0

Without using a calculator how to solve $x^x = 100$ ? A way of finding an approximation to 2 decimals would be good neough.

I know about the Lmabert W function but one cannot compute it mentally. This is why I believe my question not to be a duplicate.

4 Answers4

2

If you know that $\log2\approx0.3$ and $\log3\approx0.48$ (where "$\log$" is log base $10$), then you can show that

$$\begin{align} \log(3.6^{3.6}) &=3.6\log(3.6)=7.2(\log2+\log3)-3.6\\ &\approx7.2\cdot0.78-3.6\\ &=0.001((75-3)(75+3))-3.6\\&=0.001(75^2-9)-3.6\\ &=0.001(5625-9)-3.6\\ &=5.616-3.6=2.016\\ &\approx2 \end{align}$$

Note the "trick" $(a5)^2=(a(a+1))25$ for squaring $75$ (i.e., $56=7\cdot8$).

Remark: $\log2\approx0.3$ comes from $2^{10}=1024\approx10^3$. The estimate for $\log3$ can be obtained with a bit a rounding from $3^4\approx80$.

Barry Cipra
  • 79,832
  • How do you get the $3.6\log(3.6)=7.2(\log2+\log3)-3.6$ ? – astudentofmaths Dec 20 '17 at 15:20
  • 1
    @Arthurim, $\log(3.6)=\log36-\log10=2\log6-1=2(\log2+\log3)-1$ – Barry Cipra Dec 20 '17 at 15:24
  • Just, how do you get to first estimate 3.6? We know that x is between 3 and 5, but unless you knoiw the answer you have to go by trial and error right? – astudentofmaths Dec 20 '17 at 15:34
  • @Arthurim, I started with the obvious bounds $3\lt x\lt4$, as in gimusi's answer, then knew that $\log3.6$ would be easy to estimate from easy estimates of $\log2$ and $\log3$. What's missing, of course, is any estimate of how close $3.6$ is to the actual answer. That would require a good deal more work. – Barry Cipra Dec 20 '17 at 21:12
1

You can try:

$3^3=27$

$4^4=256$

thus $3<x<4$.

Then you can try with $3.5^{3.5}$ and so on.

user
  • 154,566
0

You can try to approximate the final solution using rational numbers $p/q$. For this you try to find $p$ and $q$ so that

$$\left(\frac pq\right)^{p/q}\approx100\qquad\Leftrightarrow\qquad p^p\approx 100^q\cdot q^p.$$

With some time and patience, the latter form can be computed on paper. What makes an approximation good? Probably when you have that the quotient of $p^p$ and $100^q\cdot q^p$ is near one.


An algorithm

A good starting point (as seen in @gimusi's answer) would be $p=7$ and $q=2$. You would find that $823543=7^7 < 100^2\cdot 2^7=1280000$. Now you can do the following:

  • If $p^p < 100^q\cdot q^p$ (as in this example), then replace $p\mapsto 3p$ and $q\mapsto2q$.
  • If $p^p > 100^q\cdot q^p$, then only replace $q\mapsto2q$.

This is basically binary bisection. So according to above, the next approximation to check would be $p=21$ and $q=8$. And you do this again and again until the result is good enough for you. The exact result is irrational, so you cannot hope to find it this way.

M. Winter
  • 29,928
0

You could potentially use a sequence of approximations. Suppose you want to approximate this with rational numbers, so let $x=\frac{a}{b}$.

$$ (\frac{a}{b})^{\frac{a}{b}}=100 $$

Rearranging this gives $(100b)^b=a^a$. Now let $b=1,2,3\ldots$ and find a suitable $a$ that bounds the answer above or below. For example if $b=1$:

$$ 27=3^3<100=(100\cdot1)^1<4^4=256, $$

$b=2$: $$ 3125=5^5< 40,000=(100\cdot 2)^2 <6^6=46656, $$

and $b=3$:

$$ 16,777,216=8^8< 27,000,000=(100\cdot 3)^3 <9^9=387,420,489, $$

one can obtain an upper and lower sequence of rational approximations:

$$ 3,\frac{5}{2},\frac{8}{3}\ldots,\\ 4,\frac{6}{2},\frac{9}{3}\ldots\\ $$

although this approach doesn't appear to converge terribly quickly.