Monday, August 31, 2009

xth root of x

Just how big can x^(1/x) be? Assume x is real number.

Labels: ,

12 Comments:

Anonymous Anonymous said...

Max value of x is e (i.e. 2.71828183...)

August 31, 2009 2:38 PM  
Blogger Chris said...

Roll the drums ... we have a winner! First post too.

August 31, 2009 2:42 PM  
Anonymous t::b:H said...

well, i supposed something like that, but until now i came up to:
d/dx[x^(1/x)*(1/(x^2)+log(x))]

still need the root...

August 31, 2009 3:59 PM  
Anonymous Anonymous said...

Another good question Chris
I am not in any way a mathematician(did A level pure maths and just scraped through)
I just about manage to understand the answer e.
one question from me.... what happens when the value of x approaches infinity?
I'm not trying to outfox you, I only wondered.

August 31, 2009 4:14 PM  
Blogger Chris said...

Hi t::b:H (you're gonna be tbH when I answer, I get spots in front of my eyes reading your isospin name;)

I've lost my notes, but can see you've gone wrong there. I converted to e^((ln(x)/x) first. The rest's a doddle.

August 31, 2009 4:54 PM  
Blogger Chris said...

I don't want to give too much away. If you don't know calculus, you'll find it a tough nut to crack. Same goes for part 2 of "Hold on".

Unless there seems to be no point, I always publish a (over)full answer. Although if I'm lazy, I'll simply provide a link.

August 31, 2009 4:59 PM  
Blogger Chris said...

... I realised that it was a genuine question. There no site rule to say you can't try to outfox me though ;)

I would have been curious too. But I've seen the graph. Here you go:
http://en.wikipedia.org/wiki/E_%28mathematical_constant%29
About half way down.

August 31, 2009 5:27 PM  
Blogger Chris said...

Time's up.
Let y = x^(1/x) = e^(ln(x)/x)

dy/dx = y' = e^(ln(x)/x)(1/(x^2)-ln(x)/(x^2))
= e^(ln(x)/x)(1-ln(x))/(x^2) [don't you love in-line maths!]

For a critical point need y' = 0 => 1 = ln(x) => x = e

So the maximum value is e^(1/e) = 1.44466786...

So, silly old me, has incorrectly announced a winner in error.
I asked for max value of x^(1/x), not for the corresponding x.
However, that's nit-picking. So I'll let the winner list stand.

August 31, 2009 6:44 PM  
Blogger Chris said...

The ever present e has reminded me of a lecture I went to which include a reasonable argument, based on cost, that the best number of logical states for a computer to work with is e. So 2 or 3 is nearest. I notice that most logic circuits are tri-state => they can disconnect themselves from the circuit as well as putting out a 0 or 1.

August 31, 2009 6:46 PM  
Anonymous t.:b:H said...

hmmmrmrmramsrsar....

thanks, got involved in a chainrule derivation war... that e^ln is for sure the better solution =)

September 1, 2009 2:30 AM  
Blogger Chris said...

Hi t:b:H. I can't even remember d/dx(a^x) = ln(a)*a^x. I always do a^x = e^(x ln(a)) first. I've not had to this stuff in real life for about 40 years and that was only to get a degree (in electronics). But it's a bit like riding a bike.

September 1, 2009 6:11 AM  
Blogger zwidorff said...

Hi there! e as the maxima of the x root of x is one of my favorite math items. I learned of it from e: the story of a number. i just used the euler program to confirm this. strange. e is some kind of strange fulcrum.

April 22, 2010 11:35 PM  

Post a Comment

Links to this post:

Create a Link

<< Home