Koszul Duality for Symmetric Algebras and Derived Loop Spaces

arXiv:2609.18115 · math.AG · Submitted 2026-09-16 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Mathematics. The day in mathematics: new theorems, new proofs, and what they open.

Ada: I'm Ada.

Emmy: And I'm Emmy, here with Kolya.

Ada: We're looking at the paper: "Koszul Duality for Symmetric Algebras and Derived Loop Spaces".

Emmy: Here's the gist of the paper. In this paper, Koszul duality is proven for symmetric algebras of connective complexes consisting of locally free sheaves of finite rank, which yields an equivalence between derived categories and provides a generalization to derived loop spaces Koszul Duality Equivalence The authors prove Koszul duality for symmetric algebras of complexes by using Mirkovi´c and Riche’s Koszul duality for symmetric algebras of complexes consisting of locally free sheaves on a scheme The result is formulated as follows: Let k be a field and let X be a Noetherian scheme over k Consider a complex X:= (0 → E−n → · · · → E−1 → E0 → 0) consisting of locally free sheaves of finite rank, where Ei has cohomological degree i Then we take the symmetric algebra Sym X of X This is a sheaf of dg-algebras on X.

Ada: Besides, we define an additional grading on B, other than cohomological one, by considering each Ei to have degree 1 We call this internal grading.

Emmy: Then the Koszul dual A of B is given by A:= Sym(X ∨

−1: ) = Sym(0 → (E 0) ∨ → (E − 1) ∨ → · · · → (E − n) ∨ → 0), where ∨ means dual as OX-modules, and each (E − i) ∨ has cohomological degree i + 1 and internal degree −1.

Ada: Under this setting, the Koszul duality is stated as follows: There is an equivalence of triangulated categories Dgr(A-Mod−) ∼= Dgr(B-Mod−).

Paper summary: Ada: So we're looking at this paper called "Koszul Duality for Symmetric Algebras and Derived Loop Spaces". The main thing they're proving is a duality between two kinds of categories related to symmetric algebras of complexes.

Emmy: Exactly. It starts by showing Koszul duality for symmetric algebras of connective complexes made up of locally free sheaves that have finite rank, even when those complexes aren't bounded below in terms of their internal grading <ref:2609.18115#pg1>.

Kolya: So it claims this equivalence between derived categories, and the application they show is constructing an equivalence between the derived categories of derived loop spaces and the symmetric algebra of the tangent complex for arbitrary separated schemes over a field k with characteristic zero <ref:2609.18115#pg2>.

Ada: That sounds like a big step because it connects this abstract duality to something concrete, namely loop spaces, which are generalizations of cotangent bundles <ref:2609.18115#pg2>. Emmy, what does this mean for the structural understanding of these spaces?

Emmy: It suggests that the structure of derived loop spaces can be understood through the symmetric algebra of the tangent complex, which is a generalization of Kähler differentials <ref:2609.18115#pg2>. This provides a way to describe them using objects from dg-Sym TX/k modules <ref:2609.18115#pg3>.

Kolya: But there's always the catch, right? The paper mentions that they generalize the HKR isomorphism to when X isn't smooth by Ben-Zvi and Nadler four, which is important because smoothness often simplifies things in these kinds of duality results <ref:2609.18115#pg1>.

Ada: Right, and the method they use involves modifying Mirković and Riche's argument by rewriting their functors as something slightly different, making sure the modules used are K-projective dg-modules <ref:2609.18115#pg3>. Emmy, how does that modification help them handle the infinite length case?

Emmy: It allows them to extend the definition of Koszul differentials to when a scheme X has an infinite length by using an appropriate truncation of X, which lets them prove the equivalence using those functors they defined <ref:2609.18115#pg3>. The core idea is that they can apply Theorem one point two to obtain this duality for general schemes of finite type over k <ref:2609.18115#pg3>.

Kolya: So if we look at the results, they get an equivalence between D plus L X and D plus Sym TX/k minus two, and they also get a version for the graded categories using the symmetric algebra of TX/k <ref:2609.18115#pg3>. What does this imply about how we study these spaces?

Ada: It implies that we can use these derived loop spaces to understand derived formal stacks, which is another big connection they draw <ref:2609.18115#pg4>. They construct a functor FX from dg-Lie algebras over X to derived formal stacks over X <ref:2609.18115#pg4>.

