The ring of differential operators on a nodal curve is not a bialgebroid

Myriam Mahaman · 2026

Read The ring of differential operators on a nodal curve is not a bialgebroid by Myriam Mahaman (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-2605-18568.txt.

Introduction

Myriam Mahaman shows in a 2026 note that the ring of differential operators on a nodal curve is neither locally projective nor a bialgebroid. An earlier paper proved local projectivity suffices for a bialgebroid structure on such rings over an affine variety. Using elementary methods, this note closes the nodal-curve case in the negative on both counts. The result separates nodal curves from varieties where the bialgebroid structure holds.

Catalog record

CREATORMyriam Mahaman
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE8084
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2605.18568

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

The ring of differential operators on a nodal curve is not a bialgebroid
Myriam Mahaman

Abstract
. In a previous article, we showed that local projectivity is a sufficient condition for the existence of a bialgebroid structure on the ring of differential operators on an affine variety. In this note, we show using elementary methods that the ring of differential operators on a nodal curve is neither locally projective nor does it admit a bialgebroid structure.



1991 Mathematics Subject Classification
16S32, 16T15

1.
Introduction

Let
A
be a commutative algebra over a commutative ring
k
,
and
\mathscr{D}_{A}
be its ring of
k
-linear differential operators
in the sense of Grothendieck

[
1
]
.
In the article
[
2
]
,
we raised the following question:

Question
.

For which algebras
A
is
\mathscr{D}_{A}
a Hopf algebroid?

This question was motivated by Sweedler’s results
in
[
4
]
,
where he showed that

\mathscr{D}_{A}
is a Hopf algebroid
when
A
is the coordinate ring of a smooth
affine variety.
In our work
[
2
]

we identified a more general property,
called
R
-
local projectivity

[
5
]
,
which yields a positive
answer to our question.
Furthermore, we showed that
this property is satisfied when
A

is the coordinate ring of a monomial curve,
and thereby providing
the first examples of non-smooth
varieties whose rings of differential operators
are Hopf algebroids.

When the assumption about local projectivity is removed,
the answer to our question remains unclear.
In this short note, we provide the first
explicit examples for which
there is no bialgebroid structure:
we take
A
to be
the coordinate ring of an affine nodal
curve, as studied in
[
3
]
,
and using elementary methods, we show not only that
there is no bialgebroid structure on
\mathscr{D}_{A}
,
but also that
\mathscr{D}_{A}
is not

R(A)
-locally projective.

This note is structured as follows:
we introduce our examples in Section
2
,
then we show that
\mathscr{D}_{A}
is not
R(A)
-locally projective
in Section
3
,
and finally, we show that

\mathscr{D}_{A}
does not admit a bialgebroid
structure in Section
4
.

Acknowledgements

The author thanks Ulrich Krähmer for helpful comments
on an earlier version of this manuscript.
This work was supported by the GAČR/NCN grant

Quantum Geometric Representation Theory and
Non-commutative Fibrations
24-11728K.

2.
Nodal curves

We fix our notation for the remainder of this paper:
let
k
be a field of characteristic zero;
we denote
\otimes\coloneqq\otimes_{k}
and

\operatorname{Hom}\coloneqq\operatorname{Hom}_{k}
.
Let
B\coloneqq k[t]
,

\mathscr{D}_{B}\coloneqq k\langle t,\partial\rangle/(\partial t-t\partial-1)

be its ring of
k
-linear differential operators,

I\coloneqq f\,k[t]
be an ideal of
B
such that

f=f_{1}\cdots f_{r}
for some
irreducible polynomials

f_{1},\ldots,f_{r}\in B
(
r\geq 2
)
which are pairwise coprime.
We consider the subalgebra
A\coloneqq k+I
of
B

and its ring of differential operators
\mathscr{D}_{A}
.
The following result was shown in

[
3
]
:

Lemma 2.1
.

\mathscr{D}_{A}=k+I\mathscr{D}_{B}\subseteq\mathscr{D}_{B}
.

Proof.

See
[
3
, Proposition 4.4]
.
∎

We will use the following properties to show our main results:

Lemma 2.2
.

(1)

For all
D\in\mathscr{D}_{A}
, we have
D(I)\subseteq I
.

(2)

There exists
D\in\mathscr{D}_{A}
such that

D(A)\subseteq I
and
D(I)\not\subseteq I^{2}
.

(3)

There exists
D\in\mathscr{D}_{A}
such that

D(I^{2})\not\subseteq I^{2}
.

Proof.

(1)

Let
D\in\mathscr{D}_{A}
.
There exists
\lambda\in k

and
D^{\prime}\in\mathscr{D}_{B}

such that
D=\lambda+fD^{\prime}
,
hence
D(a)=\lambda a+fD^{\prime}(a)\in I

for any
a\in I
.

(2)

Let
g\in I
.
There exists
D^{\prime}\in\mathscr{D}_{B}

such that
D^{\prime}(1)=0
and
D^{\prime}(g)=1
.
Then
D=fD^{\prime}\in\mathscr{D}_{A}
,
and we have that

D(A)\subseteq I

and
D(g)=f\not\in I^{2}
.

(3)

Let
g\in I^{2}
.
There exists
D^{\prime}\in\mathscr{D}_{B}

such that
D^{\prime}(g)=1
.
Then
D=fD^{\prime}\in\mathscr{D}_{A}
,
and we have
D(g)=f\not\in I^{2}
.
∎

Example 2.3
.

