A trigonometric approach to an identity by Ramanujan
C. Vignat · 2026
Read A trigonometric approach to an identity by Ramanujan by C. Vignat (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-07793.txt.
Introduction
Vignat writes an identity by Ramanujan using polar coordinates, with its proof reduced to checking an elementary trigonometric identity. Polar form describes points by distance from an origin and angle, rather than by sideways and up-down spots. The trig fact at the heart of the proof uses only basic sine and cosine links. This approach produces a few spinoffs on the original, each following the same polar route.
Catalog record
| CREATOR | C. Vignat |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 9365 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2603.07793
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
A trigonometric approach to an identity by Ramanujan
C. Vignat
Abstract
. An identity by Ramanujan is expressed using polar coordinates, so that its proof reduces to the verification of an elementary trigonometric identity. This approach produces a few variations on Ramanujan’s original identity.
1.
Introduction
Entry 45 in Chapter 23 of Ramanujan’s
4
th notebook
[
3
, p.102]
reads:
Proposition
.
Let
a,b,c
and
d
be any numbers such that
ad=bc.
Then
(1.1)
64\left\{\left(a+b+c\right)^{6}+\left(b+c+d\right)^{6}-\left(c+d+a\right)^{6}-\left(d+a+b\right)^{6}+\left(a-d\right)^{6}-\left(b-c\right)^{6}\right\}
\times\left\{\left(a+b+c\right)^{10}+\left(b+c+d\right)^{10}-\left(c+d+a\right)^{10}-\left(d+a+b\right)^{10}+\left(a-d\right)^{10}-\left(b-c\right)^{10}\right\}
=45\left\{\left(a+b+c\right)^{8}+\left(b+c+d\right)^{8}-\left(c+d+a\right)^{8}-\left(d+a+b\right)^{8}+\left(a-d\right)^{8}-\left(b-c\right)^{8}\right\}^{2}.
This identity is qualified by B. Berndt as ”one of the most fascinating finite identities we have ever seen”. As far as we know, two proofs of this identity are available, one by Berndt
and Bhargava
[
3
]
, and the other one by Nanjundiah
[
2
]
.
Berndt and Bhargava’s proof is deduced from a clever factorization
of a polynomial reparameterization of the identity. Nanjundiah’s proof
is the consequence of the properties of a degree 3 polynomial deduced
from the identity, and the identities satisfied by the powers of its
roots.
2.
A trigonometric approach
We adopt here a different approach based on the following elementary trigonometric
result.
Lemma 1
.
If
x_{1},y_{1},z_{1}
and
x_{2},y_{2},z_{2}
are real numbers
that satisfy the conditions
(2.1)
x_{i}+y_{i}+z_{i}=0,\,\,i=1,2
then there exists two radii
\rho_{1},\rho_{2}\geq 0
and two arguments
\theta_{1}
and
\theta_{2}
such that, for
i=1,2,
\begin{cases}x_{i}=&\rho_{i}\cos\theta_{i},\\
y_{i}=&\rho_{i}\cos\left(\theta_{i}-\frac{2\pi}{3}\right),\\
z_{i}=&\rho_{i}\cos\left(\theta_{i}+\frac{2\pi}{3}\right).\end{cases}
If additionally
(2.2)
x_{1}y_{1}+x_{1}z_{1}+y_{1}z_{1}=x_{2}y_{2}+x_{2}z_{2}+y_{2}z_{2},
then
\rho_{1}=\rho_{2}.
Notice that both parameterizations
(2.3)
\begin{cases}x_{1}=&b+c+d\\
y_{1}=&-a-b-c\\
z_{1}=&a-d\end{cases},\,\,\begin{cases}x_{2}=&a+c+d\\
y_{2}=&-a-b-d\\
z_{2}=&b-c\end{cases}
satisfy condition (
2.1
). If additionally
ad=bc
, they satisfy condition (
2.2
). Moreover, since identity (
1.1
)
is homogeneous in the variables
a,b,c,d
, we may assume
\rho=1.
With
n\geq 0,
define the functions
f_{n}\left(\theta\right)=\cos^{n}\theta+\cos^{n}\left(\theta-\frac{2\pi}{3}\right)+\cos^{n}\left(\theta+\frac{2\pi}{3}\right).
The proof of Ramanujan’s identity (
1.1
) is a consequence of the following Fourier series expansions, computed in Section 4.
Lemma 2
.
The following linearizations hold
f_{6}\left(\theta\right)=\frac{3}{32}\left(10+\cos\left(6\theta\right)\right),
f_{8}\left(\theta\right)=\frac{3}{128}\left(35+8\cos\left(6\theta\right)\right),
f_{10}\left(\theta\right)=\frac{27}{512}\left(14+5\cos\left(6\theta\right)\right).
