility holds only when $(a,b)=(2,4)$, which can easily be checked to be true.
11.04.2021 19:25
Quite straightforward. Note that $2\mid 7^a-3^b\implies 2\mid a^4+b^2$. Hence $a$ and $b$ have the same parity. If $a$ and $b$ are both odd, then $7^a\equiv (-1)^a\equiv -1\equiv (-1)^b\equiv 3^b$ (mod $4$) $\implies 4\mid 7^a-3^b$, which is contradictory to the congruence $7^a+3^b\equiv 2$ (mod $4$). Thus $a$ and $b$ are both even,let $a=2m,b=2n$. The divisibility condition becomes $7^{2m}-3^{2n}\mid 16m^4+4n^2\implies (7^m+3^n)(7^m-3^n)\mid 16m^4+4n^2$. Since $2\mid 7^m-3^n$,we have $7^m+3^n\le 8m^4+2n^2$. Then we can check case by case and get the unique solution $(a,b)=(2,4)$.
01.05.2021 00:47
This took me way too long (expected to be cuter ). This solution is essentially the same as m.essein's, but posted for fun. We have by modulo $4$ reasons, that $a$ and $b$ are even. Taking also modulo $8$, we get that $a=2x$ and $b=4y$. So essentially, we reduce our problem to the following: $$(7^x-9^y)(7^x+9^y)\mid 16(x^4+y^2).$$Thus, $7^x+9^y\leq 8(x^4+y^2)$ as $2\mid 7^x-9^y\implies 2\leq \vert 7^x-9^y \vert$ . Note that $7^x>8x^4$ for all $x>3$ and $9^y> 8y^2$ for all $y>0$. Therefore we must only consider values $x=1,2,3$. Note that if $x\leq 3$, then $y=1$ or $2$. Thus, we only need to consider pairs $(x,y)=(1,1),(1,2),(2,1),(2,2),(3,1),(3,2)$. Only solution is $(x,y)=(1,1)$. Thus, we obtain our only solution pair $\boxed{(a,b)=(2,4)}$.
27.06.2021 15:03
Notice that $7^a-3^b$ must be even, as both terms are odd. This implies that $a^4+b^2$ is even, so $a$ and $b$ have the same parity. However, if $a$ and $b$ were both odd, then we would have $$LHS\equiv (-1)^a-(-1)^b \equiv (-1)-(-1)\equiv 0 \pmod 4,$$as $a$ and $b$ are odd, and $$RHS \equiv 1+1\equiv 2 \pmod 4$$as an odd perfect square must be $1\pmod 4$, which is a contradiction. Therefore $a$ and $b$ are both even. This means that if we let $a=2c$ and $b=2d$, then $$49^c-9^d | 16c^4+4d^2.$$Then, since $49^c-9^d=(7^c+3^d)(7^c-3^d)$, we must also have $$7^c+3^d | 16c^4+4d^2,$$in particular $7^c+3^d \le 16c^4+4d^2$. We must have $(7^c-16c^4)+(3^d-4d^2)<0$. The minimum of $7^c-16c^4$ is $-1695$ at $c=4$, and $3^d-4d^2>1695$ for $d\ge 7$, so $d<7$. Similarly, the minimum of $3^d-4d^2$ is $-9$ at $d=3$, and $7^c-16c^4>9$ for $c\ge 5$, so $c<5$. Checking the remaining cases, we get that $c=1, d=2$ is the only one that works, which corresponds to $(a,b)=\boxed{(2,4)}$.
04.07.2021 16:55
Solved with L567 using the Catalan Conjecture (well technically not but yeah) - posting for storage. Notice that $2 \mid 7^a-3^b \mid a^4+b^2$ meaning that $a \equiv b \pmod{2}$ and therefore $4 \mid 7^a-3^b \mid a^4+b^2$ and consequently $2 \mid a,b$, writing $a = 2x,b=2y$, we have $$(7^x-3^y)(7^x+3^y)=7^{2x}-3^{2y} \mid 4(4x^4+y^2)$$Notice that $\mid 7^x-3^y \mid \geq 2$ (by the Catalan Conjecture) therefore $$7^x+3^y \leq 8x^4+2y^2$$Notice that $7^x > 8x^4$ for all $x \geq 4$ and $3^y > 2y^2$ for all $y \geq 3$. Checking all the cases for $x \in \{1,2,3\}$ and $y \in \{1,2\}$ gives us the only pair $(a,b) = (2,4)$.
