The θ \theta invariant recovers the Rozansky-Overbay invariant
Ramana Murugesan · 2026
Read The θ \theta invariant recovers the Rozansky-Overbay invariant by Ramana Murugesan (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-2604-27029.txt.
Introduction
Murugesan shows that the theta invariant generalizes the Rozansky-Overbay invariant, so theta covers the cases of the older one and more. This move places the older one inside a wider framework, with theta as the more general tool. The paper's contribution is this link itself: one extending another, joining their scope. Readers working with either gain a single, broader lens for the same cases.
Catalog record
| CREATOR | Ramana Murugesan |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 5226 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2604.27029
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
The θ \theta invariant recovers the Rozansky-Overbay invariant
Ramana Murugesan
Abstract
Let
K
be a knot.
[
1
]
wrote formulas for the Rozansky-Overbay knot
invariant
\rho_{1}(K)\in\mathbf{C}[T^{\pm 1}]
defined in
[
4
,
5
,
6
,
3
]
.
Later,
[
2
]
define the invariant
\theta(K)\in\mathbf{C}[T_{1}^{\pm 1},T_{2}^{\pm 1}]
.
In this paper we show that the
\theta
invariant is stronger than the Rozansky-Overbay invariant,
answering
[
2
, Conjecture 24]
:
Theorem 1
.
Let
K
be a knot. Then
\rho_{1}(K)=-\theta(K)|_{T_{1}\leftarrow 1,T_{2}\leftarrow T}
We begin by recalling one definition of the Alexander polynomial given in
[
1
]
.
Consider
an upright knot diagram
D
, which is an oriented long knot diagram so that at each crossing,
both overlapping strands point upwards. Every knot admits such a projection, since
crossings can be rotated to make them upright. Suppose we have such a diagram with
n
crossings. We start at the initial edge and label the edges
k=1,\ldots,2n+1
along
the orientation of the knot.
For each edge
k
, its
geometric rotation number
\varphi_{k}
is defined as the signed
number of times that the tangent to
k
is horizontal, where cups have positive sign and
caps have negative sign.
Then, we denote each crossing
c
by the tuple
(s,i,j)
consisting of
the sign of
c
, the incoming
overcrossing edge, and the incoming undercrossing edge respectively, and
define
X=\{c=(s,i,j):\text{ $c$ crossing}\}
. Letting
T
be an indeterminate,
define the
(2n+1)\times(2n+1)
matrix of Laurent polynomials
(1)
A=I-\sum_{c\in X}(T^{s}E_{i,i+1}+(1-T^{s})E_{i,j+1}+E_{j,j+1})
where
E_{pq}
is the elementary matrix with entry
1
at
(p,q)
and
0
elsewhere.
Note that the normalized Alexander
polynomial is given by
\Delta(K):=T^{(-\varphi(D)-w(D))/2}\det(A)
where
\varphi(D)=\sum_{k}\varphi_{k}
and
w(D)=\sum_{c}s_{c}
(where
s_{c}
is the sign of
c
.)
Now we let
G=(g_{\alpha\beta}):=A^{-1}
. For a crossing
c
define
R_{1}(c)=s(g_{ji}(g_{j+1,j}+g_{j,j+1}-g_{ij})-g_{ii}(g_{j,j+1}-1)-1/2)
Then
\rho_{1}(D)\in\mathbf{C}[T^{\pm 1}]
is defined by
\rho_{1}(D):=\Delta^{2}\left(\sum_{c}R_{1}(c)-\sum_{k}\varphi_{k}(g_{kk}-1/2)\right)
It is shown in
[
1
]
that this is invariant under the Reidemeister-type moves of
an upright knot diagram and so this defines a knot invariant
\rho_{1}(K)
.
Now to define the
\theta
invariant, take further
indeterminates
T_{1},T_{2}
, and
T_{3}=T_{1}T_{2}
and let
\Delta_{i}:=\Delta|_{T\leftarrow T_{i}}
and
G_{i}=(g_{i\alpha\beta}):=G|_{T\leftarrow T_{i}}
. For all crossings
c,c_{0},c_{1}
, and edges
k
, define
\displaystyle F_{1}(c)
\displaystyle=\ s\left[1/2-g_{3ii}+T_{2}^{s}g_{1ii}g_{2ji}-T_{2}^{s}g_{3jj}g_{2ji}-(T_{2}^{s}-1)g_{3ii}g_{2ji}\right.
