01.02.35 · foundations / algebra

Dold-Kan correspondence

shipped3 tiersLean: none

Anchor (Master): Dold 1958 *Ann. Math.* 68 (54-80); Kan 1958 *Proc. NAS* 41 (1092-1096); Goerss-Jardine Ch. III §2 with the full Eilenberg-MacLane splitting; May 1967 *Simplicial Objects in Algebraic Topology* §22; Schwede-Shipley 2003 *Topology* 42 on the monoidal refinement

Intuition Beginner

A simplicial abelian group is a sequence of abelian groups $A_{0}, A_{1}, A_{2}, \dots$ together with face maps $d_{i}$ and degeneracy maps $s_{i}$ relating each level to its neighbours. A non-negatively graded chain complex is a sequence of abelian groups $C_{0}, C_{1}, C_{2}, \dots$ together with a single differential $d_{n} : C_{n} \to C_{n - 1}$ such that applying the differential twice gives zero. These two structures look quite different on the surface: one has many face and degeneracy maps, the other has one differential per level.

The Dold-Kan correspondence says they are the same data, packaged two different ways. From a simplicial abelian group you can extract a chain complex by taking, in degree $n$ , the intersection of the kernels of all face maps except the last; the last face map descends to a differential on this intersection. Going the other direction, every chain complex can be assembled back into a simplicial abelian group whose face and degeneracy maps reproduce the chain complex when you apply the extraction. The two operations are inverse to one another, so the categories of simplicial abelian groups and non-negatively graded chain complexes are equivalent.

Why this matters. Simplicial homotopy theory and homological algebra were developed independently in the 1940s and 1950s and used different machinery: one built from face-degeneracy combinatorics, the other from chain-level homological calculations. The correspondence makes the bridge precise. Every theorem about chain complexes lifts to a theorem about simplicial abelian groups, and vice versa. The homology of the chain complex matches the simplicial homotopy groups of the simplicial abelian group in every degree.

Visual Beginner

A schematic diagram showing a simplicial abelian group on the left, drawn as a vertical column of abelian groups $A_{0}, A_{1}, A_{2}, A_{3}$ with face-and-degeneracy arrows going both up and down between adjacent levels. In the middle, a pair of horizontal arrows labelled $N$ and $K$ indicate the two halves of the equivalence. On the right, the same column of abelian groups labelled $N_{0} A, N_{1} A, N_{2} A, N_{3} A$ has a single downward arrow at each level, the chain-complex differential. The picture captures that the rich face-degeneracy structure on the left compresses, without loss of information, into the single differential on the right.

The arrow $N$ on top is the normalised-chain functor, which extracts the chain complex by intersecting kernels of early face maps. The arrow $K$ on the bottom is the Kan extension, which rebuilds the simplicial abelian group from a chain complex by attaching enough degenerate simplices in each degree to support all the missing face-degeneracy relations.

Worked example Beginner

Compute the chain complex that the normalised-chain functor extracts from the Eilenberg-MacLane simplicial abelian group $K (Z, 1)$ , and check that the answer is exactly the chain complex with a single copy of $Z$ in degree one and zero everywhere else.

Step 1. The Eilenberg-MacLane simplicial abelian group $K (Z, 1)$ in degree $n$ is the group of normalised $1$ -cocycles on the standard $n$ -simplex with values in $Z$ . Concretely, an element of $K (Z, 1)_{n}$ is an assignment $a : {0, 1, \dots, n}^{2} \to Z$ with $a (i, i) = 0$ and $a (i, j) + a (j, k) = a (i, k)$ for all $i, j, k$ . Such an assignment is determined by the values $a (i, i + 1)$ for $0 \leq i \leq n - 1$ , giving $K (Z, 1)_{n} ≅ Z^{n}$ for $n \geq 1$ and $K (Z, 1)_{0} = 0$ .

Step 2. The face maps act by pulling back along the standard face inclusions; the degeneracy maps insert a copy of the previous value. The first face map $d_{0}$ drops the first edge $a (0, 1)$ ; the middle face maps $d_{i}$ for $1 \leq i \leq n - 1$ collapse edge $i$ into edge $i + 1$ ; the last face map $d_{n}$ drops the final edge $a (n - 1, n)$ .

Step 3. The normalised chain complex $N_{n} A$ in degree $n$ is the intersection of the kernels of $d_{0}, d_{1}, \dots, d_{n - 1}$ . In degree $0$ the intersection is empty (no face maps to intersect), so $N_{0} K (Z, 1) = K (Z, 1)_{0} = 0$ . In degree $1$ the only face map is $d_{0}$ which drops the unique edge value, and its kernel inside $K (Z, 1)_{1} = Z$ is the whole of $Z$ when we instead intersect $ker d_{0}$ taken as $ker (K (Z, 1)_{1} \to K (Z, 1)_{0}) = Z$ since the target is zero. So $N_{1} K (Z, 1) = Z$ .

Step 4. In degree $n \geq 2$ the intersection $⋂_{i = 0}^{n - 1} ker d_{i}$ forces every edge value to be zero, since the system of face conditions equates each consecutive edge to a sum involving prior edges and the constraint that all of these vanish leaves only the zero element. So $N_{n} K (Z, 1) = 0$ for $n \geq 2$ .

Step 5. The differential is $d_{n}$ restricted to $N_{n} A$ . In degree $1$ there is no degree $2$ to map from, so the differential into $N_{1}$ is automatically zero; the differential out of $N_{1}$ goes into $N_{0} = 0$ , so it is also zero. The result is the chain complex with $Z$ concentrated in degree $1$ , exactly the chain complex denoted $Z [1]$ .

What this tells us. The Eilenberg-MacLane object $K (Z, 1)$ sits over the chain complex with a single $Z$ in degree one through the Dold-Kan correspondence, and the inverse functor $K$ sends $Z [1]$ back to $K (Z, 1)$ . The construction is mechanical once the intersection of kernels is computed, and the answer matches the topologist's expectation: the simplicial model of a $K (π, n)$ corresponds under Dold-Kan to the chain complex with $π$ concentrated in degree $n$ .

Check your understanding Beginner

Exercise (easy, multiple choice).

Under the Dold-Kan correspondence, simplicial abelian groups correspond to which kind of chain complex?

A. Bounded chain complexes
B. Unbounded chain complexes
C. Non-negatively graded chain complexes
D. Chain complexes of vector spaces

Hint

A simplicial object lives in non-negative simplicial degrees $0, 1, 2, \dots$ ; the corresponding chain complex inherits the same grading.

Answer

C. Non-negatively graded chain complexes.

The correspondence is between $s Ab$ , simplicial abelian groups, and $Ch_{\geq 0} (Ab)$ , chain complexes graded by the non-negative integers. The bound below is forced by the simplicial side, which begins at degree $0$ and grows upward. Unbounded chain complexes correspond instead to spectra and stable homotopy categories, not to ordinary simplicial abelian groups.

Formal definition Intermediate+

Let $A$ be an abelian category 01.02.33. Write $Δ$ for the simplex category whose objects are the finite ordinals $[n] = {0 < 1 < \dots < n}$ and whose morphisms are order-preserving maps. A simplicial object in $A$ is a functor $A : Δ^{op} \to A$ . The category of simplicial objects in $A$ is denoted $s A$ . When $A = Ab$ , the category $s Ab$ is the category of simplicial abelian groups 03.12.25.

