What category of equation is this?

What methods are available to solve it?

$2^x -x^3 = 0$ where $x\in\Bbb R$

  • $\begingroup$ According to intermediate value theorem, one root of $f(x)=2^x-x^3$ is between $1$ and $3$. $\endgroup$ – Mikasa Jun 27 '12 at 11:48
  • $\begingroup$ Since it is not equivalent to a polynomial equation, it would be called a transcendental equation. BTW, that Wikipedia article needs some expert cleanup. $\endgroup$ – hardmath Jun 27 '12 at 13:08

You can find a solution in terms of the Lambert W Function. Rewrite as:

$$ 1 = \frac{x^3}{2^x} = x^3 \exp(-x\log 2) $$

and take the real cube root:

$$ 1 = x \exp \left(-\frac{x\log 2}{3}\right) $$

Now multiply by $-\log 2/3$:

$$ -\frac{\log 2}{3} = -\frac{x\log 2}{3} \exp\left(-\frac{x\log 2}{3}\right) $$


$$ x = -\frac{3W_0\left( -\frac{\log 2}{3}\right)}{\log 2} $$

where $W_0$ is the principal branch of Lambert's W. The value is about 1.37.

There is another real root between 9 and 10, which is found on the second branch of the Lambert W function:

$$x = -\frac{3W_{-1}\left(-\frac{\log 2}{3}\right)}{\log 2}$$

and whose value is about 9.94.


Consider $f(n)=2^n-n^3$. $f(9)=-217$ and $f(10)=24$,therefore,there exists a root between $9$ and $10.$ You can solve $f(n)=0$ numerically using Newton-Raphson method taking $x_0=9$.

Also $f(1)=1$ and $f(2)=-6$, therefore there will be a root between 1 and 2.

For $n<1$, $f(n)$ is always positive and for $n>10$, $f(n)$ is always positive,so there are no more roots other than between $9$ and $10, 1$ and $2.$

  • $\begingroup$ For the root between 1 and 2, the fixed point iteration $x := 2^{x/3}$ seems to work well (though not as fast as Newton!), while between 9 and 10 the inverse iteration $x := 3 \log_2 x$ seems to work. $\endgroup$ – hardmath Jun 27 '12 at 13:03
  • it is an exponential equation or an exponential diophantine equation if only integers (or fractions) are allowed
  • you may find a real or imaginary solution :
    • using iterations starting with $x_0=2$ ($x_{n+1}=2^{x_n/3}$, the converse $x_{n+1}=\frac {3\ln(x_n)}{\ln 2}$ or faster Newton-Raphson iterations)
    • with the LambertW function (since $n e^{-\frac {n\ln 2}3}=1$ is wanted)
    • graphically
    • ...

This is an equation without solution, if $n$ is supposed to be integer. In order for $2^n=n^3$ to hold, $n$ should be

  • positive, since $2^n$ is,
  • a power of $2$, since $n^3$ would otherwise have an odd prime factor,
  • divisible by $3$ in order for $2^n$ to be a cube.

Obviously the last two conditions are contradictory.

Since $n$ has now been renamed $x$ and has become real, there are more solutions. Probably just the two indicated by @avatar, but some effort is required to show that this is all, since $f: x\to 2^x-x^3$ is not a convex function: it has two inflexion points, one near $0.08$ and another one near $6.3$.

  • $\begingroup$ Can you explian the third point please ? $\endgroup$ – Belgi Jun 27 '12 at 11:50
  • 1
    $\begingroup$ @Belgi: $2^n$ has an obvious prime factorization (or should I say, it is a prime factorization), and prefect cubes are precisely those integers in whose prime factorization every prime factor occurs with multiplicity divisible by $3$. $\endgroup$ – Marc van Leeuwen Jun 27 '12 at 11:54
  • 1
    $\begingroup$ 3 Unique points. I love the way you attack the problem. $\endgroup$ – Mikasa Jun 27 '12 at 11:56

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.