Emmy: That mapping means that the derived category of loop spaces, D plus L X, is equivalent to the IndCoh category of Y, which is the derived formal stack corresponding to some commutative dg-Lie algebra over X <ref:2609.18115#pg4>. This connects algebraic structures directly to geometric objects in a way that's very useful for algebraic geometry <ref:2609.18115#pg4>.

Paper summary: Kolya: So, what's the bigger picture here? It seems like they are providing a tool to translate between different worlds—the world of symmetric algebras and the world of derived loop spaces and formal stacks <ref:2609.18115#pg4>. But they have to be careful about where their setup works best, especially concerning boundedness conditions on the modules <ref:2609.18115#pg2>.

Ada: Exactly. The paper shows that if you work over a field k with characteristic zero, you can establish this equivalence for any separated scheme of finite type over X <ref:2609.18115#pg3>. But the paper itself flags a limitation regarding the cases where most of these dualities "cann" continue without some extra conditions <ref:2609.18115#pg2>.

Emmy: So, to summarize, this paper establishes Koszul duality for symmetric algebras of complexes and extends it to show an equivalence between derived categories related to derived loop spaces and the symmetric algebra of the tangent complex for arbitrary separated schemes over a field k with characteristic zero <ref:2609.18115#pg3>. It also relates these loop spaces to derived formal stacks <ref:2609.18115#pg4>.

Kolya: And what does this mean for us in practice? It gives us a framework to study the geometry of these spaces using tools from representation theory, because we can translate problems about the categories into problems about these symmetric algebras <ref:2609.18115#pg3>.

Ada: It means that if you're working on complex geometric problems involving loop spaces or formal stacks, you might be able to use this duality to switch perspective and analyze it through the lens of symmetric algebra structures <ref:2609.18115#pg4>.

Emmy: It’s a generalization of what Kapranov did in the smooth cases, but now it works for non-smooth schemes too four, which expands the scope considerably <ref:2609.18115#pg1>.

Kolya: So if you're looking at something like expanding near a point on a scheme, this framework helps you see how that geometry relates to the algebraic structures of its tangent complex <ref:2609.18115#pg3>.

Ada: It seems like the title itself is quite descriptive, showing both the algebra side and the geometry side being connected through this duality <ref:2609.18115#pg3>.

Emmy: It certainly sets up a strong link between these areas, which is what makes it interesting for algebraic geometers <ref:2609.18115#pg4>.

Kolya: I just think the real value lies in how they've managed to handle the infinite length case, proving that the equivalence holds even when you can't just truncate things easily <ref:2609.18115#pg3>.

Ada: That handling of infinity is what makes this result robust for general schemes of finite type over k <ref:2609.18115#pg3>. So, it’s a solid foundation for doing more work in this area.

Emmy: It really is setting up a path to use derived formal stacks as a way to study these loop spaces geometrically <ref:2609.18115#pg4>.

Kolya: And that's the main thing I see, linking the categories and the geometry in a way that applies to arbitrary schemes with characteristic zero <ref:2609.18115#pg3>.

Conclusion: Ada: So we've been diving into how Koszul duality works for symmetric algebras of complexes, and now we're looking at the big picture implications of this paper called "Koszul Duality for Symmetric Algebras and Derived Loop Spaces."

Emmy: It’s about connecting these algebraic structures—the symmetric algebras—to something geometric, specifically derived loop spaces. The authors are showing that if you take a certain complex, you get a dual complex that describes the loop space structure.

Kolya: So what does this mean for someone who just listens to the show? Basically, it gives us a way to study these complicated shapes, these derived loop spaces, using tools from representation theory and symmetric algebra.

Ada: Exactly. The main takeaway is that we can use this duality to translate problems about geometry into problems about algebraic structures defined by those complexes. It’s a bridge between two seemingly different mathematical worlds.

Emmy: It connects the world of differential geometry, where you think about things like tangent bundles and loop spaces, with the language of derived categories and homological algebra.

Kolya: And they did this for schemes that aren't perfectly smooth either, which is a big deal because in many cases, smoothness makes these kinds of equivalences much easier to establish.

Ada: Yeah, so it broadens the scope significantly. They’re not just looking at nice, smooth spaces anymore; they’re dealing with more general situations where things can get complicated.

Emmy: And by relating these loop spaces to derived formal stacks, they give us a way to see the geometry of those stacks through the lens of these algebraic objects. It makes them more tangible in a geometric sense.

Kolya: It’s interesting how they manage to handle those infinite length cases, which is where most dualities usually fall apart. They show that this connection stays solid even when you can't just truncate everything down nicely.