A simplicial object $A$ assigns to each ordinal $[n]$ an object $A_{n} = A ([n])$ , and to each order-preserving map an arrow in $A$ . The structural maps split into face operators $d_{i} : A_{n} \to A_{n - 1}$ for $0 \leq i \leq n$ (the contravariant image of the $i$ -th coface inclusion $[n - 1] ↪ [n]$ that skips $i$ ) and degeneracy operators $s_{i} : A_{n} \to A_{n + 1}$ for $0 \leq i \leq n$ (the contravariant image of the $i$ -th codegeneracy surjection $[n + 1] ↠ [n]$ that repeats $i$ ). These satisfy the standard simplicial identities including $d_{i} d_{j} = d_{j - 1} d_{i}$ for $i < j$ and $s_{i} s_{j} = s_{j + 1} s_{i}$ for $i \leq j$ .

The unnormalised chain complex of $A \in s A$ is the chain complex $C_{*} A$ with $C_{n} A = A_{n}$ and differential $\partial_{n} = \sum_{i = 0}^{n} (- 1)^{i} d_{i} : A_{n} \to A_{n - 1}$ . The simplicial identity $d_{i} d_{j} = d_{j - 1} d_{i}$ for $i < j$ produces $\partial^{2} = 0$ after the alternating-sum cancellation, so $C_{*} A$ is a chain complex in $Ch_{\geq 0} (A)$ .

The normalised chain complex of $A$ is the subcomplex $N_{*} A \subseteq C_{*} A$ given by $$ N_n A = \bigcap_{i = 0}^{n - 1} \ker(d_i : A_n \to A_{n-1}) $$ with the differential induced from $C_{*}$ , which on $N_{n} A$ reduces to $(- 1)^{n} d_{n}$ since the other face terms vanish on the intersection. Equivalently, $N_{*} A$ is the quotient of $C_{*} A$ by the subcomplex $D_{*} A$ spanned by the degenerate simplices, that is, by the images of the degeneracy operators $s_{i}$ . The two constructions agree because the simplicial identities supply an explicit splitting $$ C_* A = N_* A \oplus D_* A. $$ This decomposition uses the Eilenberg-MacLane operator $E_{n} : C_{n} A \to N_{n} A$ , an idempotent built as an alternating sum of compositions of degeneracy-face pairs, originating in Eilenberg-MacLane's 1953 paper on the groups $H (π, n)$ ^{[Eilenberg-MacLane 1953]}.

The normalised-chain functor $$ N : s\mathcal{A} \to \mathrm{Ch}{\ge 0}(\mathcal{A}), \qquad A \mapsto N* A, $$ is the data-extracting half of the correspondence. Its right adjoint (and inverse up to natural isomorphism) is the functor $$ K : \mathrm{Ch}{\ge 0}(\mathcal{A}) \to s\mathcal{A} $$ defined on a chain complex $(C, \partial)$ by the formula $$ K(C_)n = \bigoplus{[n] \twoheadrightarrow [k]} C_k, $$ where the sum runs over all order-preserving surjections from $[n]$ to $[k]$ for $0 \leq k \leq n$ , and the face-degeneracy structure is determined by Kan extension along the inclusion of the chain-complex category into the category of simplicial objects via the standard simplex generators. Concretely, $K$ realises a chain complex as the unique simplicial abelian group with that normalised chain complex and with degenerate simplices freely generated.

Counterexamples to common slips

The unnormalised chain complex $C_{*} A$ and the normalised chain complex $N_{*} A$ are not canonically isomorphic, only chain-homotopy equivalent. Identifying them as if they were equal misses the Eilenberg-MacLane idempotent that distinguishes degenerate from non-degenerate simplices.
The Dold-Kan correspondence is an equivalence of ordinary categories, not just of homotopy categories. The unit $A \to K N A$ and the counit $N K (C_{*}) \to C_{*}$ are isomorphisms on the nose, not merely quasi-isomorphisms.
The correspondence does not generalise to unbounded chain complexes via the same construction: $K$ produces only non-negatively-supported simplicial objects, so the equivalence is genuinely with $Ch_{\geq 0} (A)$ , not with the full category $Ch (A)$ . Spectral lifts via stable model structures are required to extend to the unbounded case.

Key theorem with proof Intermediate+

Theorem (Dold-Kan; Dold 1958, Kan 1958). Let $A$ be an idempotent-complete additive category — in particular any abelian category 01.02.33. The normalised-chain functor $N$ and the Kan-extension functor $K$ are mutually quasi-inverse equivalences of categories: $$ N : s\mathcal{A} \xrightarrow{;\sim;} \mathrm{Ch}{\ge 0}(\mathcal{A}), \qquad K : \mathrm{Ch}{\ge 0}(\mathcal{A}) \xrightarrow{;\sim;} s\mathcal{A}, $$ with unit $η : id_{s A} \sim K \circ N$ and counit $ε : N \circ K \sim id_{Ch_{\geq 0} (A)}$ , both natural isomorphisms.

Proof. The argument has three steps: construct the splitting $C_{*} A = N_{*} A \oplus D_{*} A$ that isolates the normalised part, verify that $K$ rebuilds a simplicial object whose normalised complex is the input, and check that the constructions are mutually inverse.

Step 1: the Eilenberg-MacLane splitting. For each $n$ define the degenerate subobject $D_{n} A \subseteq C_{n} A$ as the image of $\sum_{i} s_{i} : ⨁_{i = 0}^{n - 1} A_{n - 1} \to A_{n}$ , that is, the sub-abelian-group (or subobject in $A$ ) generated by everything that is a degeneracy of a lower-dimensional element. The simplicial identity $d_{i} s_{j} = s_{j - 1} d_{i}$ for $i < j$ , together with $d_{i} s_{i} = id = d_{i + 1} s_{i}$ , gives an explicit idempotent $$ E_n : C_n A \to C_n A, \qquad E_n = (\mathrm{id} - s_{n-1} d_{n-1}) (\mathrm{id} - s_{n-2} d_{n-2}) \cdots (\mathrm{id} - s_0 d_0) $$ whose image is exactly $N_{n} A = ⋂_{i < n} ker d_{i}$ and whose kernel is exactly $D_{n} A$ . Idempotence is a direct simplicial-identity calculation expanding the product; image equal to the intersection of kernels follows from the rightmost factor $(id - s_{0} d_{0})$ killing the image of $s_{0}$ and the inductive application of the remaining factors handling $s_{1}, \dots, s_{n - 1}$ . Idempotent-completeness of $A$ ensures $E_{n}$ splits as a direct-sum decomposition $C_{n} A = N_{n} A \oplus D_{n} A$ ; in an abelian category this splitting is automatic.

The differential of $C_{*} A$ respects this decomposition. On $N_{n} A$ , all face terms with $i < n$ vanish by definition of the kernel intersection, leaving $\partial ∣_{N_{n} A} = (- 1)^{n} d_{n}$ . The image of $\partial ∣_{N_{n} A}$ in $C_{n - 1} A$ lies in $N_{n - 1} A$ because each remaining face $d_{i}$ with $i < n - 1$ applied after $d_{n}$ gives $d_{i} d_{n} = d_{n - 1} d_{i}$ , which vanishes on $N_{n} A$ . So $N_{*} A$ is a subcomplex of $C_{*} A$ , and the projection $C_{*} A \to N_{*} A$ along $D_{*} A$ is a chain map.

