Define our Harmonic sequence for two numbers such that \begin{equation} a_{n+1} = \frac{2a_nb_n}{a_n + b_n} \end{equation} and our geometric sequence \begin{equation}b_{n+1} = \sqrt{a_nb_n} \end{equation} such that as $n \rightarrow \infty$ we tend towards the Geometric-Harmonic Mean.

The arithmetic-geometric mean can be defined by the following two sequences. First compute the arithmetic mean of two numbers $a,b \in \mathbb{R_+}$ such that $a_1$ = $\frac{1}{2}(a + b)$ and then compute the geometric mean such that $b_1$ = $\sqrt{ab}$. If we continue to iterate this operation we can define our two sequences {$a_n$} and {$b_n$} as follows: \begin{gather*} a_{n+1} = \frac{1}{2}(a_n + b_n) \\b_{n+1} = \sqrt{{(a_nb_n)}}. %Try and get this all under 1 square root sign \end{gather*} As ${n\to\infty}$ we approach our arithmetic-geometric mean (agM).

I have proved that the Geometric-Harmonic Mean exists and that $a_n$ = $b_n$ as $n \rightarrow \infty$. However, Mathworld states without proof that the limit for our Geometric-Harmonic mean can be written as \begin{equation*} \lim_{n \to \infty}a_n = \frac{1}{M({a_n}^{-1},{b_n}^{-1})} (*) \end{equation*}

Where M is the arithmetic geometric mean. You can find the link here: http://mathworld.wolfram.com/Harmonic-GeometricMean.html

Can anyone help me prove (*)?


Let us suppose that as per your definition $\{a_{n}\}, \{b_{n}\}$ form Harmonic-geometric sequences. Let $A_{n} = 1/a_{n}, B_{n} = 1/b_{n}$ Then clearly we have $$B_{n + 1} = \sqrt{A_{n}B_{n}}$$ and \begin{align} A_{n + 1} &= \frac{1}{a_{n + 1}}\notag\\ &= \frac{a_{n} + b_{n}}{2a_{n}b_{n}}\\ &= \dfrac{\dfrac{1}{A_{n}} + \dfrac{1}{B_{n}}}{\dfrac{2}{A_{n}B_{n}}}\notag\\ &= \frac{A_{n} + B_{n}}{2}\notag \end{align} It follows that $\{A_{n}\}, \{B_{n}\}$ form Arithmetic Geometric sequence and hence both tend to the common limit which we denote by $M(A_{1}, B_{1}) = M(a_{1}^{-1}, b_{1}^{-1})$. Now $A_{n} = 1/a_{n}$ so that $a_{n} \to 1/M(a_{1}^{-1}, b_{1}^{-1})$. Thus the desired relation is proved.

It thus goes on to show that the harmonic geometric sequences don't lead to a new concept and behaves as reciprocal of arithmetic geometric sequences.

  • $\begingroup$ Thank you! This solves my issue, and I would vote the answer up if I had but 15 reputation! $\endgroup$
    – rjadler
    Mar 15 '15 at 22:06

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.