As a consequence,
\left(f_{6}\left(\theta_{1}\right)-f_{6}\left(\theta_{2}\right)\right)\left(f_{10}\left(\theta_{1}\right)-f_{10}\left(\theta_{2}\right)\right)=\frac{405}{16384}\left(\cos\left(6\theta_{1}\right)-\cos\left(6\theta_{2}\right)\right)^{2}
and
\left(f_{8}\left(\theta_{1}\right)-f_{8}\left(\theta_{2}\right)\right)^{2}=\frac{9}{256}\left(\cos\left(6\theta_{1}\right)-\cos\left(6\theta_{2}\right)\right)^{2}
and the proof of Ramanuan’s identity (
1.1
) follows.
3.
Generalizations
Noticing that
f_{3}\left(\theta_{1}\right)=\frac{3}{4}\cos\left(3\theta_{1}\right),f_{5}\left(\theta_{1}\right)=\frac{15}{16}\cos\left(3\theta_{1}\right),f_{7}\left(\theta_{1}\right)=\frac{63}{64}\cos\left(3\theta_{1}\right),
the same arguments produce
Proposition
.
Without assuming
ad=bc,
25\left\{\left(b+c+d\right)^{3}-\left(a+b+c\right)^{3}+\left(a-d\right)^{3}\right\}\left\{\left(b+c+d\right)^{7}-\left(a+b+c\right)^{7}+\left(a-d\right)^{7}\right\}
=21\left\{\left(b+c+d\right)^{5}-\left(a+b+c\right)^{5}+\left(a-d\right)^{5}\right\}^{2}
which simplifies, replacing
b+c
by
b,
to
25\left\{\left(b+d\right)^{3}-\left(a+b\right)^{3}+\left(a-d\right)^{3}\right\}\left\{\left(b+d\right)^{7}-\left(a+b\right)^{7}+\left(a-d\right)^{7}\right\}
=21\left\{\left(b+d\right)^{5}-\left(a+b\right)^{5}+\left(a-d\right)^{5}\right\}^{2}.
This identity can be generalized to the case of two variables
\left(x_{i},y_{i},z_{i}\right),i=1,2,
noticing that
f_{3}\left(\theta_{1}\right)-f_{3}\left(\theta_{2}\right)=\frac{3}{4}\left[\cos\left(3\theta_{1}\right)-\cos\left(3\theta_{2}\right)\right]
and accordingly for the functions
f_{5}
and
f_{7},
to obtain
Proposition
.
Assume
ad-bc=0,
then
(3.1)
25\left\{\left(b+c+d\right)^{3}-\left(a+b+c\right)^{3}-\left(a+c+d\right)^{3}+\left(a+b+d\right)^{3}+\left(a-d\right)^{3}-\left(b-c\right)^{3}\right\}
\times\left\{\left(b+c+d\right)^{7}-\left(a+b+c\right)^{7}-\left(a+c+d\right)^{7}+\left(a+b+d\right)^{7}+\left(a-d\right)^{7}-\left(b-c\right)^{7}\right\}
=21\left\{\left(b+c+d\right)^{5}-\left(a+b+c\right)^{5}-\left(a+c+d\right)^{5}+\left(a+b+d\right)^{5}+\left(a-d\right)^{5}-\left(b-c\right)^{5}\right\}^{2}
Versions of these identities in the case
\rho_{1}\neq\rho_{2}
can be produced, to the price of
symmetry. For example:
Proposition
.
Assuming
ad=bc,
we have
\hskip-28.45274pt8\left(a^{2}+ab+b^{2}\right)\left(a^{2}+ac+c^{2}\right)\left\{\left(a+b+c\right)^{6}+\left(b+c+d\right)^{6}-\left(c+d+a\right)^{6}-\left(d+a+b\right)^{6}+\left(a-d\right)^{6}-\left(b-c\right)^{6}\right\}
=3a^{2}\left\{\left(a+b+c\right)^{8}+\left(b+c+d\right)^{8}-\left(c+d+a\right)^{8}-\left(d+a+b\right)^{8}+\left(a-d\right)^{8}-\left(b-c\right)^{8}\right\}
or equivalently
4\left\{\left(a+b+c\right)^{2}+\left(b+c+d\right)^{2}+\left(a-d\right)^{2}\right\}
\times\left\{\left(a+b+c\right)^{6}+\left(b+c+d\right)^{6}-\left(c+d+a\right)^{6}-\left(d+a+b\right)^{6}+\left(a-d\right)^{6}-\left(b-c\right)^{6}\right\}
=3\left\{\left(a+b+c\right)^{8}+\left(b+c+d\right)^{8}-\left(c+d+a\right)^{8}-\left(d+a+b\right)^{8}+\left(a-d\right)^{8}-\left(b-c\right)^{8}\right\}
The reason why parameterization (
2.3
) was chosen by Ramanujan still eludes us.
4.
Proofs of Lemmas 1 and 2
4.1.