Step 2: $K$ on a chain complex. Given $(C_{*}, \partial) \in Ch_{\geq 0} (A)$ , set $$ K(C_*)n = \bigoplus{\sigma : [n] \twoheadrightarrow [k], ; 0 \le k \le n} C_k^{(\sigma)}, $$ indexed by order-preserving surjections from $[n]$ to $[k]$ , with one copy of $C_{k}$ for each surjection $σ$ . The face and degeneracy operators are defined by Kan-extension formulae: a morphism $α : [m] \to [n]$ in $Δ$ acts on $σ : [n] ↠ [k]$ via the epi-mono factorisation $σ \circ α = μ \circ τ$ with $τ : [m] ↠ [j]$ surjective and $μ : [j] ↪ [k]$ injective. The induced map sends $C_{k}^{(σ)}$ to $C_{j}^{(τ)}$ as follows: if $μ$ is the identity (so $j = k$ ), use $id_{C_{k}}$ ; if $μ$ is the unique face map omitting the largest element of $[k]$ (so $j = k - 1$ ), use the differential $\partial : C_{k} \to C_{k - 1}$ ; in all other cases the induced map is zero. This collection of structural maps satisfies the simplicial identities by direct case-check on epi-mono factorisations in $Δ$ .

In the construction the summand indexed by $σ = id_{[n]}$ is the unique non-degenerate piece $C_{n}^{(id)} ≅ C_{n}$ , and every other summand $C_{k}^{(σ)}$ with $σ$ a non-identity surjection is degenerate: it lies in the image of the corresponding degeneracy operator on $K (C_{*})_{n - 1}$ . So $D_{n} K (C_{*}) = ⨁_{σ \neq = id} C_{k}^{(σ)}$ and $N_{n} K (C_{*}) = C_{n}^{(id)} ≅ C_{n}$ . The differential on $N_{n} K (C_{*})$ obtained by restricting $\sum (- 1)^{i} d_{i}$ matches $\partial : C_{n} \to C_{n - 1}$ up to the sign $(- 1)^{n}$ , which is absorbed into the standard sign convention. So $N \circ K ≅ id_{Ch_{\geq 0} (A)}$ via a natural isomorphism, establishing the counit.

Step 3: $K \circ N ≅ id_{s A}$ . For $A \in s A$ , the unit $η_{A} : A \to K (N A)$ is built from the Eilenberg-MacLane splitting. On $A_{n} = N_{n} A \oplus D_{n} A$ , the normalised part maps to the identity summand $N_{n} A^{(id)}$ of $K (N A)_{n}$ , and the degenerate part $D_{n} A$ decomposes by the structure-theorem for degenerate simplices: every degenerate element is uniquely a degeneracy of a non-degenerate simplex of strictly smaller dimension, and that lower-dimensional non-degenerate part lives in some $N_{k} A$ for $k < n$ . The component of $D_{n} A$ coming from $N_{k} A$ via the surjection $σ : [n] ↠ [k]$ maps to $N_{k} A^{(σ)} = C_{k}^{(σ)}$ in $K (N A)_{n}$ . The structure-theorem assertion that this decomposition exists and is unique is the Eilenberg-Zilber lemma for degenerate simplices, a combinatorial statement about the epi-mono factorisation system in $Δ$ . The resulting map $η_{A}$ is a level-wise isomorphism by construction: source and target have matching direct-sum decompositions, and the components agree. Naturality in $A$ is immediate from naturality of the splitting.

Both unit and counit are natural isomorphisms, so $(N, K)$ is an adjoint equivalence between $s A$ and $Ch_{\geq 0} (A)$ . $□$

Bridge. The Dold-Kan correspondence builds toward every comparison between simplicial homotopy theory and homological algebra. The foundational reason it holds is the Eilenberg-MacLane splitting $C_{*} A = N_{*} A \oplus D_{*} A$ : every simplicial abelian group decomposes once-and-for-all into a non-degenerate normalised part and a degenerate part freely generated by the degeneracy operators, and the normalised part is exactly a chain complex. This is exactly the algebraic packaging of the geometric fact that a simplicial set has a canonical filtration by non-degenerate simplices, and identifies the homological-algebra category $Ch_{\geq 0} (Ab)$ with the homotopy-theoretic category $s Ab$ on the nose, not merely up to quasi-isomorphism. The correspondence appears again in 03.12.25 (simplicial sets), where the singular simplicial set of a topological space, after free abelianisation, becomes a simplicial abelian group whose normalised chain complex is the singular chain complex; the Dold-Kan correspondence is exactly what makes "simplicial chain complex" and "singular chain complex" canonically the same object. The central insight is that the simplicial identities are not extra structure beyond a chain complex — they are exactly the bookkeeping needed to record the degeneracy-degenerate part of a chain complex, and the equivalence makes this bookkeeping invertible. Putting these together, every functorial construction on chain complexes — Tor, Ext, derived tensor products, derived Hom — lifts uniquely to a functorial construction on simplicial abelian groups, and the bridge is the canonical equivalence $N ⊣ K$ .

Exercises Intermediate+

Exercise 3 (medium, symbolic).

Show that the unnormalised differential $\partial = \sum_{i = 0}^{n} (- 1)^{i} d_{i} : C_{n} A \to C_{n - 1} A$ satisfies $\partial \circ \partial = 0$ using only the simplicial identity $d_{i} d_{j} = d_{j - 1} d_{i}$ for $i < j$ .

Hint

Expand $\partial \circ \partial = \sum_{i, j} (- 1)^{i + j} d_{i} d_{j}$ and split the double sum into $i < j$ and $i \geq j$ terms; apply the simplicial identity to one of them.

Answer

Expand $$ \partial \circ \partial = \sum_{j = 0}^n \sum_{i = 0}^{n - 1} (-1)^{i + j} d_i d_j. $$ Split the double sum into $i < j$ and $i \geq j$ pieces. For the $i < j$ terms, apply the simplicial identity $d_{i} d_{j} = d_{j - 1} d_{i}$ : $$ \sum_{i < j} (-1)^{i + j} d_i d_j = \sum_{i < j} (-1)^{i + j} d_{j - 1} d_i. $$ Substitute $j^{'} = j - 1$ so $i \leq j^{'}$ in the new index, picking up an extra sign $(- 1)^{- 1}$ that flips the overall sign: $$ \sum_{i < j} (-1)^{i + j} d_{j-1} d_i = -\sum_{i \le j'} (-1)^{i + j'} d_{j'} d_i. $$ For the $i \geq j$ piece in the original sum, re-index by writing $j$ as the inner-position index and $i$ as the outer-position, giving $\sum_{i \geq j} (- 1)^{i + j} d_{i} d_{j}$ , which equals $\sum_{j \leq i} (- 1)^{i + j} d_{i} d_{j}$ . The two sums cancel pairwise: the $(i, j)$ term of the first equals minus the $(j^{'}, i)$ term of the second with the appropriate renaming. So $\partial^{2} = 0$ . Rubric: full credit for the index-swap argument with sign tracking; partial credit for setting up the double sum without the cancellation.

Exercise 4 (medium, short-answer).

Show that the inclusion $N_{*} A ↪ C_{*} A$ is a chain-homotopy equivalence, with the projection along $D_{*} A$ as homotopy inverse.

