I'm given the repeating sequence $1,3,5,7,1,3,5,7,\dots.$ I must determine a formula for the $n$th term of the sequence.

I have tried the triangle method, but since this is a repeating series it is not working.

  • 1
    $\begingroup$ And what is your question? $\endgroup$ – abiessu Sep 18 '13 at 23:42
  • 1
    $\begingroup$ In addition to asking a question, you might talk about the "triangle method" because I've never heard of anything with that name. $\endgroup$ – Eric Stucky Sep 18 '13 at 23:45
  • $\begingroup$ @EricStucky I would guess that OP means finding the binomial transform, which can be done with a triangle-shaped diagram of successive differences. It works great for finding equations to some sequences, but not here $\endgroup$ – Cocopuffs Sep 19 '13 at 0:01

There is a way to get a formula for the sequence that allows you to simply plug in a number. Let's call this number $n$. We'll start with something simple and build up to the grand finale of four numbers.

Then, the idea is to start with something that repeats. For example, $(-1)^n$ repeats. Its sequence goes $1,-1,1,-1,\dots$

So if we take $\displaystyle\frac{(-1)^n+1}{2}$ we get $1,0,1,0,\dots$ We can then multiply this by a number to get a sequence that repeats this number. So we have:

$$a\cdot\displaystyle\frac{(-1)^n+1}{2} = a,0,a,0,\dots$$

Next, we change the power of $n$, by adding one to it. This gives $\displaystyle\frac{(-1)^{\color{red}{n+1}}+1}{2}$, and we can then get:

$$b\cdot\displaystyle\frac{(-1)^{n+1}+1}{2} = 0,b,0,b,\dots$$

Adding the two equations gives:

$$a\cdot\displaystyle\frac{(-1)^n+1}{2} + b\cdot\displaystyle\frac{(-1)^{n+1}+1}{2} = a,b,a,b,\dots$$

To get a sequence of four numbers, we use something that repeats every four terms. This number is $(e^{2 \pi i/4})^n$, where $e$ is the exponential, and $i$ is the imaginary number unit. The sequence goes $1, i, -1, -i, 1, i, -1, -i,\dots$

The next trick is to eliminate part of this sequence. If we take the cosine of this number, we get

$$\cos{\left((e^{2 \pi i/4})^n \right)} = 1, 0, -1, 0, 1, 0, -1, 0,\dots$$ $$c \cdot \cos{\left((e^{2 \pi i/4})^n \right)} = c, 0, -c, 0, c, 0, -c, 0,\dots$$

Again, if we add one to the power, we get:

$$d \cdot \cos{\left((e^{2 \pi i/4})^{n+1} \right)} = 0, -d, 0, d, 0, -d, 0, d,\dots$$

Collecting all of the equations together, we get:

$$a\cdot\displaystyle\frac{(-1)^n+1}{2} + b\cdot\displaystyle\frac{(-1)^{n+1}+1}{2} + c \cdot \cos{\left((e^{2 \pi i/4})^n \right)} + d \cdot \cos{\left((e^{2 \pi i/4})^{n+1} \right)}$$ $$ = a+c, b-d, a-c, b+d,\quad a+c, b-d, a-c, b+d,\dots$$

...So we have to solve:

$$a+c = 1$$ $$b-d = 3$$ $$a-c = 5$$ $$b+d = 7$$

This can be solved with linear algebra, for instance. We end up with:

$$a = 3$$ $$b = 5$$ $$c = -2$$ $$d = 2$$

...So our final equation is:

$$3\cdot\displaystyle\frac{(-1)^n+1}{2} + 5\cdot\displaystyle\frac{(-1)^{n+1}+1}{2} - 2 \cdot \cos{\left((e^{2 \pi i/4})^n \right)} + 2 \cdot \cos{\left((e^{2 \pi i/4})^{n+1} \right)}$$ $$ = 1,3,5,7,1,3,5,7,\dots$$

This formula has several advantages over taking the modulus. For instance, you can integrate it, since it's continuous. You can also take sums of it without much trouble. I hope this was intriguing!


This is a different question than your previous one. In this case, it appears the blocks are all four positions long. So I would say $2[n \pmod 4]+1$ if we start counting $n$ at zero.


Do you want the $\;n-$th term of the sequence $\;\{a_n\}=\{1,3,5,7,1,3,5,7,1\ldots\}\;$ , so

$$a_n=\begin{cases}1&,\;\;n=1\pmod 4\\3&,\;\;n=2\pmod 4\\5&,\;\;n=3\pmod 4\\7&,\;\;n=0\pmod 4\end{cases}$$


HINT: You probably won’t find a nice formula for the $n$-th term using familiar functions, but you can always use a multipart formula:

$$x_n=\begin{cases} 1,&\text{if }n=4k+1\text{ for some integer }k\\ 3,&\text{if }\ldots\\ 5,&\text{if }\ldots\\ 7,&\text{if }\ldots\\ \end{cases}$$

Can you fill in the other three cases?

Note that I’m assuming that you’re indexing the sequence as $\langle x_1,x_2,x_3,\ldots\rangle$; if you’re starting with $x_0$, as I normally would, the first condition would be $$\text{if }n=4k\text{ for some integer }k\;.$$

  • $\begingroup$ I believe that we can get a formula with the familiar functions; see my answer. $\endgroup$ – Matt Groff Sep 19 '13 at 15:32
  • $\begingroup$ @Matt: I know that we can, and I debated doing it; but as I said, most students probably won’t, unless they’ve seen the trick before. $\endgroup$ – Brian M. Scott Sep 19 '13 at 18:11
  • $\begingroup$ I believe I ended up doing a bad job explaining my answer anyways. It doesn't fit very easily into a short, concise answer. Not to conversate, but sorry if I came across as haughty. $\endgroup$ – Matt Groff Sep 20 '13 at 2:22
  • $\begingroup$ @Matt: No problem; my comment could have meant that I didn't think that there was such a formula. $\endgroup$ – Brian M. Scott Sep 20 '13 at 3:45

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.