17.07.2021 06:04
Solution from Twitch Solves ISL: The only answer is $(a,b) = (2,4)$, which works as $7^2 - 3^4 = -32$ divides $2^4 + 4^2 = 32$. To show this is the only solution, we note the following: The left-hand side is always even, which implies $a$ and $b$ have the same parity. This in turn implies the left-hand side is $0 \bmod 4$. Hence $a^4+b^2 \equiv 0 \pmod 4$ which means $a$ and $b$ are even. This means that $7^a -3^b \equiv 0 \pmod 8$, which further means $b \equiv 0 \pmod 4$. This means we may let $a = 2x$ and $b = 2y$ with $y$ even to obtain \[ \left( 7^x - 3^y \right)\left( 7^x + 3^y \right) \mid (2x)^4 + (2y)^2. \]To reduce to finite casework we contend: Claim: $2 \cdot 3^a > 16a^4 + 4a^2$ once $a \ge 10$. Proof. $3^{10} = 243^2 = 2 \cdot59049 = 177147 > 160400$. $\blacksquare$ Hence we are done if $\max(x,y) \ge 10$. Next Claim: $7^x > 16x^4 + 4 \cdot 10^2$ for $x \ge 6$ Proof. $7^6 = 343^2 > 100000 > 16 \cdot 6^4 + 400$. $\blacksquare$ So we may assume $x \le 5$ as well. This gives us the following cases. In all situations, note that $(2 \cdot 5)^4 + (2 \cdot 10)^2 = 10400$ and consequently if $(7^x-3^y)(7^x+3^y) > 10400$ there is nothing to verify. These cases are marked as ``big'' in the table. \[ \begin{array}{rr|ccccc} && 7 & 49 & 343 & 2401 & 16807 \\ && x=1 & x=2 & x=3 & x=4 & x=5 \\ \hline 9 &y=2 & \text{OK} & 40 \cdot 58 \nmid 272 & \text{big} & \text{big} & \text{big} \\ 81 &y=4 & 74 \cdot 88 \nmid 48 & 32 \cdot 130 \nmid 320 & \text{big} & \text{big} & \text{big} \\ 729 &y=6 & \text{big} & \text{big} & \text{big} & \text{big} & \text{big} \\ 6561 &y=8 & \text{big} & \text{big} & \text{big} & \text{big} & \text{big} \\ \end{array} \]
18.02.2022 15:32
LHS is even, so RHS is also even, so $a$ and $b$ have same parity. This implies $7^a-3^b\equiv 0\pmod 4$, so $a$ and $b$ are even, which implies $7^a-3^b\equiv 0\pmod 8$, so $b\equiv 0\pmod 4$. Set $a=2x$ and $b=2y$ where $y$ is even. So $(7^x-3^y)(7^x+3^y)\mid 16x^4+4y^2$. We note that if $y>3$, then $3^y>4y^2$. If $x>4$, then $7^x> 16x^4$. So if $x>4$ and $y>3$, then $(7^x+3^y)> 16x^4+4y^2$, a contradiction. Case 1: $x\le 4$. If $x=1$, then $3^y+7\le 16+4y^2$. So $y<4$. Now $3^3+7=34\nmid (16+4\cdot 3^2)=52$. Thus, $y<3$. If $y=1$ or $y=2$, then $2(3^y+7)=16+4y^2$, which implies $7-3^y\mid 2$, so $y=2$. Indeed, $\boxed{(a,b)=(2,4)}$ is valid. If $x=2$, then $3^y+49\le 256+4y^2$. So $y<6$. If $y=5$, then LHS is $292$ and RHS is $356$, contradiction. If $y=4$, then LHS is $130$ and RHS is $320$, contradiction. Oops I forgot $y$ is even. Much less casework. If $y=2$, then we get $58\mid 272$, contradiction. If $x=3$, then $3^y+343\le 1296+4y^2$. So $y<7$. If $y=6$, then $1072\mid 1440$, contradiction. If $y=4$, then $424\mid 1360$, contradiction. If $y=2$, then $352\mid 1312$, contradiction. If $x=4$, then $2401+3^y\le 4096+4y^2$. So $y<7$. If $y=6$, then $3130\mid 4240$, contradiction. If $y=4$, then $2482\mid 4160$, contradiction. If $y=2$, then $2410\mid 4112$, contradiction. Case 2: $x>4$ and $y=2$. So we have $7^x+9\mid 16x^4+16$. So $x<5$, a contradiction. We have exhausted all cases, so we are done.
