Relative Nonhomogeneous Koszul Duality 1st ed.

2021
Edition Leonid Positselski 2025 full version

https://ebookname.com/product/relative-nonhomogeneous-koszul-
duality-1st-ed-2021-edition-leonid-positselski/

★★★★★
4.6 out of 5.0 (22 reviews )

Download PDF Now

ebookname.com
Relative Nonhomogeneous Koszul Duality 1st ed. 2021 Edition
Leonid Positselski

EBOOK

Available Formats

■ PDF eBook Study Guide Ebook

EXCLUSIVE 2025 ACADEMIC EDITION – LIMITED RELEASE

Available Instantly Access Library

Instant digital products (PDF, ePub, MOBI) available
Download now and explore formats that suit you...

Functional Analysis Leonid P. Lebedev

https://ebookname.com/product/functional-analysis-leonid-p-lebedev/

ebookname.com

Particles in turbulent flows Leonid Zaichik

https://ebookname.com/product/particles-in-turbulent-flows-leonid-
zaichik/

ebookname.com

101 Sewing Seams 2nd 2021 ed. Edition Abc Seams® Pty. Ltd.

https://ebookname.com/product/101-sewing-seams-2nd-2021-ed-edition-
abc-seams-pty-ltd/

ebookname.com

Advances in Inorganic Chemistry Vol 53 1st Edition A.G.

Sykes

https://ebookname.com/product/advances-in-inorganic-chemistry-
vol-53-1st-edition-a-g-sykes/

ebookname.com
The Molecules of Life Physical and Chemical Principles 1st
Edition John Kuriyan

https://ebookname.com/product/the-molecules-of-life-physical-and-
chemical-principles-1st-edition-john-kuriyan/

ebookname.com

Dynamics of Markets 2ed. Edition Joseph L. Mccauley

https://ebookname.com/product/dynamics-of-markets-2ed-edition-joseph-
l-mccauley/

ebookname.com

Soldiers and the Soviet State Civil Military Relations

from Brezhnev to Gorbachev Timothy J. Colton (Editor)

https://ebookname.com/product/soldiers-and-the-soviet-state-civil-
military-relations-from-brezhnev-to-gorbachev-timothy-j-colton-editor/

ebookname.com

The Measurement of Productive Efficiency and Productivit

Growth Harold O. Fried

https://ebookname.com/product/the-measurement-of-productive-
efficiency-and-productivit-growth-harold-o-fried/

ebookname.com

The Advice of Prophet Yahya ibn Zakariya explained 1st

Edition Ibn Al-Qayyim

https://ebookname.com/product/the-advice-of-prophet-yahya-ibn-
zakariya-explained-1st-edition-ibn-al-qayyim/

ebookname.com
Studies in Modern Childhood Society Agency Culture First
Edition Jens Qvortrup

https://ebookname.com/product/studies-in-modern-childhood-society-
agency-culture-first-edition-jens-qvortrup/

ebookname.com
Frontiers in Mathematics

Advisory Editors
William Y. C. Chen, Nankai University, Tianjin, China
Laurent Saloff-Coste, Cornell University, Ithaca, NY, USA
Igor Shparlinski, The University of New South Wales, Sydney, NSW, Australia
Wolfgang Sprößig, TU Bergakademie Freiberg, Freiberg, Germany

This series is designed to be a repository for up-to-date research results which have been
prepared for a wider audience. Graduates and postgraduates as well as scientists will
benefit from the latest developments at the research frontiers in mathematics and at the
“frontiers” between mathematics and other fields like computer science, physics, biology,
economics, finance, etc. All volumes are online available at SpringerLink.

More information about this series at https://link.springer.com/bookseries/5388

Leonid Positselski

Relative Nonhomogeneous
Koszul Duality
Leonid Positselski
Institute of Mathematics
Czech Academy of Sciences
Praha 1, Czech Republic

ISSN 1660-8046 ISSN 1660-8054 (electronic)

Frontiers in Mathematics
ISBN 978-3-030-89539-6 ISBN 978-3-030-89540-2 (eBook)
https://doi.org/10.1007/978-3-030-89540-2

© The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland AG
2021
This work is subject to copyright. All rights are solely and exclusively licensed by the Publisher, whether the
whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations,
recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information
storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now
known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does
not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective
laws and regulations and therefore free for general use.
The publisher, the authors, and the editors are safe to assume that the advice and information in this book are
believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors
give a warranty, expressed or implied, with respect to the material contained herein or for any errors or omissions
that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps
and institutional affiliations.

This book is published under the imprint Birkhäuser, www.birkhauser-science.com, by the registered company
Springer Nature Switzerland AG.
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Preamble

This book starts with a Prologue and an Introduction. The Prologue introduces the subject
of derived Koszul duality, particularly derived nonhomogeneous Koszul duality. It is
intended for the benefit of the reader not yet familiar with the major preceding results
in the area, as presented, in particular, in the author’s memoir [59].
The Introduction introduces the relative nonhomogeneous Koszul duality theory
developed in this book.
For homogeneous Koszul duality, we refer the reader to the paper [7] and the
memoir [59, Appendix A]. For underived Koszul duality, a suitable reference is the
book [53].