Hint

The composition $N_{*} ↪ C_{*} ↠ N_{*}$ is the identity; the composition $C_{*} ↠ N_{*} ↪ C_{*}$ is the Eilenberg-MacLane idempotent $E$ . Exhibit a chain homotopy from $E$ to $id_{C_{*}}$ using the degeneracy operators.

Answer

Define $h_{n} : C_{n} A \to C_{n + 1} A$ by $h_{n} = \sum_{k = 0}^{n} s_{k} \cdot E_{n}^{(k)}$ , where $E_{n}^{(k)} = (id - s_{n - 1} d_{n - 1}) \dots (id - s_{k} d_{k})$ is the partial Eilenberg-MacLane idempotent killing degeneracies $s_{k}, \dots, s_{n - 1}$ but not earlier ones. A direct simplicial-identity calculation gives $\partial h_{n} + h_{n - 1} \partial = id_{C_{n} A} - E_{n}$ , where $E_{n} = E_{n}^{(0)}$ is the full idempotent. So $h$ is a chain homotopy from the identity to the idempotent $E$ projecting onto $N_{*} A$ . The inclusion $ι : N_{*} A ↪ C_{*} A$ and the projection $π : C_{*} A ↠ N_{*} A$ satisfy $π \circ ι = id_{N_{*} A}$ exactly and $ι \circ π = E_{n}$ chain-homotopic to $id_{C_{*} A}$ via $h$ . So $ι$ and $π$ are mutually inverse chain-homotopy equivalences. The precise formula for $h$ originates in Eilenberg-MacLane 1953 ^{[Eilenberg-MacLane 1953]}. Rubric: full credit for the homotopy formula plus the verification that $E$ is homotopic to the identity; partial credit for stating the existence without an explicit formula.

Exercise 5 (medium, symbolic).

Compute the normalised chain complex of the simplicial abelian group $Z [Δ^{1}]$ , the free simplicial abelian group on the standard simplex $Δ^{1}$ , and identify the result.

Hint

The standard simplex $Δ^{n}$ has non-degenerate simplices in degrees $0$ through $n$ . The free abelian-group functor $Z [-]$ converts a simplicial set to a simplicial abelian group level-wise.

Answer

The standard simplex $Δ^{1}$ has two non-degenerate vertices $v_{0}, v_{1}$ in degree $0$ and one non-degenerate edge $e$ in degree $1$ . In higher degrees every simplex is degenerate. So $Z [Δ^{1}]_{0} = Z^{2}$ generated by $v_{0}, v_{1}$ ; $Z [Δ^{1}]_{1} = Z^{3}$ generated by the edge $e$ and the two degeneracies $s_{0} v_{0}, s_{0} v_{1}$ ; and in degree $n \geq 2$ every generator is degenerate, so the simplicial abelian group has $Z^{n + 2}$ generators all in the degenerate part. The normalised chain complex extracts exactly the non-degenerate part: $N_{0} = Z^{2}$ , $N_{1} = Z$ generated by $e$ , and $N_{n} = 0$ for $n \geq 2$ . The differential $d_{1} : N_{1} \to N_{0}$ acts as $e \mapsto v_{1} - v_{0}$ . The resulting chain complex is $Z (v_{1} - v_{0}) Z^{2}$ , which is the cellular chain complex of the unit interval. Its homology is $H_{0} = Z$ (one path component, generated by either vertex modulo the boundary) and $H_{n} = 0$ for $n \geq 1$ . Rubric: full credit for the chain complex with differential, plus identification with the interval; partial credit for the chain groups alone.

Exercise 6 (medium, numeric).

For a chain complex $C_{*}$ concentrated in degrees $0$ and $1$ with $C_{0} = Z$ , $C_{1} = Z$ , and differential $\partial : C_{1} \to C_{0}$ given by multiplication by $2$ , compute the rank of $K (C_{*})_{2}$ .

Hint

The formula $K (C_{*})_{n} = ⨁_{[n] ↠ [k]} C_{k}$ sums over order-preserving surjections.

Answer

$5$ . The surjections $[2] ↠ [k]$ are: one identity surjection $[2] \to [2]$ (so $k = 2$ , giving $C_{2} = 0$ , contributing $0$ ); three surjections $[2] \to [1]$ (mapping $(0, 1, 2)$ to $(0, 0, 1)$ , $(0, 1, 1)$ , or $(0, 0, 1)$ — concretely there are exactly two non-identity surjections $[2] ↠ [1]$ given by the two degeneracies $s_{0}, s_{1}$ , plus the identity, so three including the identity; for $k = 1$ specifically there are exactly two surjections $[2] ↠ [1]$ , namely $s_{0}$ and $s_{1}$ , each contributing $C_{1} = Z$ , total rank $2$ ); and one surjection $[2] \to [0]$ (the constant map, contributing $C_{0} = Z$ , rank $1$ ). Adding the contributions: $0 + 2 \cdot 1 + 1 = 3$ — wait, the count of order-preserving surjections $[n] ↠ [k]$ is $(k n)$ . For $n = 2$ : $(2 2) = 1$ identity, $(1 2) = 2$ surjections to $[1]$ , $(0 2) = 1$ surjection to $[0]$ . So $K (C_{*})_{2} = C_{2} \oplus C_{1}^{\oplus 2} \oplus C_{0} = 0 \oplus Z^{2} \oplus Z = Z^{3}$ , rank $3$ . The answer is $3$ . Rubric: full credit for $3$ with the surjection count; partial credit for the formula without the count.

Exercise 7 (hard, short-answer).

State and outline a proof of the Eilenberg-Zilber theorem in its Dold-Kan form: for simplicial abelian groups $A, B \in s Ab$ , the normalised chain complex of the diagonal tensor product is quasi-isomorphic to the tensor product of the normalised chain complexes: $$ N(A \otimes B) \xrightarrow{;\sim;} NA \otimes NB. $$

Hint

Use the shuffle map (Eilenberg-MacLane) and the Alexander-Whitney map as mutually-inverse-up-to-homotopy chain-level comparisons. Reduce to the unnormalised statement and use the Eilenberg-MacLane idempotent.

Answer

Define the shuffle map $\nabla : N A \otimes N B \to N (A \otimes B)$ on simplicial generators by $$ \nabla(a \otimes b) = \sum_{(\mu, \nu)} \mathrm{sgn}(\mu, \nu) , s_\nu(a) \otimes s_\mu(b), $$ where the sum runs over $(p, q)$ -shuffles of $[p + q]$ — pairs $(μ, ν)$ of complementary order-preserving injections $μ : [p] ↪ [p + q]$ , $ν : [q] ↪ [p + q]$ — and $s_{μ}, s_{ν}$ are the iterated degeneracy operators along the complements. The Alexander-Whitney map $AW : N (A \otimes B) \to N A \otimes N B$ acts on a $p + q$ -simplex $(a, b)$ by $$ \mathrm{AW}(a \otimes b) = \sum_{p + q = n} \overline{d}^q(a) \otimes \underline{d}^p(b), $$ where $\overline{d}^{q} = d_{n} d_{n - 1} \dots d_{p + 1}$ deletes the last $q$ vertices and $\underline{d}^{p} = (d_{0})^{p}$ deletes the first $p$ vertices. Two facts: (i) $AW \circ \nabla = id_{N A \otimes N B}$ on the nose, by the shuffle-Alexander-Whitney telescope identity; and (ii) $\nabla \circ AW$ is chain-homotopic to $id_{N (A \otimes B)}$ via an explicit homotopy first constructed by Eilenberg-MacLane 1953 ^{[Eilenberg-MacLane 1953]}. Combining (i) and (ii) gives the quasi-isomorphism $N (A \otimes B) ≃ N A \otimes N B$ . The unnormalised statement $C_{*} (A \otimes B) ≃ C_{*} A \otimes C_{*} B$ follows by the chain-homotopy equivalence between $C_{*}$ and $N_{*}$ on each factor (Exercise 4). Rubric: full credit for both formulas plus the statement of the homotopy; partial credit for the shuffle map alone.