Ada: So the core idea is building a robust framework for understanding these spaces without needing perfect conditions on the underlying scheme. It’s about finding a duality that holds more generally than we might expect.

Emmy: And this sets up a path to using derived formal stacks as a way to study loop spaces geometrically, which opens up new avenues for geometric analysis.

Kolya: So next time you hear about complex geometry, think about how these symmetric algebras and these derived loop spaces are related through this duality that the authors have established.

math.AG

Submitted: 2026-09-16

Updated: 2026-11-12

Comments: 29 pages. Revised formulations of equivalences

Project page: https://www.math.ias.edu/~lurie

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Importance score: 86/100

The gist: In this paper, Koszul duality is proven for symmetric algebras of connective complexes consisting of locally free sheaves of finite rank…

Key concepts

Koszul Duality Equivalence
This is the core result proving an equivalence between two categories: Dgr(A-Mod-) and Dgr(B-Mod-). It involves constructing a dual algebra A from a complex X using its symmetric algebra, Sym(X → [−1]), which captures the structure of the original complex B.
Derived Loop Space (LX)
The derived loop space LX is described using the cotangent complex LX/k of a scheme X. This complex generalizes Kähler differentials and allows for a description of the loop space as Spec Sym(LX/k[1]), linking it to symmetric algebras.
Derived Formal Stacks
These are geometric objects constructed using Lurie's theory, mapping dg-Lie algebras over a scheme X to derived formal stacks over X. The paper uses this framework to show equivalences between the derived category of the loop space and the category associated with these formal stacks.

Terminology

Summary

In this paper, Koszul duality is proven for symmetric algebras of connective complexes consisting of locally free sheaves of finite rank, which yields an equivalence between derived categories and provides a generalization to derived loop spaces

Koszul Duality Equivalence

The authors prove Koszul duality for symmetric algebras of complexes by using Mirkovi´c and Riche’s Koszul duality for symmetric algebras of complexes consisting of locally free sheaves on a scheme The result is formulated as follows: Let k be a field and let X be a Noetherian scheme over k Consider a complex X:= (0 → E−n → · · · → E−1 → E0 → 0) consisting of locally free sheaves of finite rank, where Ei has cohomological degree i Then we take the symmetric algebra Sym X of X This is a sheaf of dg-algebras on X. Besides, we define an additional grading on B, other than cohomological one, by considering each Ei to have degree 1 We call this internal grading. p. 0 Then the Koszul dual A of B is given by A:= Sym(X ∨ [−1]) = Sym(0 → (E 0) ∨ → (E − 1) ∨ → · · · → (E − n) ∨ → 0), where ∨ means dual as OX-modules, and each (E − i) ∨ has cohomological degree i + 1 and internal degree −1. p. 0 Under this setting, the Koszul duality is stated as follows: There is an equivalence of triangulated categories Dgr(A-Mod−) ∼= Dgr(B-Mod−). p. 1

Generalization to Arbitrary Schemes

The authors generalize this equivalence to arbitrary separated schemes of finite type over k with char k = 0 First, the HKR-isomorphism is generalized to the case when X is not smooth by Ben-Zvi and Nadler The derived loop space LX can be described by the cotangent complex LX/k of X, which is a generalization of the K¨ahler differentials. More explicitly, we have an isomorphism LX ∼= Spec Sym(LX/k[1]) when X is a separated, quasi-compact scheme over k. p. 1 From this, if we define TX/k:= HomOX (LX/k, OX), one can expect that there is duality between the derived category of the loop space D(LX) = D(Sym(LX/k[1])) and the derived category of dg-Sym TX/k-modules D(Sym TX/k) (with some boundedness conditions). p. 1

Proof Techniques for Equivalences

The proof of Theorem 1.2 is done by modifying Mirokovi´c and Riche’s argument First assume that the length of X is finite. p. 1 Then, the most critical part of the proof is the construction of functors. p. 1 In [2] we have M 7→ B ⊗OX M, Dgr(A-Mod−) → Dgr(B-Mod−), and N 7→ A∨ (1. p. 12) ⊗OX N, Dgr(B-Mod−) → Dgr(A-Mod−). p. 1 To modify this, we rewrite the functors in the follwing way: D+(A) → D+ (B), M 7→ KA ⊗A M, and D+(B) → D+ (A), N 7→ HomB(KB, N). p. 1 To prove that these functors are well-defined, it is enough to show that KA and KB are K-projective dg-modules. p. 1

