Are there numbers that can be written as a power using their own base ten digits in order?

98 Views Asked by At

The number twenty seven, written in base eight, has the property that, if its last digit is written as a superscript, the resulting expression is still twenty seven.

$$33 \rightarrow 3^{3} = 33 \qquad (\mathrm{VIII})$$

The same thing happens in base two, if one raises more digits.

$$11011 = 11^{011} \qquad (\mathrm{II})$$

There's also one of these "height independent" numbers in base six: sixteen.

$$24 = 2^{4} \qquad (\mathrm{VI})$$

Is there a name for these kind of numbers? I think they are a subset of Friedman numbers in their respective bases, but with only one operation and no reordering. Also, are there any numbers that have this property in base ten? I think that if there are any, they must be greater than $10^{14}$.


Some extra info, in case it helps proving something:

If raising the last digits of the number leads to the expression $\alpha^{\beta}$ , then the original number must be the concatenation of $\alpha$, a (possibly empty) string of zeroes, and $\beta$. If the number is written in base b and it has the "digit height independence" property,

$$\alpha^{\beta} = \alpha b^{n} + \beta$$

Solving for b, so that we can tell in which base does a given number have the desired property,

$$b = \sqrt[n]{\alpha^{\beta - 1} - \frac{\beta}{\alpha}}$$

In order to actually search for a specific $b$, it's enough to focus in the case with $n = 1$, and double check results that are perfect powers.

$$b = \alpha^{\beta - 1} - \frac{\beta}{\alpha}$$

The formula makes sense for $\alpha, \beta \ge 2$. Since $\alpha \le \beta$, and $b$ grows when either of them grow, if

$$2^{\beta_0-1}-\frac{\beta_0}{2} = b_0$$

then all possible values of $b \lt b_0$ must occur for pairs of $\alpha$ and $\beta$ with $\beta \ge \alpha \ge 2, \beta_0 \ge \beta \ge 2$.

I searched all those numbers for $\beta_0 = 50$, $b_0 \approx 5*10^{14}$. Since there were no powers of ten among all the $b \lt b_0$ (actually, the only power was eight), I know that if there is a $b$ that is a power of ten, it has to be at least $10^{15}$, so the actual number

$$\alpha b^{n} + \beta$$

must be even greater than that.