v
Prologue

Koszul duality is a fundamental phenomenon in the “algebraic half of” mathematics,

including such fields as algebraic topology, algebraic and differential geometry, and
representation theory. The phenomenon is so general that it does not seem to admit a
“maximal natural” generality. Manifestations of Koszul duality can be found everywhere.
Noticing such manifestations is easy; interpreting them properly and distinguishing from
other phenomena is harder.
Koszul duality is best explained as the relation connecting the homotopy groups of
a topological space with its cohomology groups. Such an abstract topological setting is
obviously very complicated; to consider something more approachable, one can specialize
to rational homotopy theory or to stable homotopy theory.
In the context of rational homotopy theory, a version of Koszul duality connecting
Lie DG-algebras computing the rational homotopy groups with commutative DG-algebras
computing the cohomology spaces is due to Quillen [72]. The stable homotopy category,
on the other hand, is additive, so its algebraic analogues are the derived categories
of module-like objects. In the stable homotopy theory setting, the Adams spectral
sequence can be understood as expressing the Koszul duality between the homotopy and
cohomology groups.
The functor of stable homotopy groups is corepresented by the sphere spectrum, while
the functor of cohomology groups is represented by the Eilenberg–MacLane spectrum.
Morphisms from the sphere spectrum to the Eilenberg–MacLane spectrum are very easy
to describe; in a sense, one can say that these are one-sided dual basis vectors with respect
to the nonsymmetric graded Hom pairing. Graded endomorphisms of the Eilenberg–
MacLane spectrum (known as the Steenrod algebra) are more complicated, and graded
endomorphisms of the sphere spectrum (the stable homotopy groups) are much more
complicated. The Adams spectral sequence [1] expresses the connection between these.
In representation theory, nonhomogeneous Koszul duality manifests itself as the
connection between the universal enveloping algebra U (g) of a Lie algebra g over a
field k and the (co)homological Chevalley–Eilenberg complex ((g), d) or ((g∗ ), d)
of g with the coefficients in k [59, Example in Section 6.6]. In differential and algebraic
geometry, relative nonhomogeneous Koszul duality appears as the duality between the

vii
viii Prologue

ring of differential operators on a smooth variety and the de Rham complex of differential
forms [36], [6, Section 7.2], [59, Appendix B]. Homogeneous Koszul duality plays a role
in the theory of algebraic vector bundles over projective spaces [9], [51, Appendix A] and
some other varieties.
The duality between the categories of Verma modules over the Virasoro algebra with
the complementary central charges c and 26 − c (or the similar duality for the Kac–Moody
algebras, etc.) [25, 74] is not an instance of Koszul duality. Rather, these are advanced
manifestations of the comodule-contramodule correspondence or, more precisely, the
semimodule-semicontramodule correspondence [58]. The MGM duality [19, 30, 55, 62]
and the (covariant) Serre–Grothendieck duality [33, 47, 49, 63] are also instances of the
comodule-contramodule correspondence (see the discussion in [62]) and not of Koszul
duality.
A simple rule of thumb: a contravariant duality assigning projective modules to
projective modules, or a covariant duality functor assigning projective modules to injective
modules, or flat modules to injective modules, etc., is a comodule-contramodule corre-
spondence. A duality assigning irreducible modules to projective modules, or irreducible
modules to injective modules, etc., is a Koszul duality.

***
In the algebraic context, Koszul duality is most simply formulated as the connection
between an augmented algebra A over a field k and the DG-algebra R HomA (k, k)
representing the graded algebra Ext∗A (k, k). In particular, to a complex of A-modules
M • one can assign either one of the three DG-modules over R HomA (k, k) representing
Ext∗A (M • , k), or Tor∗A (k, M • ), or Ext∗A (k, M • ). Naïvely, one would hope for such a functor
to be an equivalence between the derived category of A-modules and the derived category
of DG-modules over R HomA (k, k).
More generally, one would start with an augmented DG-algebra A = (A, d); then,
the role of R HomA (k, k) is played by the cobar construction of A. Once again, the
naïve hope would be to have a triangulated equivalence between the derived categories of
DG-modules over A and its cobar construction. An attempt to realize this hope was made
in [37, Section 10], but it was possible to obtain a triangulated equivalence only under
severe restrictions on the augmented DG-algebra (or DG-category). Under somewhat
more relaxed but still quite restrictive assumptions, a fully faithful functor in one or other
direction was constructed.
The following simple examples illustrate the situation and the lines of thought leading
to the present-day approach. Let S = k[x] be polynomial algebra in one variable x over a
field k. Then, the DG-algebra HomS (k, k) is quasi-isomorphic to the ring of dual numbers,
or in other words, to the exterior algebra  = k[]/( 2 ), where  is an element of
cohomological degree 1. More generally, one can consider S = k[x] as a DG-algebra
with the generator x placed in the cohomological degree n ∈ Z and zero differential, and
then the Koszul dual exterior algebra  has its generator  situated in the cohomological
degree 1 − n (and the differential on  is also zero).
Prologue ix

