Novikov–Witt Algebras Overview

• Novikov–Witt algebras are Lie algebras emerging from the Krichever–Novikov framework of meromorphic vector fields on Riemann surfaces, generalizing the classical Witt algebra.
• They feature an almost-grading determined by a splitting of marked points, which supports a triangular decomposition and leads to geometric central extensions via contour integrals.
• A secondary usage defines Novikov–Witt algebras as Novikov algebras with a product whose commutator recovers the Witt bracket, with explicitly characterized derivations and quasi-derivations.

Searching arXiv for recent and foundational papers on Novikov–Witt / Krichever–Novikov algebras and related structures. Novikov–Witt algebras are Witt-type Lie algebras arising in the Krichever–Novikov framework of meromorphic vector fields on curves with prescribed singularities. In genus zero, the phrase refers to the multi-point Krichever–Novikov type vector field algebras, while in the two-point case they reduce to the classical Witt algebra; in higher-genus and multi-point settings, the ordinary grading is replaced by an almost-grading determined by a splitting of the marked points into in-points and out-points [(Schlichenmaier, 2015); (Schlichenmaier, 2013)]. The same framework admits Lie superalgebra extensions, geometric central extensions defined by contour integrals, and representation-theoretic structures such as triangular decompositions. In a different but related usage, “Novikov–Witt algebras” also denotes certain Novikov algebras whose commutator recovers the Witt bracket, together with admissible Novikov–Witt algebras constructed by Bai and collaborators; for these, derivations and quasi-derivations have been described explicitly (Kaygorodov et al., 14 Aug 2025).

1. Geometric definition and relation to the classical Witt algebra

In the Krichever–Novikov setting, one fixes a compact Riemann surface $\Sigma=E$ of genus $g$ and a finite set $A\subset \Sigma$ of marked points where poles are allowed. A splitting

$A=I\cup O$

into two disjoint nonempty subsets is fixed, with $I=\{P_1,\dots,P_K\}$ the in-points and $O=\{Q_1,\dots,Q_M\}$ the out-points, so that $N=\#A=K+M$ (Schlichenmaier, 2013). For each integer or half-integer $\lambda$, one considers the line bundle $K^\lambda$ and the space

$$F^\lambda:=\{\, f \text{ meromorphic section of } K^\lambda \text{ on } \Sigma \mid f \text{ holomorphic on } \Sigma\setminus A \,\}.$$

The vector field algebra is then

$g$0

the Lie algebra of meromorphic vector fields holomorphic off $g$1 (Schlichenmaier, 2013).

The total space

$g$2

carries both an associative multiplication induced by tensor product and a Lie bracket. Locally,

$g$3

which makes $g$4 a Poisson algebra (Schlichenmaier, 2013). Specializing to vector fields gives the usual bracket

$g$5

in local coordinates (Schlichenmaier, 2015).

In genus zero, these algebras are precisely the “$g$6-point Witt algebras” studied as multi-point versions of Krichever–Novikov type algebras. When $g$7 and the marked points are normalized to $g$8, the basis

$g$9

satisfies

$A\subset \Sigma$0

so the Novikov–Witt algebra becomes the classical Witt algebra (Schlichenmaier, 2015). This genus-zero two-point case is therefore the basic model from which the broader Krichever–Novikov picture generalizes.

A related algebraic-geometric formulation appears for affine curves $A\subset \Sigma$1, where the Krichever–Novikov algebra is

$A\subset \Sigma$2

In this language, the Witt algebra is the genus-zero case associated with $A\subset \Sigma$3, and $A\subset \Sigma$4 is the one-puncture affine counterpart (Buzaglo, 2021). This suggests that “Novikov–Witt” is used in the literature both for the geometric multi-point genus-zero Lie algebra and, more narrowly, for its affine genus-zero realizations.

2. Almost-grading, adapted bases, and triangular decomposition

A central structural feature of Krichever–Novikov and Novikov–Witt algebras is that the usual grading of the Witt algebra is replaced, in the multi-point or higher-genus setting, by an almost-grading. The splitting $A\subset \Sigma$5 determines this structure [(Schlichenmaier, 2013); (Schlichenmaier, 2015)]. For each weight $A\subset \Sigma$6, one has

$A\subset \Sigma$7

and each homogeneous component has dimension $A\subset \Sigma$8 (Schlichenmaier, 2013).