Exercise 8 (hard, short-answer).

Show that the Dold-Kan correspondence is not strict-monoidal: $N (A \otimes B) \neq = N A \otimes N B$ in general, although the two are quasi-isomorphic via the shuffle map. Discuss the consequence for the monoidal refinement (Schwede-Shipley 2003) and explain why the Quillen equivalence between simplicial commutative rings and connective $E_{\infty}$ -algebras requires more than the underlying additive equivalence.

Hint

Compute both sides for $A = B = K (Z, 1)$ and observe that one has $Z$ in degree $2$ while the other has $Z$ in degree $1 + 1 = 2$ but with multiplicity given by the shuffle count, which exceeds the rank in $N A \otimes N B$ .

Answer

For $A = B = K (Z, 1)$ : $N A = N B = Z [1]$ (one $Z$ in degree $1$ ), so $N A \otimes N B = Z [2]$ (one $Z$ in degree $2$ ). The diagonal tensor product $A \otimes B$ as simplicial abelian groups has $(A \otimes B)_{n} = A_{n} \otimes B_{n}$ , and the unnormalised chain complex $C_{*} (A \otimes B)$ has many redundant degeneracy-degenerate generators in each degree, even after normalisation. The shuffle map identifies $Z [2] = N A \otimes N B$ as a subcomplex of $N (A \otimes B)$ , with the Eilenberg-Zilber chain homotopy showing the inclusion is a quasi-isomorphism but not an equality. The Alexander-Whitney map runs in the opposite direction; its composition with the shuffle is the identity on $N A \otimes N B$ , but the reverse composition is only chain-homotopic to the identity, not equal.

Schwede-Shipley 2003 ^{[Schwede-Shipley 2003]} resolve the monoidal issue by establishing a Quillen equivalence between simplicial $R$ -modules and non-negatively graded chain complexes of $R$ -modules as monoidal model categories, where the equivalence is at the level of homotopy categories and the underlying additive equivalence is upgraded to a homotopical-monoidal equivalence via the shuffle / Alexander-Whitney machinery. The consequence for $E_{\infty}$ -algebras: a simplicial commutative ring corresponds, under Dold-Kan plus the Schwede-Shipley refinement, to a connective $E_{\infty}$ -algebra in chain complexes, with the strict commutative multiplication on the simplicial side becoming a homotopy-coherent multiplication on the chain-complex side because the equivalence is only lax monoidal. Strict commutative ring spectra and $E_{\infty}$ -ring spectra coincide on the connective side via this Quillen equivalence. Rubric: full credit for the worked example plus the Schwede-Shipley statement; partial credit for either piece alone.

Lean formalization Intermediate+

lean_status: none — Mathlib has a partial implementation in Mathlib.CategoryTheory.Idempotents.DoldKan covering the Karoubi-enriched version. The classical statement for ordinary simplicial abelian groups is not yet packaged. The intended formalisation reads schematically:

import Mathlib.AlgebraicTopology.SimplicialObject
import Mathlib.Algebra.Homology.HomologicalComplex
import Mathlib.CategoryTheory.Idempotents.DoldKan

namespace Codex.Foundations.Algebra.DoldKan

variable {A : Type*} [Category A] [Abelian A]

/-- The normalised chain functor N : sA → Ch_{≥0}(A). -/
noncomputable def normalisedChain :
    SimplicialObject A ⥤ ChainComplex A ℕ :=
  sorry  -- N_n A = ⋂_{i < n} ker d_i with differential d_n

/-- The Kan-extension inverse K : Ch_{≥0}(A) → sA. -/
noncomputable def kanExtension :
    ChainComplex A ℕ ⥤ SimplicialObject A :=
  sorry  -- K(C)_n = ⊕_{σ : [n] ↠ [k]} C_k with Kan-extension structure maps

/-- The Dold-Kan correspondence: N and K are mutually inverse
    equivalences of categories. -/
noncomputable def doldKanEquivalence :
    SimplicialObject A ≌ ChainComplex A ℕ where
  functor := normalisedChain
  inverse := kanExtension
  unitIso := sorry  -- A ≅ K(N A) via Eilenberg-MacLane splitting
  counitIso := sorry  -- N(K C) ≅ C via the identity-surjection summand

end Codex.Foundations.Algebra.DoldKan

The proof gap is substantive but well-scoped. The Karoubi-Dold-Kan API in Mathlib.CategoryTheory.Idempotents.DoldKan provides the abstract scaffolding: it gives the equivalence between Karoubi envelopes $Karoubi (SimplicialObject A) ≃ Karoubi (ChainComplex A)$ for any additive $A$ . When $A$ is already idempotent-complete — and any abelian category is — the Karoubi envelope is equivalent to the original category, so the abstract result specialises to the ordinary equivalence. What is missing is the named specialisation theorem on ordinary $s Ab$ , plus the Eilenberg-MacLane operator $E_{n}$ as an explicit idempotent, the explicit normalised-chain formula $N_{n} A = ⋂_{i < n} ker d_{i}$ , the Kan-extension formula for $K$ on a chain complex, and the Eilenberg-Zilber-Cartier corollary $N (A \otimes B) ≃ N A \otimes N B$ with the shuffle map. Each piece is formalisable from existing Mathlib infrastructure; the consolidation into a CategoryTheory.Equivalence instance with the explicit shuffle-Alexander-Whitney monoidal structure is the upstream contribution target.

Advanced results Master

Theorem (Eilenberg-Zilber-Cartier). For simplicial abelian groups $A, B \in s Ab$ , the shuffle map $$ \nabla : NA \otimes NB \to N(A \otimes B) $$ and the Alexander-Whitney map $$ \mathrm{AW} : N(A \otimes B) \to NA \otimes NB $$ are mutually-inverse-up-to-chain-homotopy quasi-isomorphisms. The composition $AW \circ \nabla$ equals $id_{N A \otimes N B}$ on the nose; the composition $\nabla \circ AW$ is chain-homotopic to $id_{N (A \otimes B)}$ via an explicit homotopy constructed by Eilenberg-MacLane.

The shuffle map originates in Eilenberg-MacLane's 1953-1954 papers on $H (π, n)$ . The Alexander-Whitney map originates independently in Alexander's combinatorial work on the cup product and Whitney's chain-level cross product. Their identification as mutually-inverse-up-to-homotopy is the content of the Eilenberg-Zilber theorem in its modern form, proved at the chain-complex level by Cartier and refined in subsequent expositions in Goerss-Jardine Ch. III §2 and May 1967 §22.

The theorem upgrades the Dold-Kan correspondence to a lax-monoidal equivalence: the equivalence $N ⊣ K$ is compatible with tensor products up to coherent homotopy, but not strictly. The unit of the equivalence is the standard tensor unit $Z$ in degree $0$ , fixed by both $N$ and $K$ ; the product structure transports via the shuffle, which is associative and commutative up to homotopy but not on the nose. The strict-monoidal failure is exactly what makes simplicial commutative rings (strict commutativity) and connective $E_{\infty}$ -algebras (homotopy-coherent commutativity) into distinct presentations of the same homotopy theory, distinguished only by the choice of strictification.