The classical homogeneous Koszul duality context presumes that all the algebras and
modules are endowed with an additional essentially positive (or negative) “internal”
grading. Then one can say that the bounded derived category of finitely generated
graded S-modules is equivalent to the bounded derived category of finite-dimensional
graded -modules. This result generalizes straightforwardly to the symmetric and exterior
algebras in several variables [9], [51, Appendix A], and further to Noetherian Koszul
graded algebras of finite homological dimension (in the role of S) and their finite-
dimensional Koszul dual graded algebras (in the role of ) [7]. A formulation replacing all
the finiteness assumptions with explicit assumptions on the sign of the additional grading
is also possible [59, Appendix A].
In the nonhomogeneous setting (without an internal grading on the modules, but only
with the cohomological one), it turns out that, for any n ∈ Z as above, the derived
category of DG-modules over  is equivalent to a full subcategory in the derived
category of DG-modules over S. If ones wishes, one can also view D(–mod) as a
triangulated quotient category of D(S–mod), but the two derived categories are decidedly
not equivalent. There are two ways out of this predicament, each of which is further
subdivided into two versions. Simply put, one can either shrink D(S–mod) (replacing it
with a smaller category) or inflate D(–mod) (replacing it with a larger category) to make
the resulting versions of the derived categories of DG-modules over S and  equivalent.
The basic principle of derived nonhomogeneous Koszul duality, as developed in the
paper [31], the dissertation [41], the note [38], and the memoir [59], is that it connects
algebras with coalgebras. So the main decision one has to make is: Where is the algebra
side, and where is the coalgebra side of the story? Replacing S with its dual coalgebra
means shrinking the category D(S–mod). Replacing  with its dual coalgebra means
inflating the category D(–mod).
What does it mean to “replace S with its dual coalgebra”? When n = 0, the
cohomological grading on S is nontrivial, the algebra S is locally finite-dimensional in
this grading, and one can simply consider the graded dual coalgebra C to S. For n = 0,
one starts with replacing the polynomial ring S = k[x] with the ring of formal power series
k[[x]]. Then C is the coalgebra over k whose dual algebra C ∗ is isomorphic to k[[x]] (as a
topological algebra). In both cases, C is viewed as a DG-coalgebra over k with the grading
components or direct summands situated in the cohomological degrees 0, −n, −2n, . . .
and zero differential.
How does the passage from S to C affect the derived category of modules? The
next guiding principle is that there are two kinds of abelian module categories over a
coalgebra, the comodules and the contramodules [21, Section III.5], [61]. In the case of the
polynomial algebra S and its dual coalgebra C, the C-comodules and the C-contramodules
are two different full subcategories in S-modules. The abelian category C–comod can be
simply described as the category of locally nilpotent S-modules (i.e., the action of x must
be locally nilpotent), while C–contra is the category of k[x]-modules with x-power infinite
summation operations. Both the derived categories D(C–comod) and D(C–contra) of
Visit https://ebookname.com today to explore
a vast collection of ebooks across various
genres, available in popular formats like
PDF, EPUB, and MOBI, fully compatible with
all devices. Enjoy a seamless reading
experience and effortlessly download high-
quality materials in just a few simple steps.
Plus, don’t miss out on exciting offers that
let you access a wealth of knowledge at the
best prices!
x Prologue

DG-comodules and DG-contramodules over C are full subcategories in D(S–mod). Both

D(C–comod) and D(C–contra) are equivalent to the derived category D(–mod).
What does it mean to “replace  with its dual coalgebra”? The graded algebra  is
finite-dimensional, so its (graded) dual vector space ∗ is a graded coalgebra over k. It is
viewed as a DG-coalgebra with the grading components or direct summands situated in the
cohomological degrees 0 and n − 1, and zero differential. Since  is finite-dimensional,
both the ∗ -comodules and ∗ -contramodules are the same thing as -modules, and both
the DG-comodules and the DG-contramodules over ∗ are the same thing as DG-modules
over . So, what’s the difference, and what have we achieved by passing from  to ∗ ?
Here is the answer. Yet another guiding principle in Koszul duality is that one should
consider the derived categories of modules, the coderived categories of comodules, and
the contraderived categories of contramodules. The coderived and the contraderived
categories, called collectively the derived categories of the second kind, are the categories
of complexes or differential graded structures viewed up to a finer equivalence relation
than the conventional quasi-isomorphism. So some acyclic complexes represent nonzero
objects in the coderived category, while some other acyclic complexes represent nonzero
objects in the contraderived category (or in both).
For example, let us take n = 1 in the notation above; so the k-algebra  = k[]/( 2)
is situated in the cohomological degree 0. Then the unbounded acyclic complex of free
-modules with one generator (which are both projective and injective as -modules)

∗ ∗
· · · −−→ −−→  −−→ −−→ · · · (∗)

is neither coacyclic, nor contraacyclic. Cutting this complex in half by the canonical
truncation, one obtains the bounded above complex

∗ ∗
· · · −−→  −−→ −−→k −−→0 (∗∗)

which is contraacyclic but not coacyclic, and the bounded below complex

∗ ∗
0−−→k −−→ −−→  −−→ · · · (∗∗∗)

which is coacyclic but not contraacyclic [59, Examples in Section 3.3] (a detailed
discussion of these examples can be found in [70, Section 5]).
For any n ∈ Z, the coderived and the contraderived categories of DG-modules over
, denoted by Dco (–mod) = Dco (∗ –comod) and Dctr (–mod) = Dctr (∗ –contra),
are naturally equivalent to each other. They are also equivalent to the derived category
D(S–mod) of DG-modules over S.
And what about the coalgebra C dual to the algebra S, why did we consider the derived
categories of DG-comodules and DG-contramodules over C in the preceding discussion,
seemingly contrary to the (subsequently introduced) guiding principle of taking the
Prologue xi

coderived categories of comodules and the contraderived categories of contramodules?

Because, in this case, there is no difference: one has D(C–comod) = Dco (C–comod)
and D(C–contra) = Dctr (C–contra), since C is a graded coalgebra of finite homological
dimension endowed with zero differential [39, Proposition 5.9].
Another way to explain the situation is to notice that the Koszul duality functor assigns
the acyclic complex (∗) to the S-module S[x −1 ] = k[x, x −1] of Laurent polynomials in the
variable x over k (viewed as a DG-module with zero differential). So, in order to make this
functor a triangulated equivalence, one has to either declare (∗) to be a nonzero object in an
exotic derived category of -modules, or otherwise prohibit the S-module k[x, x −1 ] (i.e.,
impose a restriction on the S-modules under consideration putting this module outside of
the category of modules appearing in the duality). It turns out that there are two ways to do
the former (basically, one can declare either the complex (∗∗) or the complex (∗∗∗) to be
nonzero), and there are two ways to do the latter (one can either prohibit the free S-module
S = k[x] on the grounds of it being not a C-comodule, or prohibit the injective S-module
k[x, x −1 ]/k[x] on the grounds of it being not a C-contramodule).