The adapted basis

$A\subset \Sigma$9

is characterized by prescribed orders at the in-points,

$A=I\cup O$0

together with normalization in chosen local coordinates,

$A=I\cup O$1

(Schlichenmaier, 2013). In genus zero with standard splitting $A=I\cup O$2 and $A=I\cup O$3, this basis becomes explicit. For affine coordinates $A=I\cup O$4 of the finite points, one sets

$A=I\cup O$5

and then

$A=I\cup O$6

for the vector field algebra (Schlichenmaier, 2015).

The defining property of almost-grading is the existence of finite bounds on degree shifts. In general,

$A=I\cup O$7

where $A=I\cup O$8 depend only on $A=I\cup O$9 (Schlichenmaier, 2013). For basis elements one has leading terms

$I=\{P_1,\dots,P_K\}$0

$I=\{P_1,\dots,P_K\}$1

(Schlichenmaier, 2013). In the genus-zero standard splitting, the explicit bounds are

$I=\{P_1,\dots,P_K\}$2

(Schlichenmaier, 2015).

Almost-grading induces a triangular decomposition

$I=\{P_1,\dots,P_K\}$3

with

$I=\{P_1,\dots,P_K\}$4

where $I=\{P_1,\dots,P_K\}$5 is the relevant upper bound (Schlichenmaier, 2015). In Schlichenmaier’s superalgebra notation this becomes

$I=\{P_1,\dots,P_K\}$6

with finite $I=\{P_1,\dots,P_K\}$7-part and infinite-dimensional $I=\{P_1,\dots,P_K\}$8-parts (Schlichenmaier, 2013). The significance of this decomposition is representation-theoretic: it supports highest-weight constructions, Fock space representations, and a compatible notion of creation and annihilation operators (Schlichenmaier, 2015).

Different splittings of the same set of marked points yield different almost-gradings and triangular decompositions (Schlichenmaier, 2015). This dependence is not accidental: locality of cocycles and compatibility of central extensions are defined with respect to the chosen almost-grading.

3. Krichever–Novikov pairings and geometric central extensions

The almost-graded structure is tied to a canonical bilinear pairing. For $I=\{P_1,\dots,P_K\}$9, the Krichever–Novikov pairing is

$O=\{Q_1,\dots,Q_M\}$0

where $O=\{Q_1,\dots,Q_M\}$1 is a separating cycle in $O=\{Q_1,\dots,Q_M\}$2, namely a smooth cycle separating $O=\{Q_1,\dots,Q_M\}$3 from $O=\{Q_1,\dots,Q_M\}$4 (Schlichenmaier, 2013). This pairing is nondegenerate and independent of the representative of the separating homology class, and the adapted basis is dual in the sense that

$O=\{Q_1,\dots,Q_M\}$5

(Schlichenmaier, 2013).

Central extensions are described by Lie algebra $O=\{Q_1,\dots,Q_M\}$6-cocycles. For a Lie algebra or Lie superalgebra $O=\{Q_1,\dots,Q_M\}$7, a central extension is given by adjoining a central element $O=\{Q_1,\dots,Q_M\}$8 and defining

$O=\{Q_1,\dots,Q_M\}$9

where $N=\#A=K+M$0 is a $N=\#A=K+M$1-cocycle (Schlichenmaier, 2013). In the Krichever–Novikov setting, these cocycles are geometric: they arise from contour integrals of globally defined differentials associated with vector fields and, in the super case, half-forms.

For the vector field algebra $N=\#A=K+M$2, fixing a holomorphic projective connection $N=\#A=K+M$3 and a closed curve $N=\#A=K+M$4, one defines

$N=\#A=K+M$5

(Schlichenmaier, 2013). In the genus-zero formulation of the same construction, the differential is written as

$N=\#A=K+M$6

(Schlichenmaier, 2015). Different projective connections give cohomologous cocycles, so the cohomology class depends on the cycle rather than the auxiliary choice of $N=\#A=K+M$7 [(Schlichenmaier, 2013); (Schlichenmaier, 2015)].

