# Infinitely many mutually coprime numbers in the same residue class

Suppose we are considering the residue class of all numbers $\equiv a \pmod b$, where $(a,b)=1$. Is there an elementary (and relatively simple) way to see that there is a subsequence that consists entirely of elements that are mutually coprime?

Of course, this must be a true statement, as it follows immediately by Dirichlet's Theorem. But this statement is a bit weaker, so is there perhaps a way to construct a sequence of mutually coprime integers all in the same residue class? I was thinking something like a Fermat construction (i.e. $F_n=1+\prod_1^{n-1}F_i$) would work, but I cannot see how to see this method through. Anyway, there might not even be an elementary solution, and I may be approaching my original problem in the wrong way.

• Wouldn't $a + b^k$ be such a sequence? – fleablood Feb 10 '17 at 6:28
• @fleablood Are all numbers of the form $1+2^k$ mutually coprime? Nope. – Erick Wong Feb 10 '17 at 6:38

Let $u_1=a$, $u_2=b+a$, $v_k=u_2\times u_3\times\cdots\times u_k$ for $k=2,3,\dots$, $u_{k+1}=bv_k+a$ for $k=2,3,4,\dots$.
E.g., for $a=2$, $b=5$, we get $u_1=2$, $u_2=7$, $u_3=5\times7+2=37$, $u_4=5\times7\times37+2=1297$, $u_5=5\times7\times37\times1297+2$, and so on.