04.03.2022 15:29
11.06.2022 07:39
Note that $7^a-3^b$ is even, so $a^4+b^2$ is even. Thus, $a^4,b^2$ have same parity which means $a,b$ have same parity. $~$ If $a,b$ odd then $7^a-3^b$ is divisible by $4$ but $a^4,b^2\equiv 1\pmod 4$ so $a^4+b^2\equiv 2\pmod 4.$ $~$ Thus, $a,b$ are even. Let $a=2a_1,b=2b_1$ to get $49^{a_1}-9^{b_1}\equiv 0 \pmod 8$ which implies $16a_1^4+4b_1^2\equiv 0 \pmod 8$ so $b_1$ is even. Thus, let $b=4b_2.$ $~$ We have $49^{a_1}-81^{b_2}=(7^{a_1}-9^{b_2})(7^{a_1}+9^{b_2}).$ Note that the first factor is a difference of two odd but distinct integers, so it's at least two. Thus, $7^{a_1}+9^{b_2}\le 8(a_1^2+b_2^2)$ $~$ Easy bounding gives $9^k>8k^2$ and so $7^{a_1}\le 8a_1^2$ implying that $a_1=1,2$ or $3.$ Then, more bounding gives the solution $(1,1).$ Thus, $(a,b)=(2,4)$ is the only solution
07.07.2022 19:16
. Clearly $a$ and $b$ have the same parity. Then, suppose that $a,b\equiv 1\pmod 2$. Then we have that $a^4+b^2\equiv 2\pmod 4$ while $7^a-3^b\equiv 0\pmod 4$, a contradiction. Then we may set $a=2x$ and $b=2y$ to obtain that \[7^{2x}-3^{2y}\mid 16x^4+4y^2\implies 7^x+3^y\mid 16x^4+4y^2\implies 7^x+3^y\le 16x^4+4y^2.\]Then by bounding we obtain that $x\le 4$ and $y\le 6$. Finally, plugging into our original equation gives that only $(a,b)=\boxed{(2,4)}$ works (we omit the calculations).
09.07.2022 04:27
Obviously $a,b$ are the same parity. If they are both odd then $a^4+b^2 \equiv 2 \pmod 4$ and $7^a-3^b \equiv 0 \pmod 4$, a contradiction. Thus $a=2x$ and then $b=2y$. Thus $$7^{2x}-3^{2y}|16x^4+4y^2 \implies (7^x+3^y)(7^x-3^y)|16x^4+4y^2 \implies |(7^x-3^y)|(7^x+3^y) \le 16x^4+4y^2 \implies 7^x+3^y \le 8x^4+2y^2.$$Next note that $3^y \ge 2y^2+1$ for all $y \ge 1$, so we have that $7^x+1 \le 8x^4$. This inequality gives $x=1,2,3$. If $x=1$ then we have that $3^y \le 2y^2+1$ or $y=1,2$, and we get the solution $(x,y)=(1,2)$ so $(a,b)=(2,4)$. If $x=2$, then $3^y \le 79+2y^2$, so $y=1,2,3,4$ upon checking gives no solution. If $x=3$, then $3^y \le 305+2y^2$ which gives $y=1,2,3,4,5$ which upon checking gives no solutions, so the sole solution is $(a,b)=(2,4)$.
21.07.2022 02:42
Missed the obvious bounding at the end :c (oops this looks like asdf's sol) The key claim is that Claim: $2|a$ and $4|b$. Proof: Since $7^a - 3^b$ is even, $a$ and $b$ must have the same parity. Thus $7^a - 3^b \equiv 0 \pmod{4}$ and so $2|a, b$. Then $7^a-3^b \equiv 1 - 1 = 0 \pmod{8}$ and so $4|b$. Let $a = 2c$ and $b = 4d$. We can rewrite the original equation as $$7^{2c}-3^{4d} \mid (2c)^4 + (4d)^2,$$or, $$(7^c+3^{2d})(7^c-3^{2d}) \mid 16(c^4+d^2) \implies (7^c+3^{2d}) \leq 16(c^4+d^2).$$Bound. This is left as an exercise to the reader. The answer is $(a, b) = \boxed{(2, 4)}$. $\square$
01.08.2022 13:40