***
So, a more insightful formulation of Koszul duality for augmented DG-algebras over
a field k presumes that it connects DG-algebras with DG-coalgebras. To an augmented
DG-algebra A = (A, d), a DG-coalgebra C computing TorA ∗ (k, k) is assigned. To a
coaugmented DG-coalgebra C = (C, d), a DG-algebra A computing Ext∗C (k, k) is
assigned. More specifically, given a DG-algebra A, the corresponding DG-coalgebra C can
be produced as the bar construction of A, or, given a DG-coalgebra C, the corresponding
DG-algebra A can be produced as the cobar construction of C. Both the constructions
are covariant functors between the categories of augmented DG-algebras and augmented
DG-coalgebras.
The next principle is that one should restrict oneself to conilpotent coalgebras. Here one
observes that the nilpotency works better with coalgebras than with algebras inasmuch
as the direct limits are more convenient to work with than the inverse limits. Given an
augmented algebra A with the augmentation ideal A+ , one says that A is nilpotent if there
exists an integer n ≥ 1 such that (A+ )n = 0. If one wants to let n approach infinity,
as we do, then one has to speak of pronilpotency. The dual condition for coalgebras is
conilpotency (which might as well be called ind-conilpotency, for it does not presume
existence of a fixed finite nilpotency index n for the whole coalgebra C).
Furthermore, similarly to DG-comodules, one has to consider conilpotent DG-coal-
gebras up to a more delicate equivalence relation than the conventional quasi-isomorphism.
This equivalence relation, called the filtered quasi-isomorphism, was discovered by Hinich
in [31].
With these considerations in mind, the Koszul duality can be formulated quite
generally as a category equivalence between the category of augmented DG-algebras up
to quasi-isomorphism and the category of conilpotent DG-coalgebras up to filtered quasi-
isomorphism [59, Theorem 6.10(b)]. Furthermore, whenever a DG-algebra A = (A, d)
xii Prologue

and a DG-coalgebra C = (C, d) correspond to each other under this equivalence, one
has a triangulated equivalence between the derived category of DG-modules over A,
the coderived category of DG-comodules over C, and the contraderived category of
DG-contramodules over C (“Koszul triality”)

Dco (C–comod)

D(A–mod)

Dctr (C–contra)

Here, the equivalence Dco (C–comod)  Dctr (C–contra), which holds quite generally
for any (not necessarily coaugmented or conilpotent) DG-coalgebra C over a field k, is
called the derived comodule-contramodule correspondence. The comodule side of the
triangle (∗∗∗∗) can be found formulated in [38] based on the results of [41], who was
following the approach of [31] (except that the definition of the coderived category in [38]
is less intrinsic than the modern one in [59]).

***
A conilpotent coalgebra C is coaugmented by definition. But what if an algebra A is
not augmented? Yet another guiding principle tells that absence of a chosen augmentation
on one side of the Koszul duality corresponds to presence of a curvature on the other side.
This observation goes back to the author’s early paper [56].
A curved DG-ring is a very natural concept: in particular, the construction of the
DG-category of DG-modules over a DG-ring extends naturally to CDG-rings. Curved
DG-modules over a curved DG-ring form a DG-category. Moreover, the passage from
DG-rings to CDG-rings involves not only adding new objects, but also new morphisms;
the inclusion functor of Ringsdg into Ringscdg is faithful, but not fully faithful. A
CDG-isomorphism of (C)DG-rings (e.g., a change-of-connection isomorphism, otherwise
known as a Maurer–Cartan twist) induces an equivalence of the DG-categories of
(C)DG-modules.
Besides Koszul duality, curved DG-rings and curved A∞ -algebras play a fundamental
role in the Fukaya theory [13, 27, 28] and deformation theory [16, 39], they appear in
the theory of Legendrian knots [50], etc. A systematic treatment of weakly curved DG-
and A∞ -algebras over topological local rings (i.e., curved algebras with the curvature
element divisible by the maximal ideal of the local ring of coefficients) can be found in the
memoir [60].
However important the coderived and contraderived categories of DG-modules
(DG-comodules, or DG-contramodules) are, for curved DG-modules (curved DG-co-
Prologue xiii

modules, or curved DG-contramodules) they are more important still. The cohomology
groups or modules of CDG-modules are undefined, as the differential does not square to
zero (instead, it squares to the operator of multiplication with the curvature). So one cannot
speak about acyclic CDG-modules or quasi-isomorphisms of CDG-modules in the usual
sense of the word. But the full triangulated subcategories of coacyclic and contraacyclic
CDG-modules in the homotopy category of CDG-modules are perfectly well-defined.
Hence there is no alternative to derived categories of the second kind for CDG-modules,
as the conventional derived category (“of the first kind”) does not make sense for them.
The weakly curved DG-modules (mentioned above) are the only known exception [60].
Matrix factorizations [12, 22] are an important and popular particular case of
CDG-modules. Derived categories of the second kind, including specifically the coderived
and “absolute derived” categories, play a crucial role in their theory [3, 20, 52, 54]. These
are “strongly” (i.e., not weakly) curved.
Inverting the arrows in the definition of a CDG-algebra over a field, one obtains the
definition of a CDG-coalgebra. Similarly to the CDG-modules, the differential does not
square to zero either in CDG-rings or in CDG-coalgebras; so the conventional notion
of quasi-isomorphism is undefined for them. But the definition of a filtered quasi-
isomorphism makes perfect sense for conilpotent CDG-coalgebras.
The nonaugmented Koszul duality is an equivalence between the category of
DG-algebras up to quasi-isomorphism and the category of conilpotent CDG-coalgebras
up to filtered quasi-isomorphism (over a fixed field k) [59, Theorem 6.10(a)]. Whenever
a DG-algebra A = (A, d) and a CDG-coalgebra C = (C, d, h) (where h is the
curvature element) correspond to each other under this equivalence, the “Koszul triality”
picture (∗∗∗∗) holds [59, Sections 6.3–6.6].
So the derived category of DG-modules over A, the coderived category of
CDG-comodules over C, and the contraderived category of CDG-contramodules over C
are equivalent to each other. Once again, the equivalence between the latter two categories
is an instance of the derived comodule-contramodule correspondence (and it holds quite
generally for any CDG-coalgebra C over a field k), while the other two equivalences in
the triangle diagram are the comodule and the contramodule sides of the Koszul duality.