Locality is the condition that aligns central extensions with the almost-grading. A cocycle is local if there exist bounds $N=\#A=K+M$8 such that

$N=\#A=K+M$9

(Schlichenmaier, 2013). A cocycle is bounded from above if $\lambda$0 whenever $\lambda$1 for some $\lambda$2 (Schlichenmaier, 2013). Integrating over a separating cycle yields a local class, while integrating over a small circle around a marked point yields a class that is bounded from above but in general not local with respect to the same splitting [(Schlichenmaier, 2013); (Schlichenmaier, 2015)].

This geometric distinction leads to the basic classification results. For the vector field algebra $\lambda$3, the space of bounded cohomology classes has dimension $\lambda$4, with basis given by cocycles integrated around the in-point circles $\lambda$5, whereas the space of local classes is one-dimensional and generated by the separating-cycle class $\lambda$6 (Schlichenmaier, 2013). In genus zero, the universal central extension has center of dimension $\lambda$7, corresponding to

$\lambda$8

and is generated by residue cocycles around punctures (Schlichenmaier, 2015). Only one of these classes is local for a fixed splitting, and that class plays the role of the Virasoro cocycle (Schlichenmaier, 2015).

4. Lie superalgebras of Novikov–Witt type

The Krichever–Novikov formalism extends from the vector field algebra to a Lie superalgebra

$\lambda$9

where the odd part consists of meromorphic half-forms holomorphic outside $K^\lambda$0 (Schlichenmaier, 2013). The brackets are defined by extending the Lie bracket on $K^\lambda$1, the Lie derivative action of vector fields on half-forms, and the product of half-forms: $K^\lambda$2 (Schlichenmaier, 2013). This structure satisfies the super-Jacobi identity and is called the Krichever–Novikov Lie superalgebra (Schlichenmaier, 2013).

With the homogeneous bases

$K^\lambda$3

the leading terms of the super-brackets are

$K^\lambda$4

$K^\lambda$5

$K^\lambda$6

where the higher-degree terms are bounded in the sense of the almost-grading (Schlichenmaier, 2013). The even part carries integer degrees and the odd part half-integer degrees (Schlichenmaier, 2013).

The superalgebra admits an even central extension given by the same vector-field cocycle on $K^\lambda$7 together with an odd-odd part

$K^\lambda$8

(Schlichenmaier, 2013). In the genus-zero exposition, this is written as

$K^\lambda$9

(Schlichenmaier, 2015). The difference in normalization reflects the conventions adopted in the two sources.

The classification parallels the vector field case. For the superalgebra $$F^\lambda:=\{\, f \text{ meromorphic section of } K^\lambda \text{ on } \Sigma \mid f \text{ holomorphic on } \Sigma\setminus A \,\}.$$0 with even central element, the space of bounded cohomology classes has dimension $$F^\lambda:=\{\, f \text{ meromorphic section of } K^\lambda \text{ on } \Sigma \mid f \text{ holomorphic on } \Sigma\setminus A \,\}.$$1, the space of local classes is one-dimensional, and for a fixed splitting there is, up to equivalence and rescaling, a unique nontrivial almost-graded central extension represented by the separating-cycle cocycle $$F^\lambda:=\{\, f \text{ meromorphic section of } K^\lambda \text{ on } \Sigma \mid f \text{ holomorphic on } \Sigma\setminus A \,\}.$$2 (Schlichenmaier, 2013). By contrast, every bounded cocycle defining a central extension with odd central element is a coboundary, so all such extensions split (Schlichenmaier, 2013).

