I have an equation that I've been trying to solve for fun algebraically after helping someone learn logarithm and exponent rules: $$3^x+11^{2x^2+4x-7}=311$$ So far, no manipulation using logarithms or the changing of bases has helped. I know it is probably not possible to solve it like this, but I was wondering if there is any method to getting an exact solution, not a numerical one (Desmos gives ~-3.3868 and ~1.38616). If there is no method, how come? I feel like there should be some expression that perfectly captures the solution, whether it be algebraic, complex, or something else. My goal is to figure out how to solve these more difficult exponential equations in case someone asks me. Sorry if this question is relatively simple compared to others on this site, but I don't know where to ask this.
-
2An easier example is already "unsolvable" – Quý Nhân May 27 '25 at 03:48
1 Answers
There is solution as an infinite series.
$$f(x)=3^x+11^{2x^2+4x-7}-311$$
I assume that you noticed thet $f(x)$ is very stiff and that the positive solution is rather close to $x=\frac 32$.
So, consider instead a smooth transform of $f(x)$ such as $$g(x)=(2x^2+4x-7)\log(11)-\log(311-3^x)$$ and expand it as a Taylor series around $x=\frac 32$.
This gives $$g_p(x)=\sum_{n=0}^p a_n\,\left(x-\frac{3}{2}\right)^n+O\left(\left(x-\frac{3}{2}\right)^{p+1}\right)$$ The $a_n$ as quite nasty but this does not matter.
Now, use power series reversion to get explicitly $x_{(p)}$ using the explicit formula given by Morse and Feshbach for the $n^{\text{th}}$ term of the inverse series.
For example $$x_{(1)}=\frac 32-\frac{48347 \left(7 \log (11)-2 \log \left(311-3 \sqrt{3}\right)\right)}{27 \log (3)+933 \sqrt{3} \log (3)+966940 \log (11)}=1.38875$$ while the solution is $1.38616$.
Playing with $p$ and converting the result to decimals $$\left( \begin{array}{cc} p & x_{(p)} \\ 1 & 1.38875190259 \\ 2 & 1.38627322206 \\ 3 & 1.38616299535 \\ 4 & 1.38615686078 \\ 5 & 1.38615647853 \\ 6 & 1.38615645301 \\ 7 & 1.38615645122 \\ 8 & 1.38615645109 \\ 9 & 1.38615645108 \\ \end{array} \right)$$
and we could continue forever.
For example $x_{(15)}=\color{red}{1.38615645108209659}12$
For the negative root, do the same around $x=-\frac 72$; for this case $$x_{(1)}=-\frac 72+\frac{\left(\sqrt{3}-25191\right) \left(8 \log (3)+7 \log (11)-2 \log \left(25191-\sqrt{3}\right)\right)}{2 \left(\sqrt{3} \log (3)-251910 \log (11)+10 \sqrt{3} \log (11)\right)}$$ which is $-3.38936$ while the solution is $-3.38680$.
For example $x_{(15)}=\color{red}{-3.386801949501781337}46$
- 4,757
- 289,558
-
What do you mean by $f(x)$ being "stiff"? I've not seen that term used before. – Teepeemm May 27 '25 at 18:54
-
Adding on to @Teepeemm's comment, how did you get the 'smooth transform' of f(x), and how does it help? I'm assuming it makes it easier to write as a Taylor series (or even makes it possible?). – McKale May 27 '25 at 23:52
-
@Teepeemm. Wrong word ! Sorry. I wanted to mean that $f(x)$ varies very fast. – Claude Leibovici May 28 '25 at 03:00
-
@McKale. Wrong word ! Sorry. I wanted to mean that $f(x)$ varies very fast. – Claude Leibovici May 28 '25 at 03:01
-
To answer McKale's question, it looks like you set $f(x)=0$, solved for the $11$ part, took logarithms, and used that to create your $g$. It looks like power series reversion needs $g^{-1}$. How did you find that? Or am I misunderstanding what you did next? (Would Newton's method work for $f$ or $g$?) – Teepeemm May 28 '25 at 17:59