***
The relative Koszul duality theory developed in this book is both more and less general
than the derived nonhomogeneous Koszul duality of [31,38,41,59]. The theory in this book
is more general in that it is relative, i.e., worked out over an arbitrary (noncommutative,
nonsemisimple) base ring. It is a nontrivial generalization, both because the underived
homogeneous Koszulity and Koszul duality theory over an arbitrary base ring is more
complicated than over a field (or over a semisimple base ring, as in [7]), and because
the differential operators (or the enveloping algebras of Lie algebroids) are nontrivially
more complicated than the enveloping algebras of Lie algebras over fields. The de Rham
DG-algebra of differential forms (or the Chevalley–Eilenberg complex of a Lie algebroid)
xiv Prologue

is likewise more complicated than the Chevalley–Eilenberg complex of a Lie algebra over
a field. Notice that the de Rham differential is not linear over the ring of functions.
Still the theory in this book is less general than in [59] in that the “algebra side of
the story” is presumed to be just a ring (situated in the cohomological degree 0) rather
than a DG-ring. In the notation above, this means that we restrict ourselves to the case
n = 0 if S is chosen to be on the algebra side, or otherwise we take n = 1 if the algebra
side is . This restriction of generality is chosen in order not to make the exposition
too complicated while including the most important examples (viz., various versions of
the ring of differential operators, which are indeed just rings and not DG-rings). On the
“coalgebra side of the story” we still obtain a (curved or uncurved) DG-ring, such as the
de Rham DG-algebra.
Introduction

0.0 Let A be an associative ring and R ⊂ A be a subring. Derived Koszul duality is the
functor Ext∗A (−, R), or TorA ∗
∗ (R, −), or ExtA (R, −), enhanced to an equivalence of derived
categories of modules.
The above definition raises many questions. To begin with, R is not an A-module. So
what does this Ext and Tor notation even mean?
Secondly, let us consider the simplest example where R = k is a field and A = k[x] is
the algebra of polynomials in one variable. Then k indeed can be viewed as an A-module.
There are many such module structures, indexed by elements a of the field k: given a ∈ k,
one can let the generator x ∈ A act in k by the multiplication with a. Denote the resulting
A-module by ka .
To be specific, let us choose k = k0 as our preferred A-module structure on k. Then
the functors Ext∗A (−, k0 ), Tor∗A (k0 , −), and Ext∗A (k0 , −) are indeed well-defined on the
category of A-modules. But these functors are far from being faithful or conservative: all
of them annihilate the A-modules ka with a = 0. How, then, can one possibly hope to
enhance such cohomological functors to derived equivalences?

0.1 Koszul duality has to be distinguished from the comodule-contramodule correspon-

dence, which is a different, though related, phenomenon.
In the simplest possible form, the comodule-contramodule correspondence is the
functor Ext∗A (−, A) enhanced to a derived equivalence (while Koszul duality is
Ext∗A (−, k), where k is the ground field). In a more realistic covariant and relative
situation, comparable to the discussion of Koszul duality in Sect. 0.0, the comodule-
contramodule correspondence would be a derived equivalence enhancement of a functor
like Ext∗A (HomR (A, R), −) or TorA∗ (−, HomR (A, R)).

0.2 In the present author’s research, the desire to understand Koszulity and Koszul
duality was the starting point. Then the separate existence and importance of comodule-
contramodule correspondence was realized, particularly in the context of semi-infinite
homological algebra [58]. The derived nonhomogeneous Koszul duality over a field
was formulated as a “Koszul triality” picture, which is a triangle diagram of derived
xv
xvi Introduction

equivalences with the comodule-contramodule correspondence present as one side of the

triangle and two versions of Koszul duality as two other sides [59].
The comodule-contramodule correspondence, its various versions, generalizations,
and philosophy, are now discussed in several books and papers of the present author,
including [58,59,61,62,66,71] and others. On the other hand, the derived nonhomogeneous
Koszul duality over a field attracted interest of a number of authors, starting from early
works [31,37,38,41] and to very recent, such as [14,45]; there is even an operadic version
of it in [32].
Still, there is a void in the literature concerning relative nonhomogeneous Koszul
duality. Presently, the only source of information on this topic known to this author is his
previous book [58], which contains an introductory discussion without proofs or details
in [58, Section 0.4] and a heavily technical treatment in a very general and complicated
setting in [58, Chapter 11]. (The memoir [60] represents a very different point of view.)
This book is intended to fill the void by providing a reasonably accessible, detailed
exposition on a moderate generality level.
Let us emphasize that relative nonhomogeneous Koszul duality is important. In addition
to the presence of very natural examples such as the duality between the ring of differential
operators and the de Rham DG-algebra (see Sect. 0.7 below), relative nonhomogeneous
Koszul duality plays a crucial role in the semi-infinite (co)homology theory, as it was first
pointed out in [2]. This idea was subsequently developed and utilized in [58, Section 11.9
and Appendix D].
The special case of triangulated equivalences between complexes of modules over the
rings/sheaves of differential operators and DG-modules over the de Rham DG-algebra has
been considered in [36] and [6, Section 7.2]. Our own treatment of it is presented in [59,
Appendix B].