Theorem (Moore complex agrees with normalised complex). For a simplicial abelian group $A$ , the Moore complex $$ M_n A = \bigcap_{i = 1}^{n} \ker(d_i : A_n \to A_{n-1}) $$ with differential $d_{0}$ is canonically isomorphic to the normalised chain complex $N_{n} A = ⋂_{i = 0}^{n - 1} ker d_{i}$ with differential $\pm d_{n}$ , via the reflection $[n] \to [n]$ , $i \mapsto n - i$ in the simplex category.

The Moore complex and the normalised complex are two ways of choosing which $n$ face maps to use as kernel-defining constraints. The reflection automorphism of $[n]$ exchanges them and gives the canonical isomorphism. Some references (May, Goerss-Jardine) prefer the Moore convention; others (Weibel, Gelfand-Manin) prefer the normalised convention. The Dold-Kan correspondence holds with either convention.

Theorem (Schwede-Shipley monoidal Dold-Kan; Schwede-Shipley 2003, Topology 42). Let $R$ be a commutative ring. The Dold-Kan correspondence extends to a Quillen equivalence $$ N : s(R\text{-}\mathrm{Mod}) \rightleftarrows \mathrm{Ch}_{\ge 0}(R\text{-}\mathrm{Mod}) : K $$ of monoidal model categories, where both sides carry the projective model structure (cofibrations are degree-wise mono, weak equivalences detected by homotopy groups / homology). The Quillen equivalence is lax monoidal via the shuffle / Alexander-Whitney structure, and the homotopy categories are equivalent as symmetric monoidal categories.

The consequence: the category of simplicial commutative $R$ -algebras and the category of connective $E_{\infty}$ -algebras over $R$ (in chain complexes) have equivalent homotopy categories. This bridges the strictly-commutative-multiplication world of simplicial rings with the up-to-homotopy-commutative world of $E_{\infty}$ -rings, on the connective side. The non-connective side requires stabilisation and is the subject of Robalo's spectral-Dold-Kan and the Lurie / Higher Algebra framework.

Theorem (Dold-Kan in a general additive category). For any idempotent-complete additive category $A$ , the normalised-chain functor $N$ and the Kan extension $K$ are mutually inverse equivalences $$ N : s\mathcal{A} \xrightarrow{\sim} \mathrm{Ch}{\ge 0}(\mathcal{A}), \qquad K : \mathrm{Ch}{\ge 0}(\mathcal{A}) \xrightarrow{\sim} s\mathcal{A}. $$ Idempotent-completeness is required precisely so that the Eilenberg-MacLane operator $E_{n}$ splits as a direct-sum decomposition; in a general additive category without this hypothesis, only the Karoubi-envelope version holds.

In particular the correspondence applies to $A = R - Mod$ for any ring $R$ , $A = Coh (X)$ for a Noetherian scheme $X$ , or $A = Sh (X, Ab)$ for sheaves of abelian groups on a topological space. The Karoubi-envelope reformulation (developed in Mathlib's CategoryTheory.Idempotents.DoldKan namespace) extends the equivalence to all additive categories at the cost of replacing both sides with their Karoubi envelopes.

Theorem (Hochschild-style chain complexes). For a commutative ring $R$ and a simplicial $R$ -module $A$ , the normalised chain complex $N A$ computes the homotopy groups $\pi_ A $e x a c tl y :$ \pi_n A = H_n(NA) $f or e v er y$ n \ge 0$.*

This identification is the simplicial-side computation that makes Dold-Kan a workhorse of homological algebra. Tor and Ext groups, which are defined via projective or injective resolutions, can equivalently be computed via simplicial resolutions and the normalised chain complex; the two computations agree by Dold-Kan. Hochschild homology of a commutative ring is the canonical example: $HH_{*} (R)$ is the homotopy of the simplicial commutative ring $R \otimes_{R \otimes R} R$ , equivalently the homology of its normalised chain complex.

Synthesis. The Dold-Kan correspondence is the foundational reason simplicial homotopy theory and homological algebra share a common algebraic core. The central insight is that the normalised-chain functor $N$ and the Kan extension $K$ are mutually inverse equivalences $s A ≃ Ch_{\geq 0} (A)$ for every idempotent-complete additive category $A$ , with the Eilenberg-MacLane splitting $C_{*} A = N_{*} A \oplus D_{*} A$ as the explicit witness. Putting these together, the apparently distinct languages of face-degeneracy operators on the simplicial side and single-differential complexes on the chain-complex side are two presentations of the same data. The bridge appears again in 03.12.25 (simplicial sets), where the free simplicial-abelian-group functor $Z [-]$ applied to the singular simplicial set $Sing (X)$ produces a simplicial abelian group whose normalised chain complex is the singular chain complex of $X$ ; Dold-Kan identifies the singular-chain-complex calculation with the simplicial-homotopy-group calculation in every degree, so $H_{n} (X; Z) = π_{n} Z [Sing (X)]$ on the nose.

The correspondence also identifies the Eilenberg-MacLane simplicial abelian groups $K (π, n)$ with the chain complexes $π [n]$ concentrated in a single degree: $N K (π, n) = π [n]$ and $K π [n] = K (π, n)$ . This pattern recurs in 03.12.05 (Eilenberg-MacLane spaces) where the same identification on the topological side underlies the representability of cohomology by $K (π, n)$ . The lax-monoidal refinement via the shuffle / Alexander-Whitney machinery (Eilenberg-Zilber-Cartier) is exactly what makes Tor and Ext computable via simplicial resolutions and what makes Hochschild homology a derived tensor product. The Schwede-Shipley monoidal refinement (2003) builds toward the modern theory of connective $E_{\infty}$ -rings, where simplicial commutative rings and connective $E_{\infty}$ -algebras in chain complexes are identified as the same homotopy theory. The foundational reason all of these constructions agree is precisely the equivalence $N ⊣ K$ : every functorial construction on chain complexes lifts uniquely to a functorial construction on simplicial abelian groups, and the simplicial-side and chain-complex-side calculations match degree-by-degree.

Full proof set Master

Theorem (Dold-Kan), proof. Given in the Intermediate-tier section: the Eilenberg-MacLane operator $E_{n} = \prod_{k = 0}^{n - 1} (id - s_{k} d_{k})$ is an idempotent on $C_{n} A$ whose image is $N_{n} A$ and whose kernel is $D_{n} A$ , giving the splitting $C_{n} A = N_{n} A \oplus D_{n} A$ . The Kan-extension functor $K$ on a chain complex rebuilds a simplicial abelian group whose normalised complex is the input via the formula $K (C_{*})_{n} = ⨁_{σ : [n] ↠ [k]} C_{k}$ . The unit $η : id_{s A} \to K \circ N$ and counit $ε : N \circ K \to id_{Ch_{\geq 0} (A)}$ are natural isomorphisms by construction. $□$

Proposition (the Eilenberg-MacLane operator is idempotent). For each $n$ , the operator $E_{n} = (id - s_{n - 1} d_{n - 1}) \dots (id - s_{0} d_{0})$ satisfies $E_{n}^{2} = E_{n}$ .

