Want to share your content on R-bloggers? click here if you have a blog, or here if you don't.
a1+…+a10=100
and
a1+2a2+…+10a10=2a2x….x10a10,
which reduces the number of unknowns from 100 to 10 (or even 9). Furthermore, the fact that the (first) sum of the ai‘s is less than 100 implies that the (second) sum of the iai‘s is less than 1000, hence iai is less than 1000. This reduces the number of possible ten-uplets enough to allow for an enumeration, hence the following R code:
bounds=c(100,trunc(log(1000)/log(2:10))) for (i2 in 0:bounds[2]) for (i3 in 0:bounds[3]) for (i4 in 0:bounds[4]) for (i5 in 0:bounds[5]) for (i6 in 0:bounds[6]) for (i7 in 0:bounds[7]) for (i8 in 0:bounds[8]) for (i9 in 0:bounds[9]) for (i10 in 0:bounds[10]){ A=c(i2,i3,i4,i5,i6,i7,i8,i9,i10) if (sum(A)<101){ A=c(100-sum(A),A) if (sum((1:10)*A)==prod((1:10)^A)) print(A) }}
that produces two answers
[1] 97 0 0 2 0 0 1 0 0 0 [1] 95 2 3 0 0 0 0 0 0 0
i.e. either 97 1′s, 2 4′s and 1 7, or 95 1′s, 2 2′s and 3 3′s. I would actually love to see a coding solution that does not involve this pedestrian pile of “for”. And a mathematical solution based on Diophantine equations. Rather than the equally pedestrian solution given by Le Monde this weekend.
Filed under: Books, Kids, R Tagged: Diophantine equations, Le Monde, mathematical puzzle, R
R-bloggers.com offers daily e-mail updates about R news and tutorials about learning R and many other topics. Click here if you're looking to post or find an R/data-science job.
Want to share your content on R-bloggers? click here if you have a blog, or here if you don't.