Proof of Lemma
1
Denote
(x,y,z)
one of both triples
(x_{i},y_{i},z_{i}),\,\,i=1,2
and define
r^{2}=x^{2}+y^{2}+z^{2}.
Assuming
x+y+z=0
produces
\frac{r^{2}}{2}=\frac{x^{2}+y^{2}+z^{2}}{2}=-\left(xy+xz+yz\right)=-xy-z\left(x+y\right)=-xy+\left(x+y\right)^{2}=x^{2}+y^{2}+xy
so that
\frac{r^{2}}{2}=X^{2}+Y^{2}
with the new variables
X=x+\frac{y}{2},\,\,Y=\frac{\sqrt{3}}{2}y.
Define an argument
\alpha
such that
X=\frac{r}{\sqrt{2}}\cos\alpha,\,\,Y=\frac{r}{\sqrt{2}}\sin\alpha
and deduce
x=\frac{r}{\sqrt{2}}\cos\alpha-\frac{1}{2}r\sqrt{\frac{2}{3}}\sin\alpha,\,\,y=r\sqrt{\frac{2}{3}}\sin\alpha
that reduces to
x=r\sqrt{\frac{2}{3}}\cos\left(\alpha+\frac{\pi}{6}\right),y=r\sqrt{\frac{2}{3}}\cos\left(\alpha-\frac{\pi}{2}\right)
and produces
z=-x-y=r\sqrt{\frac{2}{3}}\cos\left(\alpha+\frac{5\pi}{6}\right).
Defining
\rho=r\sqrt{\frac{2}{3}},\,\,\theta=\alpha+\frac{\pi}{6}
produces the result.
4.2.
A trigonometric identity
Proposition
.
For two integers
p
and
N
, we have the linearization
\frac{1}{N}\sum^{N-1}_{k=0}\cos^{2p}\left(\theta+\frac{2k\pi}{N}\right)=\sum^{p}_{\underset{2m\equiv 0\mod N}{m=0}}a_{2m,p}\cos\left(2m\theta\right)
with the Fourier coefficients
a_{2m,p}=\begin{cases}\frac{\binom{2p}{p+m}}{2^{2p-1}},&m\neq 0,\,\,m\leq p\\
\frac{\binom{2p}{p}}{2^{2p}},&m=0.\end{cases}
In the case of an odd power,
\frac{1}{N}\sum^{N-1}_{k=0}\cos^{2p+1}\left(\theta+\frac{2k\pi}{N}\right)=\sum^{p}_{\underset{2m+1\equiv 0\mod N}{m=0}}a_{2m+1,p}\cos\left(\left(2m+1\right)\theta\right)
with
a_{2m+1,p}=\frac{\binom{2p+1}{p-m}}{2^{2p+1}},\,\,0\leq m\leq p.
Proof.
The left-hand side is a periodic function of
\theta
with period
T=2\pi
: its Fourier coefficients are obtained as
a_{n,p}=\frac{1}{2\pi}\int^{2\pi}_{0}\frac{1}{N}\sum^{N-1}_{k=0}\cos^{2p}\left(\theta+\frac{2k\pi}{N}\right)e^{-\imath n\theta}d\theta=\frac{1}{2\pi N}\sum^{N-1}_{k=0}\int^{2\pi}_{0}\cos^{2p}\left(\theta+\frac{2k\pi}{N}\right)e^{-\imath n\theta}d\theta.
Using the change of variable
\eta=\theta+\frac{2k\pi}{N}
and noticing
that, by periodicity, the domain of integration can be shifted back to
\left[0,2\pi\right],
we deduce
a_{n,p}=\frac{1}{2\pi N}\sum^{N-1}_{k=0}\int^{2\pi}_{0}\cos^{2p}\left(\eta\right)e^{-\imath n\left(\eta-\frac{2k\pi}{N}\right)}d\eta=\frac{1}{2\pi}\int^{2\pi}_{0}\cos^{2p}\left(\eta\right)e^{-\imath n\eta}\left(\frac{1}{N}\sum^{N-1}_{k=0}e^{\imath 2\pi\frac{kn}{N}}\right)d\eta.
The inner sum is evaluated as the indicator function
\sum^{N-1}_{k=0}e^{\imath 2\pi\frac{kn}{N}}=\begin{cases}0,&n\not\equiv 0\mod N\\
1,&n\equiv 0\mod N\end{cases}
In the case
n\equiv 0\mod N,
symmetry arguments show that the integral
vanishes for odd values of
m,
and, for even values of
n=2m,
a_{2m,p}=\frac{1}{2\pi}\int^{2\pi}_{0}\cos^{2p}\left(\eta\right)e^{-\imath 2m\eta}d\eta=\begin{cases}\frac{\binom{2p}{p+m}}{2^{2p-1}},&m\neq 0\\
\frac{\binom{2p}{p}}{2^{2p}},&m=0.\end{cases}
Notice that this result implies that
a_{2m,p}=0
for
m>p.
The odd power case is proved in the same way.