ISL Marabot solve Clearly, $7^a-3^b$ is even. So, parity of $a,b$ must be same. Therefore, $4|7^a-3^b$. Which gives, $a,b$ are even. Let, $a= 2x, b = 2y$. So, $(7^x+3^y)(7^x-3^y)|16x^4+4y^2$ Here, $7^x- 3^y$ is even and not $0$. So, $(7^x+3^y)|8x^4+2y^2$ $\implies 7^x+3^y \leq 8x^4+2y^2$ But, $3^y > 2y^2 \ \forall y\in \mathbb N$. So, $7^x > 8x^4$ which gives $x\leq 3$. And, for each $x$ we get $y < 3$. And by checking all of the pairs we can see that only $(x,y) = (1,2)$ works. Which gives the only solution $(a,b) = (2,4). \ \ \ \blacksquare$
04.09.2022 20:49
We have that $7^a-3^b$ is even so we need the parities of $a,b$ to be the same. So we have that $4|7^a-3^b$ which implies that $a,b$ are even. We now let $a=2x$ and $b=2y$ so we have that $7^2x-3^2y=(7^x+3^y)(7^x-3^y)|16x^4+4y^2$ this implies that $7^x>8x^4$ else there is a contradiction so we have that $x \leq 3$ and $y<3$ so the only pair is $(x,y)=(1,2)$ which means that $(a,b)=(2,4)$ QED
28.05.2023 22:59
Take mod $2$ $7^a-3^b \equiv 0 \pmod 2$ so $a^4+b^2 \equiv 0 \pmod 2$ $a+b \equiv 0 \pmod 2$ $a \equiv b \pmod 2$ Take mod $4$ $(-1)^a-(-1)^b \equiv 0 \pmod 4$ if $a$ and $b$ have the same parity, so $a^4+b^2 \equiv 0 \pmod 4$ and $a$ and $b$ are all even Let $a=2c, b=2d$ $49^c-9^d$ divides $16c^4+4d^2$ Take mod $8$ $49^c-9^d \equiv 0 \pmod 8$ So $16c^4+4d^2 \equiv 0 \pmod 8$ $4d^2 \equiv 0 \pmod 8$ $d^2 \equiv 0 \pmod 2$ $d \equiv 0 \pmod 2$ Let $d=2e$ $49^c-81^e$ divides $16c^4+16e^2$ $7^{2c}-9^{2e}$ divides $16c^4+16e^2$ $(7^c-9^e)(7^c+9^e)$ divides $16c^4+16e^2$ $|7^c-9^e|(7^c+9^e) \le 16c^4+16e^2$ $7^c+9^e \le 16c^4+16e^2$ For $e \ge 2$, $9^e > 16e^2$ because $e=2$ satisfies it and multiplying by $9$ is more than multiplying by $\frac{(e+1)^2}{e^2}$ so it will always satisfy For $c \ge 5$, $7^c > 16c^4$ because $c=5$ satisfies it and multiplying by $7$ is more than multiplying by $\frac{(c+1)^4}{c^4}$ so it will always satisfy So for $e \ge 2, c \ge 5$, $7^c+9^e > 16c^4+16e^2$ So to satisfy $7^c+9^e \le 16c^4+16e^2$, $e=1$ and $c=1,2,3,4$ Plug in all possible values of $c$ and $e$ to find that only $c=1, e=1$ works and results in $\boxed{a,b=(2,4)}$
08.06.2023 16:25
The answer is $(a,b)=\boxed{(2,4)}$ only, which clearly works. Note that the LHS is a multiple of $4$, so the RHS must also be a multiple of $4$, so $a$ and $b$ are even. However, this makes the LHS a multiple of $8$, so $b$ is a multiple of $4$. Let $a=2m$ and $b=4n$. Then \[(7^m-9^n)(7^m+9^n)\mid 16(m^4+n^2).\]Since the first factor is even, we have \[7^m+9^n\mid 8(m^4+n^2)\rightarrow (7^m-8m^4)+(9^n-8n^2)\le 0.\] Let $f(m)=7^m-8m^4$ and $g(n)=9^n-8n^2$. Then we can see that $f(1)=-1$, $f(2)=-79$, $f(3)=-305$, $f(4)=353$, and $f(n)>353$ for all positive integers $n\ge 5$. We can also see that $g(1)=1$, $g(2)=49$, $g(3)=657$, and $g(n)>657$ for all positive integers $n\ge 4$. Therefore, we have to check $(m,n)=(1,1),(2,1),(2,2),(3,1),(3,2)$. If $(m,n)=(2,1)$, then $58\mid 8(17)$ which is not true. If $(m,n)=(2,2)$, then $130\mid 8(20)$ which is not true. If $(m,n)=(3,1)$, then $352\mid 8(82)$ which is not true. If $(m,n)=(3,2)$, then $424\mid 8(85)$ which is not true. If $(m,n)=(1,1)$, then $(a,b)=(2,4)$. QED.