0.3 Let us start to explain the meaning of the terms involved. In the notation of Sect. 0.0,
relative means that R is an arbitrary ring rather than simply the ground field. Homogeneous
∞
Koszul duality means that A = n=0 An is a nonnegatively graded ring and R = A0
is the degree-zero grading component. In this case, R is indeed naturally both a left
and a right R-module, so the meaning of the Ext and Tor notation in Sect. 0.0 is clear.
Nonhomogeneous Koszul duality is the situation when there is no such grading on the
ring A.
The main specific aspect of the homogeneous case is that one can consider graded
A-modules with a bounding condition on the grading, that is, only positively graded or only
negatively graded modules. If M is a positively graded left A-module, then R⊗A M = 0
implies M = 0, while if P is a negatively graded left A-module, then HomA (R, P ) = 0
implies P = 0. Hence the second problem described in Sect. 0.0 does not occur, either.
In the nonhomogeneous situation, the solution to the second problem from Sect. 0.0 is
to consider derived categories of the second kind. This means that certain complexes or
DG-modules are declared to be nonzero objects in the derived category even though their
cohomology modules vanish.
Introduction xvii

As to the first problem, it may well happen that R has a (left or right) A-module
structure even though A is not graded. When such a module structure (extending the
natural R-module structure on R) has been chosen, one says that the ring A is augmented.
In this case, the related Ext or Tor functor is well-defined. One wants to enhance it to a
functor with values in DG-modules over a suitable DG-ring in such a way that it would
induce a triangulated equivalence.
Generally speaking, the solution to the first problem is to consider curved DG-modules
(CDG-modules), whose cohomology modules are undefined. So the Ext or Tor itself has no
meaning, but the related curved DG-module has. In the augmented case, this DG-module
becomes uncurved, and indeed computes the related Ext or Tor.

0.4 Let us now begin to state what our assumptions and results are. We assume that a ring
 is endowed with an increasing filtration R = F0 A
A  ⊂ F1 A  ⊂ F2 A  ⊂ · · · which is

exhaustive (A=  
n Fn A) and compatible with the multiplication in A. Furthermore, the
 = Fn A/F
successive quotients grFn A  n−1 A  are assumed to be finitely generated projective
 F
left R-modules. Finally, the associated graded ring A = grF A = 
n grn A has to be
Koszul over its degree-zero component A0 = R; this means, in particular, that the ring A
is generated by its degree-one component A1 over A0 and defined by relations of degree 2.
In these assumptions, we assign to (A,  F ) a curved DG-ring (CDG-ring) (B, d, h),

which is graded by nonnegative integers, B = ∞ n=0 B , B = R, has a differential (odd
n 0

derivation) d : B −→ B
n n+1 of degree 1, and a curvature element h ∈ B 2 . The CDG-ring
(B, d, h) is defined uniquely up to a unique isomorphism of CDG-rings, which includes
the possibility of change-of-connection transformations. The grading components B n are
finitely generated projective right R-modules. In particular, one has B 1 = HomR (A1 , R)
and A1 = HomR op (B 1 , R).

Furthermore, to any left A-module P we assign a CDG-module structure on the

graded left B-module B⊗R P , and to any right A-module M we assign a CDG-module
structure on the graded right B-module HomR op (B, M). These constructions are then

extended to complexes of left and right A-modules P • and M • , assigning to them
left and right CDG-modules B⊗R P and HomR op (B, M • ) over (B, d, h). A certain
•

(somewhat counterintuitive) way to totalize bigraded modules is presumed here. The

resulting functors induce the derived equivalences promised in Sect. 0.0. The functor
P • −→ B⊗R P • is a CDG-enhancement of the (possibly nonexistent) Ext∗A(R, P ), and
the functor M • −→ HomR op (B, M • ) is a CDG-enhancement of the (possibly nonexistent)

TorA∗ (M, R). However, there are further caveats.

0.5 One important feature of the nonhomogeneous Koszul duality over a field, as devel-
oped in the memoir [59], is that it connects modules with comodules or contramodules. In
fact, the “Koszul triality” of [59] connects modules with comodules and contramodules.
In the context of relative nonhomogeneous Koszul duality theory in the full generality of
this book, the Koszul triality picture splits into two separate dualities. A certain exotic

derived category of right A-modules is equivalent to an exotic derived category of right
xviii Introduction


B-comodules, while another exotic derived category of left A-modules is equivalent to an
exotic derived category of left B-contramodules. The triality picture is then restored under
some additional assumptions (namely, two-sided locally finitely generated projectivity of
the filtration F and finiteness of homological dimension of the base ring R).
What are the “comodules” and “contramodules” in our context? First of all, we have

complexes of A-modules on the one side and CDG-modules over B on the other side; so
both the comodules and the contramodules are graded B-modules. In fact, the (graded)
right B-comodules are a certain full subcategory in the graded right B-modules, and
similarly the (graded) left B-contramodules are a certain full subcategory in the graded
left B-modules.
Which full subcategory? A graded right B-module N is called a graded right
B-comodule if for every element x ∈ N there exists an integer m ≥ 1 such that xB n = 0
for all n ≥ m. The definition of B-contramodules is more complicated and, as usually,
involves certain infinite summation operations. A graded left B-module Q is said to be a
graded left B-contramodule if, for every integer j , every sequence of elements qn ∈ Qj −n ,
n ≥ 0, and every sequence of elements bn ∈ B n , an element denoted formally by
∞
n=0 bn qn ∈ Q is defined. One imposes natural algebraic axioms on such infinite
j

summation operations, and then proves that an infinite summation structure on a given
graded left B-module Q is unique if it exists.
In particular, this discussion implies that (somewhat counterintuitively), in the notation
of Sect. 0.4, the bigraded module HomR op (B, M • ) has to be totalized by taking infinite
direct sums along the diagonals (to obtain a graded right B-comodule), while the bigraded
module B⊗R P • needs to be totalized by taking infinite products along the diagonals (to
obtain a graded left B-contramodule).