Proof. Compute $E_{n}^{2}$ by expanding both copies. Each factor $(id - s_{k} d_{k})$ is an idempotent in its own right because $s_{k} d_{k}$ is itself idempotent: $(s_{k} d_{k}) (s_{k} d_{k}) = s_{k} (d_{k} s_{k}) d_{k} = s_{k} \cdot id \cdot d_{k} = s_{k} d_{k}$ , using the simplicial identity $d_{k} s_{k} = id$ . The product of idempotents is idempotent provided the factors commute, and the factors $(id - s_{k} d_{k})$ for varying $k$ commute up to terms whose product vanishes by the simplicial identities $d_{i} s_{j} = s_{j - 1} d_{i}$ for $i < j$ , which kill cross-terms involving $s_{k} d_{k}$ followed by $s_{j} d_{j}$ with $j < k$ . A careful expansion of the product, with the simplicial identities applied to commute degeneracies past face operators, gives $E_{n} \cdot E_{n} = E_{n}$ . Image: every element of $im E_{n}$ has the form $E_{n} (x)$ , and applying $d_{i}$ for $i < n$ to $E_{n} (x)$ kills the result by the simplicial identity $d_{i} (id - s_{i} d_{i}) = d_{i} - d_{i} s_{i} d_{i} = d_{i} - d_{i} = 0$ , so $im E_{n} \subseteq ⋂_{i < n} ker d_{i} = N_{n} A$ . The reverse inclusion follows because $E_{n}$ acts as the identity on $N_{n} A$ : each factor $(id - s_{k} d_{k})$ applied to an element $x \in ⋂_{i < n} ker d_{i}$ kills the second term $s_{k} d_{k} (x) = s_{k} \cdot 0 = 0$ for every $k < n$ . Kernel: by direct computation, $ker E_{n}$ is the image of the degeneracies $\sum_{i} im s_{i} = D_{n} A$ . $□$

Proposition (every degenerate simplex is uniquely a degeneracy of a non-degenerate one — Eilenberg-Zilber lemma). For every $x \in A_{n}$ in a simplicial abelian group $A$ , there exist unique data $(k, σ, y)$ where $0 \leq k \leq n$ , $σ : [n] ↠ [k]$ is an order-preserving surjection, and $y \in N_{k} A$ is a non-degenerate simplex, such that $x = \sigma^ y $w h er e$ \sigma^ $i s t h e i t er a t e dd e g e n er a cy o p er a t or a ssoc ia t e d t o$ \sigma$.

Proof. Existence follows from the Eilenberg-MacLane splitting in a slightly more refined form. Decompose $A_{n}$ as the direct sum $⨁_{σ} σ^{*} (N_{k} A)$ where $σ$ ranges over order-preserving surjections $[n] ↠ [k]$ with $k \leq n$ . Each summand $σ^{*} (N_{k} A)$ is the image of $N_{k} A$ under the iterated degeneracy along the inverse of $σ$ in $Δ$ ; these images are linearly independent because the simplicial identities forbid relations between distinct $σ$ . So every $x \in A_{n}$ has a unique decomposition $x = \sum_{σ} σ^{*} (y_{σ})$ with $y_{σ} \in N_{k} A$ . Uniqueness of the dominant non-degenerate part follows from the dimension count: at most one $y_{σ}$ can be non-zero in the formula $x = σ^{*} y$ if we require $y \in N_{k} A$ and $x$ to be a single non-degenerate-pulled-back element. Existence of a single dominant $σ$ for a given $x$ follows because the Eilenberg-MacLane idempotents project $x$ onto its component in the canonical summand. The uniqueness statement is the combinatorial content of the Eilenberg-Zilber lemma. $□$

Theorem (Eilenberg-Zilber-Cartier), proof outline. The shuffle map $\nabla : N A \otimes N B \to N (A \otimes B)$ and the Alexander-Whitney map $AW : N (A \otimes B) \to N A \otimes N B$ are defined by the explicit formulas in Exercise 7. The composition $AW \circ \nabla = id_{N A \otimes N B}$ is verified by a direct shuffle-counting argument: the Alexander-Whitney map evaluates $\nabla (a \otimes b)$ on the leading face-deletion factors, and the shuffle-counting telescope reduces to the identity. The composition $\nabla \circ AW$ is chain-homotopic to $id_{N (A \otimes B)}$ via the Eilenberg-MacLane chain homotopy, constructed inductively: at each simplicial level, the homotopy interpolates between a $(p + q)$ -simplex and its shuffle-of-Alexander-Whitney decomposition by inserting partial degeneracies. The full construction appears in Eilenberg-MacLane 1953 ^{[Eilenberg-MacLane 1953]} and Goerss-Jardine Ch. III §2 ^{[Goerss-Jardine 1999]}. $□$

Theorem (Moore complex agrees with normalised complex), proof. The reflection $ρ : [n] \to [n]$ , $i \mapsto n - i$ , is an automorphism of $[n]$ in the simplex category $Δ$ (it is order-reversing, but the reflection-induced functor on simplicial objects swaps $d_{i}$ with $d_{n - i}$ and $s_{i}$ with $s_{n - i - 1}$ ). Pulling back along $ρ$ exchanges the kernel intersection $⋂_{i = 0}^{n - 1} ker d_{i}$ (normalised) with $⋂_{i = 1}^{n} ker d_{i}$ (Moore), and similarly exchanges the differential $\pm d_{n}$ with $d_{0}$ . The resulting isomorphism between normalised and Moore complexes is canonical because the reflection is unique up to a sign choice. $□$

Theorem (Dold-Kan in a general additive category), proof. The construction of $N$ and $K$ uses only the additive structure plus the splitting of idempotents. In an idempotent-complete additive category, the Eilenberg-MacLane operator $E_{n}$ splits as a direct-sum decomposition $C_{n} A = N_{n} A \oplus D_{n} A$ , and the rest of the argument proceeds identically. Idempotent-completeness is required precisely at the splitting step: without it, $E_{n}$ is an idempotent endomorphism but its image and kernel may not be retracts of $A_{n}$ , and the equivalence holds only after passing to Karoubi envelopes. The Karoubi-envelope version is proved in Mathlib's CategoryTheory.Idempotents.DoldKan and gives the abstract scaffold for the idempotent-complete case. $□$

Theorem (Schwede-Shipley monoidal refinement), stated without proof — see Schwede-Shipley 2003 Topology 42 ^{[Schwede-Shipley 2003]}. The full monoidal Quillen equivalence requires verification that the shuffle / Alexander-Whitney structure satisfies the lax-monoidal axioms (associativity and unit constraints) up to coherent homotopy, plus a Quillen-pair compatibility check at the model-category level. Both verifications are technical but routine given the additive Dold-Kan equivalence; the consequences for connective $E_{\infty}$ -rings are substantial and use the full Hovey monoidal-model-category framework.