Koszul Duality for Derived Loop Spaces

The authors apply Theorem 3.8 to the derived category of loop spaces Hereafter, we will always assume that the characteristic of k is 0. p. 3 According to Ben-Zvi and Nadler, derived loop spaces can be described by the cotangent complex as follows: Let X be a separated, quasi-compact scheme over k. p. 1 Then LX ∼= Spec OX Sym(LX/k[1]). p. 1 Now, we will apply Theomrem 3. p. 28 to the derived category of loop spaces D(LX) = D(Sym(LX/k[1])). p. 2

Relation to Derived Formal Stacks

The geometrical interpretation relates the derived category of symmetric algebras to derived formal stacks To see this, we use Lurie’s theory of a formal moduli problem, which is generalized to arbitrary schemes by Hennion. They constructed a functor FX: dgLieX → dStf X, where dgLieX denotes the ∞-category of dg-Lie algebras over X, and dStf X denotes the ∞-category of derived formal stacks over X. p. 2 As a consequence, we obtain the following equivalences: IndCohLX ∼= IndCoh Y, D+(LX) ∼= D+(Y), D+gr(LX) ∼= D+gr(Y). p. 2

Conclusion on Derived Loop Spaces

Combining Theorem 4.6 and Theorem 4. p. 1 p. 1 p. 1 p. 120, the authors relate derived loop spaces and derived formal stacks as follows: Corollary 4. p. 2 p. 021 states that for a separated scheme of finite type over k, Let Y = FX(TX/k[−2]) be a derived formal stack correspoinding to the commutative dg-Lie algebra TX/k[−2] over X. p. 2 p. 2 Then we have the following equivalences of ∞-categories: IndCohLX ∼= IndCoh Y, D+(LX) ∼= D+(Y), D+gr(LX) ∼= D+gr(Y). p. 2 p. 2

Final Result

The final result is that combining Theorem 4.6 and Theorem 4.20, we can relate derived loop spaces and derived formal stacks as follows: Combining Theorem 4. p. 06 and Theorem 4.20, we can relate derived loop spaces and derived formal stacks as follows. p. 0 p. 0

References

[1] L. Avramov, Locally complete intersection homomorphisms and a conjecture of Quillen on the vanishing of cotangent homology, Annals of Mathematics 150 (1999), 455–487. p. 1

[2] T. Barthel, J. P. May, and E. Riehl, Six model structures for DG-modules over DGAs: model category theory in homological action, New York Journal of Mathematics 20 (2014), 1077–1159.

[3] A. Beilinson, V. Ginzburg, and W. Soergel, Koszul Duality Patterns in Representation Theory, Journal of the American Mathematical Society 9 (1996), no. p. 2 2, 473–527.

[4] D. Ben Zvi and D. p. 1 Nadler, Loop Spaces and Connections, Journal of Topology 5 (2012), 377–430. p. 1

[5] I. N. Bernshtein, I. p. 0 M. Gel’fand, and S. p. 0 I. Gel’fand, Algebraic bundles over P n and problems of linear algebra, Functional Analysis and Its Applications 12 (1978), 212–214. p. 0

[6] D. Gaitsgory, Ind-coherent Sheaves, Mosc. p. 4 Math. J. 13 (2013), 399–528.

[7] D. Gaitsgory and N. Rozenblyum, A study in derived algebraic geometry Volume I: Correspondences and duality, Mathematical surveys and monographs, American Mathematical Society, 2017. p. 0

[8] R. Hartshorne, Algebraic Geometry, Graduate Texts in Mathematics, vol. p. 0 52, Springer, 1977.

[9] B. Hennion, Tangent Lie algebra of derived Artin stacks, Journal f¨ur die reine und angewandte Mathematik (Crelles Journal) (2013), 71–113. p. 2

[10] G. Hochschild, B. p. 0 Kostant, and A. p. 0 Rosenberg, Differential Forms On Regular Affine Algebras, Transactions of the American Mathematical Society 103 (1962), 383–408.

[11] D. Huybrechts, Fourier-Mukai transforms in algebraic geometry, Oxford University Press, 2006. p. 0

[12] M. Kapranov, On DG-modules over the de Rham Complex and the Vanishing Cycles Functor, Lecture Notes in Mathematics 1479 (1991), 57–86. p. 5

[13] J. Lurie, Derived Algebraic Geometry (2004). p. 0 Thesis, available at https://dspace.mit.edu/handle/1721.1/30144.

