Alexis Lemaire got famous after finding the 13th root of a computer-generated 200-digit number without calculator. In this article, they say that there are "with 393 trillion possible answers to choose from". At first, I thought that it was the number of permutations of digits for each possible answer, but this wouldn't give the $393$ trillion, I guess. Does it actually mean anything or is it just a misunderstanding on the part of the journalists?
-
I've read non-scientific articles with huge mistakes. Perhaps purposeful mistakes. They don't really care if it's a permutation or a combination, maybe a set. That being said, I can't answer your question. – OFRBG Oct 21 '16 at 12:43
-
@O.VonSeckendorff Yes, I suspect it's a mistake too. – Red Banana Oct 21 '16 at 12:50
-
You forgot the word “trillion” in the title. – Hans Lundmark Oct 21 '16 at 15:03
-
@HansLundmark Yes, thanks. – Red Banana Oct 21 '16 at 15:08
-
@OppaHilbertStyle This is an example of a great question, +1 – Oct 24 '16 at 08:13
-
@Bacon Sarcasm? – Red Banana Oct 24 '16 at 09:04
2 Answers
We have to compute the $13$-th root of an unknown number $N$ such that $10^{199}\le N<10^{200}$ so that $\;10^{199/13}\le N^{1/13}<10^{200/13}$ : all the choices are not possible but only the values from $ 2,030,917,620,904,736\;$ to $\;2,424,462,017,082,328$. This gives at most : $$10^{200/13}-10^{199/13}\approx 393,544,396,177,593\quad\text{possibilities}$$ $$-$$ For something 'simpler' suppose known the $60000$ digits of the perfect power $\;N=k^{11001}\;$ and try to find the positive integer $k$. You'll notice that $k$ can only take the values from $284420$ to $284478$.
Remembering the two last digits of the $11001$-th powers of $\;(284420+i)\;$ for $i$ from $0$ to $58$ : $$0, 21, 72, 23, 24, 25, 76, 27, 28, 29, 0, 31, 32, 33, 84, 75, 36, 37, 88, 39, 0, 41, 92, 43, 44, 25, 96, 47, 48, 49, 0, 51, 52, 53, 4, 75, 56, 57, 8, 59, 0, 61, 12, 63, 64, 25, 16, 67, 68, 69, 0, 71, 72, 73, 24, 75, 76, 77\quad\text{(and some tricks to distinguish the degenerate cases like $0$) }$$ should help you, after just a smart glance at the $60000$ digits of the $11001$-th power, to provide instantly the wished $11001$-th root !
But to choose between $59$ values we don't really need to memorize all these values (nor even require $N$ to be a perfect power). Mental calculators usually know their common logarithms and may use :
$$N^{1/11001}\approx 284419+59.53\;\log_{10}N_2,\\\quad\text{with $N_2$ defined by $\,N=N_2\;10^{60000-1}\;$ and}\;\,1\le N_2<10$$
this is easily obtained from $\;N^{1/11001}\approx 10^{\large{\frac{60000-1}{11001}+\frac{\log_{10}(N_2)}{11001}}}\approx 10^{\large{\frac{60000-1}{11001}}}\left(1+\frac{\ln(10)}{11001}\log_{10}(N_2)\right)$.
A working precision of $2$ digits for $N_2$ should be enough while replacing $59.53$ by $60$ gives an error bounded by $\,0.48$.
We may thus "reduce the possibilities" by using the most significant digits and, for a perfect power only, by exploiting the less significant ones (as explained by Barry Cipra).
It should be clear that all this doesn't explain the speed of A. Lemaire and others who used specific techniques like memorizing the different $13$-th roots possible for the first digits (specifically for $200$ digits numbers) as well as for the last digits.
Many of the methods used are neatly exposed in Ron Doerfler and Miles Forster article :
"The $13$-th Root of a $100$-Digit Number" starting with the methods used by Wim Klein (from Ron Doerfler's blog). Let's conclude with some of his wise words :
“What is the use of extracting the $13$-th root of $100$ digits? ’Must be a bloody idiot,’ you say.
No. It puts you in the Guinness Book, of course”

- 43,021
- 5
- 86
- 140
-
1
-
@OppaHilbertStyle: Ok here only the fixed digit $2$ but I saw compittions where the $10000$-th root was computed by 'head' and this gave quite a fiew fixed digits (I edited my answer...). – Raymond Manzoni Oct 21 '16 at 12:58
-
Minors corrections "between..." should be "from $2030917620904736$ to $2424462017082328$ and "a fiew" $\to$ "a few" ;-) – Raymond Manzoni Oct 21 '16 at 13:49
-
1
-
2
-
Thanks to notice @Barry Cipra (I thought at it and forgot...) Cheers, – Raymond Manzoni Oct 21 '16 at 17:02
This is just an elaboration on Raymond Manzoni's answer, addressing the OP's question whether the journalist misunderstood something.
Here are the lede paragraphs from the news story (which features a photo illustration of Lemaire pondering the 200-digit number):
When the answer is 2,407,899,893,032,210 you know the question is tough.
Not so tough, however, that Alexis Lemaire could not work it out in his head. His challenge yesterday was to come up with the 13th root of a computer-generated 200-digit number.
And, with 393 trillion possible answers to choose from, the PhD student made it almost look easy.
The "393 trillion" almost certainly came from a press release prepared by the sponsor of the challenge; no reporter (except maybe me) has the time or the expertise to make such a calculation (and I would probably get it wrong). Given that the 13th root of the 200-digit number was an integer, it seems fairly clear that what the computer did was pick a (random?) number $n$ such that $10^{199}\le n^3\lt10^{200}$, so that there are $\lfloor10^{200/13}\rfloor-\lfloor10^{199/13}\rfloor\approx393$ trillion choices for $n$.
It's perhaps worth noting, though, that $a^{13}\equiv a$ mod $10$ for all $a$, which means that that a quick look at the final digit of the 200-digit number tells you the final digit of the answer. So in a sense this quickly cuts the number of possible answers down to around $39.3$ trillion. And in this case, in fact, the 200-digit number ends in a string of thirteen $0$'s preceded by a $1$, which means the answer necesarrily ends in $10$, thus cutting the number of possibilities to a "manageable" $3.9$ trillion....

- 79,832
-
Thanks for the continuation Barry! (factorial Cipra :-)). From your citation I found too this link. Of course we may further reduce the possibilities from the left by evaluating $;l:=\log_{10}(N)/13;$ (and $10^l$ after that) but it seems that the $13$-th root evaluation was and 'industry' with great calculators names like Wim Klein and an excellent article by Ron Doerfler and Miles Forster "The $13$-th Root of a 100-Digit Number". Thanks again for the follow up, – Raymond Manzoni Oct 22 '16 at 16:01
-
1@RaymondManzoni, excellent link! You might edit it into your own answer as well! (Incidentally, I hope my factorial comment was understood as the gentle joke I intended it to be.) – Barry Cipra Oct 22 '16 at 17:44
-
No problem @Barry Cipra I considered it that way(!) (with some minor aftershocks... :-)). The Doerfler-Foerster paper is indeed excellent and I'll introduce it in my answer. Cheers, – Raymond Manzoni Oct 22 '16 at 18:16