Given the linear diophantine $3^n x - 2^n y = 1$, ($n$ being a natural parameter) it's well known that it has infinitely many solutions.

My question is whether we can find a generic particular solution of it in terms of $n$.

I inspected some values using the online linear diophantine solver > https://www.math.uwaterloo.ca/~snburris/htdocs/linear.html yet found no pattern.

I also tried to conjecture the inverse of $3^n$ modulo $2^n$ in terms of $n$ assuming a formula could be found but in vain.

Any ideas are welcome. Thanks.



Since it is $$ 3^{\,n} x - 2^{\,n} y = 3^{\,n} x + 2^{\,n} \left( { - y} \right) = 1 = \gcd \left( {2^{\,n} ,3^{\,n} } \right) $$ then the Bezout Identity assures that we can find two integers such that the identity above is satisfied.

The two integers $x,y,$ will come by applying the Extended Euclidean Algorithm to the couple $3^n , \, 2^n$, or by expanding $(3/2)^n$ into a Continued Fraction.

But in fact, the process does not look to show any discernible path at the varying of $n$.


Lets see the result needs to be odd ( 1 mod 2) . That requires x be 1 mod 2 . The result is also, 1 mod 3 leading to if n is odd y is 2 mod 3, if n is even y is 1 mod 3 . These two combine because the result is Also 1 mod 6 (1 mod 2 and 1 mod 3 together) etc. It's Also 1 mod n but unless n is prime we don't get much. if we say n is prime, we get x and y both being 1 mod n works ( may not be the only way though).


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.