14.08.2023 23:37
This one's a few minute headsolve. Note that the LHS is divisible by 2, so a,b have same parity; in particular, if a,b odd then LHS is 0 mod 4, while RHS is 2 mod 4, contradiction, so we deduce a=2c,b=2d. Now the problem is evident because $7^c+3^d\mid 7^{2c}-3^{2d}\mid 16c^4+4d^2$, which it's obvious LHS increases faster so by obvious bounding we can manually check solutions that's only $2,4$.
21.12.2023 02:00
As $7^a-3^b$ is even, $a, b$ are of the same parity; on the other hand, this implies $8 \mid 7^a - 3^b$, which forces $2 \mid a$ and $4 \mid b$ for mod $8$ reasons. On the other hand we must now have $7^{a/2} + 3^{b/2} \leq a^4+b^2$. By comparing the $a$ and $b$-terms, it suffices to check $(a, b) = (2, 4), (4, 4), (6, 4), (8, 4)$, of which only $(2, 4)$ works.
18.08.2024 11:09
2007 N1 First of all, $a,b$ are of same parity implying that $ 8|7^a-3^b,$ so $2| a$ and $4|b$ , We can after that easily get that only $(2,4)$ works.
18.08.2024 16:41
Quite a nice problem It's easy to find by $mod 4$ that $2 \mid a$ and $4 \mid b$ So suppose $a=2x$ and $b=4y$, we get that: $(7^x+9^y)(7^x-9^y) \mid 16(x^4+y^2)$ which means that $16(x^4+y^2) \ge 7^x+9^y$ It's obvious that $x \ge 5$ doesn't work and by casework($x=4, 3, 2, 1$) we get that the only solution is x=y=1 which means that $(a;b)=(2;4)$
28.08.2024 05:25
bad formatting because copied from my mathdash solution note that 7^a - 3^b divides to, so a and b have the same parity, both odd causes 7^a - 3^b to be divsiible by 4 whereas a^4 + b^2 = 1 + 1 = 2 mod 4, so fail. then a,b, are both even, taking mod 8 gives a^4 + b^2 divides 8, so 4 | b. now factor (7^a - 3^b) = (7^x- 3^y)/2 *2(7^x + 3^y). now we then have 2(7^x + 3^y) <= 16x^4 + 4y^2, so 7^x + 3^y <= 8x^4 + 2y^2. now we must have one of the inequalities of x < 4, y < 2, true, otherwise we can just sum 7^x >8x^4, 3^y > 2y^2 and win. since y is even, we must have x < 4. x = 3 gives 343 + 3^y | 648 + 2y^2, clearly the left side is more than half the right side so we must have 3^y = 648 - 343 + 2y^2. this cleraly fails by size for y > 5, and y < 5 LHS is 2 digits RHS is 3. then y = 5 fails by direct computation. now we try x = 2, this gives 49 + 3^y <= 128 + 2y^2, for y > 5 we can win by size, y = 5 doesnt work by 2| y, y = 4 fails by direct computation, y = 3 fails by 2 | y, y = 2 fails by szie. now we try x = 1, or 7 + 3^y <= 8 + 2y^2, y > 2 fails by size, y = 2 works and yields the only solution that works, (2,4). it is trivial to see that this works.
04.01.2025 03:15
By mod $4$ both $a, b$ are even. Then mod $8$ yields $4|b.$ Thus let $a=2x, b=4y.$ Then this yields $$(7^x-9^y)(7^x+9^y) \, | \, 16(x^4+y^2) \implies 16(x^4+y^2) \geq 7^x+9^y.$$Note that the RHS increases faster than the LHS for $x \geq 5, y \geq 2.$ If $y \geq 2,$ testing $x=1, 2, 3, 4$ yields no solutions. Therefore $y=1 \implies x = 1, 2, 3, 4.$ Only $x=1$ works meaning that the only solution is $(a, b) = (2, 4).$
04.01.2025 03:47
09.01.2025 09:46
We claim $(2, 4)$ is the only solution. Consider taking both sides mod $4$. We get that clearly the LHS is even, so $a$ and $b$ have equal parity. However, if they are both odd, the LHS is divisible by $4$ but the RHS is not. We perform the substitution $a=2x$, $b=2y$. Then we get $$(7^x-3^y)(7^x+3^y)|16x^4+4b^2.$$It must follow that since $7^x-3^y\neq 0$ and $7^x-3^x$ is even, it follows that $$2(7^x+3^y)|16x^4+4b^2.$$Now note that by this rule $x\leq 3$. Testing and bounding $y$ gives the desired result.