In the classical genus-zero two-point case, the construction recovers the Neveu–Schwarz superalgebra. With

$$F^\lambda:=\{\, f \text{ meromorphic section of } K^\lambda \text{ on } \Sigma \mid f \text{ holomorphic on } \Sigma\setminus A \,\}.$$3

the central extension specializes to

$$F^\lambda:=\{\, f \text{ meromorphic section of } K^\lambda \text{ on } \Sigma \mid f \text{ holomorphic on } \Sigma\setminus A \,\}.$$4

$$F^\lambda:=\{\, f \text{ meromorphic section of } K^\lambda \text{ on } \Sigma \mid f \text{ holomorphic on } \Sigma\setminus A \,\}.$$5

together with the corresponding odd-odd relation, whose precise central term depends on normalization (Schlichenmaier, 2013). This is the super-analogue of the Witt-to-Virasoro passage.

5. Explicit genus-zero multi-point cases

The genus-zero case on $$F^\lambda:=\{\, f \text{ meromorphic section of } K^\lambda \text{ on } \Sigma \mid f \text{ holomorphic on } \Sigma\setminus A \,\}.$$6 provides the most explicit realization of Novikov–Witt algebras (Schlichenmaier, 2015). Fix marked points

$$F^\lambda:=\{\, f \text{ meromorphic section of } K^\lambda \text{ on } \Sigma \mid f \text{ holomorphic on } \Sigma\setminus A \,\}.$$7

normalize $$F^\lambda:=\{\, f \text{ meromorphic section of } K^\lambda \text{ on } \Sigma \mid f \text{ holomorphic on } \Sigma\setminus A \,\}.$$8, and choose a splitting into finite in-points and the out-point at infinity. The vector field algebra then consists of meromorphic vector fields with poles only at $$F^\lambda:=\{\, f \text{ meromorphic section of } K^\lambda \text{ on } \Sigma \mid f \text{ holomorphic on } \Sigma\setminus A \,\}.$$9.

For $g$00, the almost-grading is an honest grading and the universal central extension is the Virasoro algebra. Integrating the geometric cocycle over a small circle around $g$01 yields

$g$02

so

$g$03

(Schlichenmaier, 2015). This is exactly the classical Virasoro relation.

For $g$04, one obtains the standard three-point example with

$g$05

An adapted basis of functions is

$g$06

satisfying

$g$07

(Schlichenmaier, 2015). The corresponding vector fields are

$g$08

with Lie brackets

$g$09

$g$10

$g$11

(Schlichenmaier, 2015). These formulas display the almost-grading explicitly: only a finite band of higher degrees can occur, and in this case $g$12 (Schlichenmaier, 2015).

The same three-point example illustrates the distinction between local and merely bounded cocycles. The two independent cocycles may be taken as residues at $g$13 and at $g$14: $g$15 (Schlichenmaier, 2015). Their explicit values on the basis are given in detail in the source, and only $g$16 is local with respect to the chosen splitting; $g$17 is bounded from above but not local (Schlichenmaier, 2015). Accordingly, the central extension remains compatible with the almost-grading precisely when the nonlocal class is excluded (Schlichenmaier, 2015).

These genus-zero formulas are significant because they show that the passage from Witt/Virasoro to multi-point Novikov–Witt algebras does not merely add punctures: it changes grading into almost-grading, enlarges the space of geometric cocycles, and makes the choice of splitting structurally decisive.

6. Alternative Novikov–Witt usage: Novikov products, admissible variants, and quasi-derivations

A different but established usage of “Novikov–Witt algebra” concerns Novikov algebras whose commutator Lie algebra is the Witt algebra (Kaygorodov et al., 14 Aug 2025). In this setting, one works not with meromorphic vector fields directly but with a Novikov product $g$18 satisfying the left-symmetric and right-commutative identities: $g$19

$g$20

(Kaygorodov et al., 14 Aug 2025).

The Novikov–Witt algebras $g$21 have basis $g$22, parameters $g$23, $g$24, $g$25, and product

$g$26

Their commutator reproduces the Witt bracket: $g$27 (Kaygorodov et al., 14 Aug 2025). Admissible Novikov–Witt algebras $g$28 are defined by

$g$29

again with

$g$30

(Kaygorodov et al., 14 Aug 2025).