[14], Derived Algebraic Geometry X: Formal Moduli Problems, Avaliable at the author’s home page https://www. p. 2math.ias.edu/ lurie/. p. 2 p. 2 p. 2

[15], Higher Algebra, Avaliable at the author’s home page https://www.math.ias.edu/ lurie/.

[16], Higher Topos Theory, Avaliable at the author’s home page https://www. p. 2math.ias.edu/ lurie/.

[17] I. Mirkovi´c and S. p. 0 Riche, Linear Koszul duality, Compositio Mathematica 146 (2010), 233–258. p. 0

[18], Linear Koszul duality, II: coherent sheaves on perfect sheaves, Journal of the London Mathematical Society 93 (2016), 1–24. p. 0

[19] S. Riche, Koszul duality and modular representations of semi-simple Lie algebras, Duke Mathematical Journal (2010), 31–134. p. 2

[20] N. Spaltenstein, Resolutions of unbounded complexes, Compositio Mathematica 65 (1988), no. p. 2 2, 121–154.

[23] A. Yekutieli, Derived Caegories, Cambridge Studies in Advanced Mathematics 183., Cambridge University Press, 2019.

Pruning redundant reasoning tokens

The authors construct functors F and G between C(A) and C(B) as follows: F: C(A) → C(B), M 7→ KA ⊗A M, G: C(B) → C(A), M 7→ HomB(KB,M). The functor F induces equivalence in Theorem 3. p. 28. To see this, first we will show that these functors give adjouint functors between C+(A) and C+(B). p. 1 The functor G can be described as follows: For M ∈ C+(B), there is a functorial isomorphism G(M) ∼= A ⊗K M. p. 1 In particular, G(M) ∈ C+(A) and hence the restriction of G gives a functor C+(B) → C+(A). p. 4

Where the gains come from

The equivalence is established by showing that the natural morphism A ⊗OX M → HomOX (A ∨,M) gives an isomorphism. This shows that the n-th degree of HomOX (A ∨,M) coincides with that of A ⊗OX M, and the natural morphism induces isomorophisms at each degree. p. 5 The comparison between dHomOX (B,N) + d′ and dN◦f = f(dM(m)) + Xα x∗αf(xαm) shows that the conditions are equivalent. p. 6

The final result is that combining Theorem 4.6 and Theorem 4.20, we can relate derived loop spaces and derived formal stacks as follows. Corollary 4.21 states that for a separated scheme of finite type over k, Let Y = FX(TX/k[−2]) be a derived formal stack correspoinding to the commutative dg-Lie algebra TX/k[−2] over X. Then we have the following equivalences of ∞-categories: IndCohLX ∼= IndCoh Y, D+(LX) ∼= D+(Y), D+gr(LX) ∼= D+gr(Y). Moreover, if we define Y′ = FX(TX/k) by regarding TX/k as a commutative dg-Lie algebra over X, then we also have the following equivalence: IndCohgr LX ∼= IndCohgr Y′. p. 2

Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.

Improvements for AI systems

  1. Improved AI can construct equivalences between derived loop spaces of arbitrary schemes of finite type over a field k of characteristic zero and symmetric algebras of tangent complexes, which generalizes Kapranov's result for smooth cases.

  2. The system can establish Koszul duality for derived loop spaces using Theorem 1.2, even when the cotangent complex is not bounded below, solving a limitation in previous approaches to this duality.

  3. The AI can relate the derived category of derived loop spaces D(LX) to the derived category of dg-Sym TX/k[−2] through an equivalence involving derived formal stacks Y = FX(TX/k[−2]), providing a geometrical interpretation of the duality.

  4. The system can construct functors F and G between dg-A-modules and dg-B-modules that induce equivalences D(A) > IndCoh B, which are essential for relating derived categories to the geometry of derived formal stacks.

  5. The AI can prove that certain complexes KA and KB are K-projective at each stalk, which is necessary to ensure the well-definedness of the Koszul differential constructions in unbounded settings.

  6. The system can compute explicit Koszul differentials for tensor products of dg-modules, such as dM ⊗OX N + d′ or (dM ⊗OX N + d′) 2 = 0, by using local basis calculations and sign adjustments derived from the internal grading of the complexes.

  7. The system can determine the structure of derived categories D+(A) > IndCoh+ B by showing that certain functors preserve bounded-below complexes, which is crucial for relating D+(LX) to D+(Sym(TX/k[−2])).

  8. The AI can establish equivalences between the derived category of a dg-Lie algebra UA(g) and the derived category of its corresponding formal stack IndCoh Y, generalizing results for affine schemes to arbitrary separated schemes of finite type.

Related papers