We assume
A
to be the coordinate ring of
the nodal cubic
y^{2}=x^{2}(x+1)
.
Under the identifications
x\mapsto t^{2}-1

and

y\mapsto t(t^{2}-1)
, we obtain
that
A=k+I

where
I=(t^{2}-1)k[t]
.
Straightforward computations show that
the operator

(t^{2}-1)\frac{\mathrm{d}}{\mathrm{d}t}

satisfies the condition (2)
in Lemma
2.2
,
and that
(t^{2}-1)\frac{\mathrm{d}^{2}}{\mathrm{d}t^{2}}

satisfies condition (3).

3.
Local projectivity

The notion of
R
-local projectivity
was originally introduced
in
[
5
]

for an arbitrary
A
-module
M
.
Let us formulate how it looks
for the case
M=\mathscr{D}_{A}
:
we consider the map

\Psi\colon\mathscr{D}_{A}\to\operatorname{Hom}(A\otimes\mathscr{D}_{A},\mathscr{D}_{A}),\quad D\mapsto\Psi_{D},

where
\Psi_{D}(a\otimes D^{\prime})=D(a)D^{\prime}
.

Definition 3.1
.

We say that
\mathscr{D}_{A}
is

R(A)
-locally projective

if for all
D\in\mathscr{D}_{A}
, we have
D\in\operatorname{Im}(\Psi_{D})
,
i.e. there exist
a_{1},\ldots,a_{n}\in A

and
D_{1},\ldots,D_{n}\in\mathscr{D}_{A}
such that

D=\sum_{i=1}^{n}D(a_{i})D_{i}.

Lemma 3.2
.

Let
D\in\mathscr{D}_{A}
such that
D(A)\subseteq I
.
Then
E(I)\subseteq I^{2}
for all
E\in\operatorname{Im}(\Psi_{D})
.

Proof.

Let
a\in A
and
D^{\prime}\in\mathscr{D}_{A}
.
We have
\Psi_{D}(a\otimes D^{\prime})=D(a)D^{\prime}
,
where
D(a)\in I
and
D^{\prime}(I)\subseteq I
,
hence
D(a)D^{\prime}(I)\subseteq I^{2}

and the result follows.
∎

Theorem 3.3
.

\mathscr{D}_{A}
is not
R(A)
-locally projective.

Proof.

We know from Lemma
2.2
(2)
that there exists

D\in\mathscr{D}_{A}
such that
D(A)\subseteq I

and
D(I)\not\subseteq I^{2}
.
Then it follows from Lemma
3.2

that
D\not\in\operatorname{Im}(\Psi_{D})
,
hence
\mathscr{D}_{A}
is not
R(A)
-locally projective.
∎

4.
Bialgebroid structure

Let us show that there can be no bialgebroid
structure on
\mathscr{D}_{A}
.
Here
\mathscr{D}_{A}
is endowed with the canonical counit

\varepsilon\colon\mathscr{D}_{A}\to A
,
D\mapsto D(1)
,
and we show that there can be no associated
comultiplication.
See
[
2
]

for the full definition.

We consider the maps

\displaystyle\mu_{*}\colon

\displaystyle\mathscr{D}_{A}\otimes_{A}\mathscr{D}_{A}\to\operatorname{Hom}(A\otimes A,A),

\displaystyle D_{1}\otimes_{A}D_{2}

\displaystyle\mapsto\big(a\otimes b\mapsto D_{1}(a)D_{2}(b)\big),

\displaystyle\mu^{*}\colon

\displaystyle\mathscr{D}_{A}\to\operatorname{Hom}(A\otimes A,A),

\displaystyle D

\displaystyle\mapsto\big(a\otimes b\mapsto D(ab)\big),

where the tensor product
\mathscr{D}_{A}\otimes_{A}\mathscr{D}_{A}

is obtained via
the left
A
-module structure of
\mathscr{D}_{A}
.

Lemma 4.1
.

There can be no map
\Delta\colon\mathscr{D}_{A}\to\mathscr{D}_{A}\otimes_{A}\mathscr{D}_{A}

such that

\mu^{*}(D)=\mu_{*}(\Delta(D))

for all
D\in\mathscr{D}_{A}
.

Proof.

On the one hand, we have
\varphi(I\otimes I)\subseteq I^{2}

for all
\varphi\in\operatorname{Im}\mu_{*}
.
Indeed, for any
D_{1},D_{2}\in\mathscr{D}_{A}
,
we have
D_{i}(I)\subseteq I
for
i=1,2
,
therefore

\mu_{*}(D_{1}\otimes_{A}D_{2})(a\otimes b)=D_{1}(a)D_{2}(b)\in I^{2}

for all
a,b\in I
,
hence
\mu_{*}(D_{1}\otimes_{A}D_{2})(I\otimes I)\subseteq I^{2}
.

On the other hand, we know from
Lemma
2.2
(3)
that there exists
D\in\mathscr{D}_{A}

such that
D(I^{2})\not\subseteq I^{2}
.
It follows that

\mu^{*}(D)(I\otimes I)\not\subseteq I^{2}
,
hence
\mu^{*}(D)\not\in\operatorname{Im}\mu_{*}
,
and there can be no map
\Delta

satisfying the given property.
∎

Theorem 4.2
.

There is no bialgebroid structure
on
\mathscr{D}_{A}
for which the map
D\mapsto D(1)

is the counit.

Proof.

The axioms defining a bialgebroid require
the comultiplication
\Delta\colon\mathscr{D}_{A}\to\mathscr{D}_{A}\otimes_{A}\mathscr{D}_{A}

to satisfy the property in Lemma
4.1
;
however, as we have shown, there can be no such map.
∎

Remark 4.3
.

It remains unclear whether in general
R
-local projectivity
is necessary for the existence of a bialgebroid structure,
since the arguments we used to disprove

Read the complete text