Ratio of sum of digits functions in two bases

Imagine you are a digital detective trying to crack a code hidden inside numbers. In the world of mathematics, every number can be written down in different "languages" called bases. The most common one is base 10, where we use digits 0 through 9. But you could just as easily write numbers in base 2 (using only 0s and 1s) or base 3 (using 0, 1, and 2). Think of a base like a different set of building blocks: in base 10, a number is a tower of blocks where each level represents a power of 10; in base 2, it's a tower of powers of 2.

Now, imagine you have a special tool that counts the total value of all the blocks in a number's tower. Mathematicians call this the "sum of digits." If you take the number 19, in base 10 it's $1 + 9 = 10$. But if you write 19 in base 2, it looks like $10011_2$, and the sum of those digits is $1+0+0+1+1 = 3$. The big question that has puzzled mathematicians for decades is: What happens if you compare these sums across two completely different languages? If you pick two bases that don't share a simple mathematical relationship (like 2 and 3, or 5 and 7), do the sums of their digits ever line up perfectly? Do they ever hit a specific ratio, like one being exactly twice the other? This isn't just a game of numbers; it touches on deep mysteries about how randomness and order coexist in mathematics, helping us understand the hidden structure of the universe's most basic building blocks.

This paper, written by Pascal Jelinek, dives straight into that mystery. The author proves a powerful new rule about how these digit sums behave when you switch between two "independent" bases. Before this, we knew that the ratio of these sums could get arbitrarily close to any positive number, but we didn't know if they could actually hit every single rational number (like 1/2, 3/4, or 7/11) exactly. Jelinek shows that they do. In fact, for any two bases that are multiplicatively independent (meaning one isn't just a power of the other, like 2 and 4), and for any target ratio you can name, there are infinitely many numbers where the sum of the digits in the first base divided by the sum of the digits in the second base equals that ratio exactly.

The paper doesn't just say "it happens"; it provides a rigorous mathematical proof that this happens infinitely often. It even gives a formula to estimate how many such numbers you can find below a certain size, showing that while they might be rare, they are definitely there. The author also solves the other side of the coin: what if the bases are related, like base 4 and base 16? In that case, the paper proves that the ratio can only fall within a specific, limited range, and it describes exactly which ratios are possible and which are impossible.

To visualize this, imagine two different types of dice. One die has faces numbered 0 to 9 (base 10), and another has faces 0 to 2 (base 3). You roll a number of times to build a number. The "sum of digits" is just the total score you get. Jelinek's work is like proving that no matter what score ratio you want to achieve between the two dice (say, the base 10 score is always 1.5 times the base 3 score), there is an infinite sequence of rolls that will make that exact ratio happen. He also proves that if the dice are somehow linked (like one is just a re-skinned version of the other), you can't get just any ratio; you're stuck with a specific set of outcomes.

The proof is a masterclass in using "arithmetic progressions." Think of this as a line of numbers where each step is the same size, like 5, 10, 15, 20. The author shows that if you pick the right starting point and the right step size, the digit sums along that line behave in a very predictable, almost magical way. They distribute themselves evenly across all possible remainders, allowing the author to "tune" the numbers until the ratio hits the target perfectly. This method extends previous work that only solved the problem for specific cases (like base 2 and base 3) to a general rule that works for almost any pair of bases you can imagine.

In the case of dependent bases (like base 4 and base 16), the paper acts as a strict gatekeeper. It proves that you cannot achieve just any ratio. For example, if you are comparing base $b^k$ and base $b^\ell$, the ratio of the digit sums is trapped between $1/(b^\ell - 1)$ and $b^k - 1$. The author demonstrates that every single rational number within this trap is attainable infinitely many times, but nothing outside of it is possible. This gives us a complete map of the territory: we now know exactly which ratios are possible and which are forbidden, depending on the relationship between the two bases.

So, what's the takeaway? If you pick two bases that are mathematically "strangers" to each other, the universe of numbers is rich enough to contain an infinite supply of numbers that satisfy any digit-sum ratio you can dream up. If the bases are "relatives," the possibilities are more limited, but still perfectly predictable. Jelinek's work turns a vague hope that "maybe these ratios exist" into a concrete, proven fact, filling in the missing pieces of a puzzle that mathematicians have been trying to solve for years. It's a reminder that even in the rigid world of integers, there is a surprising amount of flexibility and infinite variety waiting to be discovered.