For these Novikov algebras, derivations, $g$31-derivations, and quasi-derivations are parameter dependent. For $g$32, the derivation algebra is

$g$33

with

$g$34

(Kaygorodov et al., 14 Aug 2025). In all cases,

$g$35

(Kaygorodov et al., 14 Aug 2025). The quasi-derivations decompose as

$g$36

(Kaygorodov et al., 14 Aug 2025).

When $g$37, a distinguished family is

$g$38

while for $g$39 the quasi-derivations are described by finitely supported parameter families $g$40 via explicit formulas (Kaygorodov et al., 14 Aug 2025). For admissible Novikov–Witt algebras $g$41, one likewise has

$g$42

with

$g$43

and

$g$44

(Kaygorodov et al., 14 Aug 2025). The quasi-derivations again split as derivations, $g$45-derivations, and an explicit family $g$46.

This second usage should not be conflated with the Krichever–Novikov vector field algebra itself. The former concerns Novikov algebra structures whose commutator is Witt, whereas the latter concerns Lie algebras of meromorphic vector fields on punctured curves. The connection is that both recover the Witt Lie bracket, but they belong to different algebraic categories.

7. Structural consequences, representation theory, and algebraic limits

The almost-grading and locality of cocycles have direct implications for representation theory. In the genus-zero multi-point setting, the triangular decomposition supports highest-weight modules, semi-infinite wedge representations, and creation/annihilation operator formalisms compatible with the chosen splitting (Schlichenmaier, 2015). For affine Krichever–Novikov algebras, the Sugawara construction yields an energy–momentum representation of the centrally extended vector field algebra provided the level is non-critical, and the resulting stress–energy field implements a projective action that becomes a genuine Lie representation after passing to the appropriate central extension (Schlichenmaier, 2015). A plausible implication is that the unique local central extension for a fixed splitting plays, in the multi-point setting, the same organizing role that the Virasoro extension plays in the classical two-point theory.

The framework also encompasses related algebras beyond vector fields. The genus-zero Krichever–Novikov theory treats the algebra of functions, differential operators, current algebras, affine Lie algebras, and Lie superalgebras, all with analogous almost-graded structures and geometric cocycles (Schlichenmaier, 2015). For functions, the geometric cocycles are

$g$47

and in genus zero every $g$48-invariant cocycle is a linear combination of these residue cocycles (Schlichenmaier, 2015). For differential operators $g$49, there is also a mixing cocycle

$g$50

with an affine connection $g$51 (Schlichenmaier, 2015).

From a ring-theoretic viewpoint, Novikov–Witt and Krichever–Novikov algebras exhibit a strong negative property: the universal enveloping algebra of any Krichever–Novikov algebra is not noetherian (Buzaglo, 2021). More precisely, for any affine curve $g$52,

$g$53

is not noetherian, extending the known result for the Witt algebra (Buzaglo, 2021). The proof proceeds by embedding suitable subalgebras of $g$54 into $g$55; since the enveloping algebras of these subalgebras are already non-noetherian, the same holds for the full Krichever–Novikov algebra (Buzaglo, 2021). The same conclusion extends to the KN-type algebras

$g$56

for finite-dimensional $g$57 (Buzaglo, 2021).

This non-noetherianity does not contradict the representation-theoretic usefulness of Novikov–Witt algebras; rather, it marks a limit of classical ring-theoretic techniques. The sources explicitly note that non-noetherian enveloping algebras complicate finite-generation arguments for ideals and annihilators and indicate that many infinite-dimensional Lie algebras of geometric origin resist standard noetherian machinery (Buzaglo, 2021).

Taken together, these results place Novikov–Witt algebras at a confluence of several themes: generalization of the Witt and Virasoro algebras to curves with multiple punctures and higher genus, geometric classification of central extensions via contour integrals and homology classes, superalgebra extensions of Neveu–Schwarz type, and, in a separate Novikov-algebra sense, parameter-dependent derivation and quasi-derivation theories [(Schlichenmaier, 2013); (Schlichenmaier, 2015); (Kaygorodov et al., 14 Aug 2025)]. The unifying principle is that the Witt bracket persists across these settings, while grading, cohomology, and deformation behavior become sensitive to the underlying geometry or to the chosen Novikov product.