\displaystyle\quad\left.+(T_{3}^{s}-1)g_{2ji}g_{3ji}-g_{1ii}g_{2jj}+2g_{3ii}g_{2jj}+g_{1ii}g_{3jj}-g_{2ii}g_{3jj}\right]
\displaystyle\quad+\frac{s}{T_{2}^{s}-1}[(T_{1}^{s}-1)T_{2}^{s}\left(g_{3jj}g_{1ji}-g_{2jj}g_{1ji}+T_{2}^{s}g_{1ji}g_{2ji}\right)
\displaystyle\quad+(T_{3}^{s}-1)g_{3ji}(1-T_{2}^{s}g_{1ii}+g_{2ij}+(T_{2}^{s}-2)g_{2jj}(T_{1}^{s}-1)(T_{2}^{s}+1)g_{1ji}]
\displaystyle F_{2}(c_{0},c_{1})
\displaystyle=\frac{s_{1}(T_{1}^{s_{0}}-1)(T_{3}^{s_{1}}-1)g_{1j_{1}i_{0}}g_{3j_{0}i_{1}}}{T_{2}^{s_{1}}-1}(T_{2}^{s_{0}}g_{2i_{1}i_{0}}+g_{2j_{1}j_{0}}-T_{2}^{s_{0}}g_{2j_{1}i_{0}}-g_{2i_{1}j_{0}})
\displaystyle F_{3}(k)
\displaystyle=\varphi_{k}(g_{3kk}-1/2)
Then
\theta(D)\in\mathbf{C}[T_{1}^{\pm 1},T_{2}^{\pm 1}]
is defined by
\theta(D):=\Delta_{1}\Delta_{2}\Delta_{3}\left(\sum_{c}F_{1}(c)+\sum_{c_{0},c_{1}}F_{2}(c_{0},c_{1})+\sum_{k}F_{3}(k)\right)
Although it is only obvious from the formula that
\theta(D)
is a rational
function,
[
2
]
show that the denominators of the form
(T_{2}^{s}-1)
in
F_{2}
vanish in the summation above, so
\theta(D)
is a Laurent polynomial. They also show
that
\theta(D)
is invariant under Reidemeister-type moves, and so defines a knot invariant
\theta(K)
.
We now prove the theorem.
Proof of Theorem
1
.
Let
D
be a upright knot diagram representing
K
.
Since
T_{3}=T_{1}T_{2}=T
we have
\displaystyle g_{ii}
\displaystyle=g_{2ii}=g_{3ii}
\displaystyle\Delta
\displaystyle=\Delta_{2}=\Delta_{3}
\displaystyle\Delta_{1}
\displaystyle=\Delta|_{T\leftarrow 1}=1
where the last equality comes from the normalization of
\Delta
. Since
F_{2}\equiv 0
and
F_{3}(k)=\varphi_{k}(g_{kk}-1/2)
it suffices to show that
R_{1}(c)=-F_{1}(c)
for each
c=(s,i,j)
.
With the above equations,
F_{1}(c)=\frac{s}{2}[1+2(-1+T^{s})g^{2}_{ji}+2g_{ji}(1+g_{ij}-2g_{jj})+2g_{ii}(-1-(-1+T^{s})g_{ji}+g_{jj})]
To show that this equals
-R_{1}(c)
, we need to eliminate the
g_{j,j+1}
and
g_{j+1,j}
terms in
R_{1}(c)
.
Recall that
G=A^{-1}
. Multiplying (
1
) by
G
,
\displaystyle 1
\displaystyle=(AG)_{jj}=g_{jj}-g_{j+1,j}
\displaystyle 0
\displaystyle=(GA)_{j,j+1}=g_{j,j+1}-(1-T^{s})g_{ji}-g_{jj}
Thus:
\displaystyle R_{1}(c)
\displaystyle=s\left[\vphantom{\frac{1}{2}}g_{ji}((g_{jj}-1)+(g_{jj}+(1-T^{s})g_{ji})-g_{ij})\right.
\displaystyle\quad-g_{ii}(g_{jj}\left.+(1-T^{s})g_{ji}-1)-\frac{1}{2}\right]
\displaystyle=\frac{s}{2}[-1+2(1-T^{s})g_{ji}^{2}+2g_{ii}(-g_{jj}-(1-T^{s})g_{ji}+1)
\displaystyle\quad+2g_{ji}(2g_{jj}-1-g_{ij})]
\displaystyle=-F_{1}(c)