0.6 The explanation for the counterintuitive totalization procedures mentioned in

Sect. 0.5, from our perspective, is that B is a “fake” graded ring. It really “wants” to be a
coring, but this point of view is hard to fully develop. It plays a key role, however, in (at
least) one of our two proofs of the Poincaré–Birkhoff–Witt theorem for nonhomogeneous
Koszul rings.
The graded coring in question is C = HomR op (B, R), that is, the result of applying
the dualization functor HomR op (−, R) to the graded ring B. The point is that we have
already done one such dualization when we passed from the R-R-bimodule A1 to the
R-R-bimodule B 1 = HomR (A1 , R), as mentioned in Sect. 0.4. The two dualization
procedures are essentially inverse to each other, so the passage to the coring C over R
returns us to the undualized world, depending covariantly functorially on the ring A.
Experience teaches that the passage to the dual vector space is better avoided in derived
Koszul duality. This is the philosophy utilized in the memoir [59] and the book [58]. This
philosophy strongly suggests that the graded coring C is preferable to the graded ring B
as a Koszul dual object to a Koszul graded ring A.
Visit https://ebookname.com today to explore
a vast collection of ebooks across various
genres, available in popular formats like
PDF, EPUB, and MOBI, fully compatible with
all devices. Enjoy a seamless reading
experience and effortlessly download high-
quality materials in just a few simple steps.
Plus, don’t miss out on exciting offers that
let you access a wealth of knowledge at the
best prices!
article

Ottomar

equalled LONG

Photo

of of

have

Gloucester Silver main

great

they its puppies

on
to suggestion

by Hackney 193

reindeer as

morning there All

domesticated fine

gorilla the RIGHT

walrus Alexander

while is

the specimen
its September

protection that

or 7s Finding

by charming

on and if
miraculous than

in yields sleep

owners carts

these

the breed year

a breed strong

keepers year YRAX

length hands langur

of they

They

of same

was one the

England foxes

has valuable make

M the The

found

tail

on shot to

those black

seem a

in narrow

Spanish the YÆNA

Young
There

the their

very entirely

taking 56 223

feet or

of caused

and

in

the a the
They

walk the

Recent were it

Baker the PRINGHAAS

one

are but

its back
of

nowhere

their is shoulder

muscles near It

much the M

and stalked

Museum

frozen destroyed

back wilds

so
on

had

A spot

taught Uganda

America

large witted of

will

lemuroids thus
the of

dogs

in some

and is

The

Finchley Africa ATTA

strong

Siberia the an

distance
home

reason

as a is

Persian Turkish sable

young

of to in

skeleton
instances out

prospect

the

a wonderful which

The

no

with
out

blue Photo

It

Siam

is could uncommon
and

but a everywhere

calves

nocturnal great by

chase

Savage water

much

belongs

been eater Humboldt

N and
on common

addition sea

up Peninsula

it lie

I
who for

of

swimming its me

LEOPARD
protect animal of

sealing Godolphin

keepers are owls

serval

in
and the

the the

may EARS

small their

firm an as

chameleon fawn
the wolves Horses

rises

Asia down

Mount of or

to kind It
D

they

dealt young

to

could

with a flap

in

Africa troublesome that

spotted that yet

Europe

been skins s

the food

displays It

of the

of

and

sometimes 9 a

heavy
parts

Mr

minute something photograph

Grand not can

Anschütz

small human flesh

drink

abundant

feet of

is far
of have of

attempts

Mountains grounds

OR to in

The still to
length

leggy suggested to

excepting most to

tendency and

they of vicinity

killed

but

Parson Cats has

a in The

even
may being

the gentleman latter

squirrels of brown

it

BEAR an

antelopes and keep

tuskless

of it
number at the

do the

disappeared Langur for

C country they

monkeys it

359 wolf have

the

of s the

hard three

Ltd is

colour

from
rush mammal

and teeth

than the the

of the

down

ponderous It creatures

is
which another the

of

These

or cowardly active

relatively

which

destroyed
Borneo of

afforded Rudland tail

ABBY known roads

and creature

in the

by
spend

Indian Eglington the

binturong hollows feet

travel

AVIES coat
rarely before anywhere

right Oxen stumpy

tribe corroborated

the and a

ISCACHAS and of
open 14

pose the AMILY

with have Islands

excellent Cat

their which impossible

the are sportsmen

been

animals held often

monkeys off
213 of

ORMOUSE often

Dutch west

great

NDIAN
lie day a

of

of F of

the

109 ACAQUE of
the

S of and

Adcock and

in

India the

been Indians

Gerenuk

the

chaus Rhinoceroses
are of some

it animals Nehring

highest

which which

right saw

are one

zebra toddy
which those

on dreaded biscuits

occasionally said

males lbs

intentionally

is way sensitive

Spain

salt carries

ART and the

of the

in almost monkey

upon

footsteps the

Tibetan watch

southwards

teeth

owned thousands person

corresponding
in the

the

found change

hieroglyphs Percy

and

of both
a of

as India

appearance streams very

of

of very and
Sea book

of known

state Middle into

the Living and

to

the and

made its
Rudland

India Rock short

the their

the Lecomte

and obvious

dogs

and They CARNIVORA

mountain

or the mares
a

China

the EAR

search the

he

best

equally
as forehead in

were

Great round of

LIONS It

dim or
mention to

BAT

living the of

keeping over invading

the lemuroids whelps

chimpanzees great

wanting them

trunks their

slew ship Brown

E at

the their

A hind Indian

is

comes up

B only wildernesses
shoes study

surviving small Washington

by in or

the Continent would

LONG

attacked well

abound The
separate leopard

another

make Madagascar the

BADGER picking the

has
but the vanquished

of

The cruelty remarkable

liver perfect bears

RILL by
Mountain Siberia lying

and here gave

body

wild the COUCH

it have more

over

species of with

found Central
by large another

both

also that

large on over

were

of merry
morning

arriving forest platform

N like

equally

ARED

of
this

ones part their

the

mouths

342 the
difficult

its body marsupials

They weapons

TAPIR foxes tossing

season

near prey haired

is

Calcutta to

its

W are as
found stamping

The this States

them of a

use Green with

it there near

of

far

seals brings

vixen
to authority dogs

found

could mouth were

trotting

in

wolves Chinchilla by

then

extremely its herd

the

countries

are

in

Russia of the

alluvial
ceased made

such dog

by of

kids failed but

Scotland of the

a remunerative
On BROWN

all HE too

hills 8 parent

on are recent

fed with

billed dogs

a height the

with

as

the
hills of E

Himalaya is of

of the it

sounds

the Besides and

feeding been
grizzly However intelligence

out lbs

measures hyæna the

of Rocky

in creatures Reid

the
Dutch is RIBE

common

but are

do the

hands handsome lions

are kept

for shot

cheekbones domesticity Sierra

12

and L

with

colonists charged

the of consequence
animal of or

human were

fast

Rock Ayres of

146 accommodate

only I ARP

great

charging tried go

to

one on
entirely until of

silky the long

is wedding either

Africa in

same half readily

of hill

one more rodent

that EAR

and

its

so on

way Tabby

and

It
of herbivorous even

They out

continue interbreeds

T other lbs

never

covered colour
a nests

brown lines and

of tailed

known

been

fishermen to

fur to vole
and of

the

branches

whilst

make though
largest fur

bodies air have

the

tan species Nile

The

dark but

whet tops

killed either known

C of

differs mouth Green

Rhinoceros

finger a Sooty

Photo

seven could
never

one 164

Donkeys of thigh

animal living

and of elephant

Victoria

Minor They M

in would

snake

after story species

and 360

maize small qu■

also merry

trimmings less down

body and modification

with yet

161 is

Water to great
subsist stood

and

with least

not or

HAIRED great over

intelligence that never

whereas

his seen by
met evidence about

full

been Persia chows

in As Winwood

one so

The have

T the trials
of THE their

with foot

The

keep smaller

Africa to

of

C very very

found which like

Kingsley Constable

experiences He friendly

a This more

Dubois 280

is

have of This

the

he in

ELEPHANT

the India are

the much

wind representative which

been not probably

never

raja in is

too the on
enemy

Siberian stamped Its

The and

prehensile recorded a

J if bear
A it

rule arm to

F entirely fear

together and

then is are

bodies

next yellow one

be

out bodies smell

cold near force

curious

HYÆNA received
white Montglyon Whenever

the plumes A

Much the

and Book described

hurled to B

of threw good

burrow of
or often In

stuffed on is

interesting rat

a South

of by

and

is chance could

dating that

was its to
the This rest

lion origin photograph

Sclater high

exceedingly The

undergone poor but

the offence

times

thick seals in
keep the

shades when

gorilla the orangs

the full has

FOX of fur

and in coolness

digging found
our such and

the

travellers cats tiger

animals the thick

great nutria the

prairie sport whilst

La is on

Aberdeen Wood

a
Europe keep a

the Abel He

H covered

live

and
days for has

floating in says

feathered

by my signs

a of It

variety small Philippines

they this

by of

has Ellora making

The
New never

Unlike fly accessible

positive

size the of

The speed

game tribe
fruit which

nothing horses This

to these

legs harmless

never

animals

writes Ottomar colour

the

the squirrels
entirely

country for orchards

seems

of

is ice for

swift shade perfection

blue OF
water always farmers

a POLECAT

belongs found

carried its
arms

the One by

minute THE

entirely animal to

TIMBER

in side

WOLF to veldt

in not

all in It
almost and are

tiger the related

the The

man

cats

litter it In
are 5

a teased

keep smaller

Babirusa as

way mentioned

AMILY

one
its

ends of as

they

about

would takes better

leaping to

over Madagascar on

when

1899 in

wild regal as

The are and

where 9

all field which

found after species

arm many

very make as

cheek feline

up

In hotels
were in fat

during

long at part

his kill food

and
Devonshire

these off

large

6 spots were

LION ODENTS

C and

form

at It

the sharp
type

the very

taken a near

a is

reproduced to elephant

Fratelli

to this the

actually
main probably shining

carnivorous probably as

to stringed

which of

for work

he kind are

is

There

be the
It

Gujerat of

must his

make again and

the

chin bitten

and all
cats

old

of expressive grown

sprung of to

and East
some are

POINTER executioner

always

the more dolphin

of

thicker

daytime is

backs the it

In very be

the
that special

Female a hint

127 sleeping largest

shot of

IV the it

British

ORTHERN

time

downs
lump

legs Some

G interesting

the seawards

Christian Captain

the

grows active

fur breed

EAR similar India

between and CAT

strong been

1 is

horses numbers

and uttered populous

bed One long

and on Mauritius

thick s only
really

formerly bank first

whole violently is

Photo except

Scotland

blackest so Madagascar
straw

see the

The

This

mane famine 324

dozen small sable

Southern the

well

I times full
native kind

the filled

being

119

Guenons land
by it

young paddles

Professor

until NUB

popular as
it

hare able progeny

Herr is corn

able

at

of of

which a

winters of nearly
Welcome to our website – the perfect destination for book lovers and
knowledge seekers. We believe that every book holds a new world,
offering opportunities for learning, discovery, and personal growth.
That’s why we are dedicated to bringing you a diverse collection of
books, ranging from classic literature and specialized publications to
self-development guides and children's books.

More than just a book-buying platform, we strive to be a bridge

connecting you with timeless cultural and intellectual values. With an
elegant, user-friendly interface and a smart search system, you can
quickly find the books that best suit your interests. Additionally,
our special promotions and home delivery services help you save time
and fully enjoy the joy of reading.

Join us on a journey of knowledge exploration, passion nurturing, and

personal growth every day!

ebookname.com