Connections Master

Chain complex in an abelian category 01.02.30. The Dold-Kan correspondence identifies $s A$ with $Ch_{\geq 0} (A)$ for every abelian category $A$ . Without the chain-complex side, the correspondence has no target; with it, the equivalence transports every theorem about $Ch_{\geq 0} (A)$ — long exact sequences, the snake lemma, derived functors — into a theorem about simplicial objects, and vice versa. The bounded-below grading on the chain-complex side is forced by the non-negative grading on the simplicial side.
Abelian category and Grothendieck axioms AB1-AB5 01.02.33. The correspondence requires only idempotent-completeness of the additive ground category, which is automatic in any abelian category. The Grothendieck AB5 axiom — exactness of filtered colimits — passes through the correspondence: a simplicial abelian group has filtered colimits computed level-wise, and the normalised chain functor preserves filtered colimits because it is built from kernels which commute with filtered colimits in an AB5 category.
Simplicial sets and geometric realisation 03.12.25. The free-abelian-group functor $Z [-] : sSet \to s Ab$ converts a simplicial set $X$ into a simplicial abelian group level-wise. Composing with the normalised-chain functor produces the singular chain complex $C_{*}^{sing} (X) = N Z [X]$ , identifying the simplicial-homotopy groups $π_{n} Z [X]$ with the homology groups $H_{n} (X; Z)$ on the nose. The Dold-Kan correspondence is what allows the singular-chain computation and the simplicial-homotopy computation to be the same calculation, not merely equivalent up to quasi-isomorphism.
Eilenberg-MacLane spaces 03.12.05. The Eilenberg-MacLane simplicial abelian group $K (π, n)$ corresponds under Dold-Kan to the chain complex $π [n]$ with $π$ concentrated in degree $n$ . The topological Eilenberg-MacLane space is the geometric realisation $∣ K (π, n) ∣$ , and its cohomology represents singular cohomology with coefficients in $π$ . The Dold-Kan identification is exactly what makes the algebraic chain complex $π [n]$ and the topological space $K (π, n)$ two presentations of the same homotopy type at the level of simplicial abelian groups.
Tensor product of modules 01.02.10. The Eilenberg-Zilber-Cartier theorem $N (A \otimes B) ≃ N A \otimes N B$ identifies the chain-complex tensor product with the simplicial-abelian-group diagonal tensor product, up to the shuffle / Alexander-Whitney chain homotopy. This is what makes Tor groups computable via simplicial resolutions: a flat simplicial resolution of an $R$ -module $M$ corresponds under Dold-Kan to a flat chain-complex resolution, and the tensor products agree up to the Eilenberg-Zilber homotopy, so $Tor_{*}^{R} (M, N)$ has matching simplicial and chain-complex calculations.

Historical & philosophical context Master

The Dold-Kan correspondence emerged simultaneously and independently from two directions in 1958. Albrecht Dold, in his 1958 paper Homology of symmetric products and other functors of complexes (Ann. Math. 68, 54-80) ^{[Dold 1958]}, constructed the normalised-chain functor $N$ from simplicial abelian groups to non-negatively graded chain complexes and established that it is an equivalence of categories. Dold's motivation was the calculation of homology of symmetric products $SP^{n} (X)$ , where the structural simplicial-abelian-group decomposition of the singular chain complex of $SP^{n} (X)$ required isolating the non-degenerate part of a simplicial abelian group as a chain complex. The explicit splitting via the Eilenberg-MacLane operator was the key tool.

Daniel Kan, in his 1958 paper Functors involving c.s.s. complexes (Proc. NAS 41, 1092-1096) ^{[Kan 1958]}, constructed the inverse functor $K$ via Kan extension and proved the equivalence from the opposite direction. Kan's motivation came from his earlier work on the combinatorial homotopy theory of simplicial sets and the Kan extension condition; the functor $K$ realises an arbitrary chain complex as a simplicial abelian group via the Kan-extension formula, giving the inverse half of the equivalence. The simultaneous publication of Dold's and Kan's papers in 1958, with complementary perspectives, established the equivalence with both directions explicit and gave the correspondence its hyphenated name.

The shuffle / Alexander-Whitney machinery that upgrades the additive equivalence to a lax-monoidal one originates in Eilenberg-MacLane's 1953-1954 papers On the groups $H (π, n)$ I-II (Ann. Math. 58, 55-106 and 60, 49-139) ^{[Eilenberg-MacLane 1953]}. The shuffle map was used by Eilenberg-MacLane to compute the homology of the spaces $K (π, n)$ , and the Alexander-Whitney map appeared in Whitney's earlier chain-level cross-product constructions. Cartier in the 1950s assembled these into the modern Eilenberg-Zilber-Cartier theorem, and Goerss-Jardine's Simplicial Homotopy Theory (Birkhäuser 1999) ^{[Goerss-Jardine 1999]} gives the definitive modern exposition with full proofs of the chain-homotopy inversions.

The monoidal refinement was completed in stages by Schwede and Shipley. Their 2003 paper Equivalences of monoidal model categories (Topology 42, 103-153) ^{[Schwede-Shipley 2003]} established the Quillen equivalence of monoidal model categories between simplicial $R$ -modules and non-negatively graded chain complexes of $R$ -modules, identifying the homotopy categories of simplicial commutative $R$ -algebras and connective $E_{\infty}$ -algebras over $R$ . This identification underlies modern derived algebraic geometry, where the equivalence between the simplicial-commutative-ring perspective (Toën-Vezzosi-Lurie) and the chain-complex / dg-algebra perspective (Kontsevich-Soibelman) on the connective side is exactly the Schwede-Shipley refinement of Dold-Kan. The Karoubi-envelope reformulation, developed in the categorical-logic literature (Bourn, Beauville, and others in the 1980s-1990s), extended the equivalence to arbitrary additive categories, and is the form formalised in Mathlib's CategoryTheory.Idempotents.DoldKan namespace.

Bibliography Master

@article{Dold1958,
  author  = {Dold, Albrecht},
  title   = {Homology of symmetric products and other functors of complexes},
  journal = {Ann. of Math. (2)},
  volume  = {68},
  year    = {1958},
  pages   = {54--80}
}

@article{Kan1958FunctorsCSS,
  author  = {Kan, Daniel M.},
  title   = {Functors involving c.s.s. complexes},
  journal = {Proc. Nat. Acad. Sci. USA},
  volume  = {41},
  year    = {1958},
  pages   = {1092--1096}
}

@article{EilenbergMacLane1953,
  author  = {Eilenberg, Samuel and MacLane, Saunders},
  title   = {On the groups $H(\pi, n)$, I},
  journal = {Ann. of Math. (2)},
  volume  = {58},
  year    = {1953},
  pages   = {55--106}
}

@article{EilenbergMacLane1954,
  author  = {Eilenberg, Samuel and MacLane, Saunders},
  title   = {On the groups $H(\pi, n)$, II: Methods of computation},
  journal = {Ann. of Math. (2)},
  volume  = {60},
  year    = {1954},
  pages   = {49--139}
}

@book{GoerssJardine1999,
  author    = {Goerss, Paul G. and Jardine, John F.},
  title     = {Simplicial Homotopy Theory},
  publisher = {Birkh{\"a}user},
  series    = {Progress in Mathematics},
  volume    = {174},
  year      = {1999}
}

@book{Weibel1994,
  author    = {Weibel, Charles A.},
  title     = {An Introduction to Homological Algebra},
  publisher = {Cambridge University Press},
  series    = {Cambridge Studies in Advanced Mathematics},
  volume    = {38},
  year      = {1994}
}

@book{GelfandManin2003,
  author    = {Gelfand, Sergei I. and Manin, Yuri I.},
  title     = {Methods of Homological Algebra},
  publisher = {Springer-Verlag},
  edition   = {2},
  year      = {2003}
}

@book{May1967Simplicial,
  author    = {May, J. Peter},
  title     = {Simplicial Objects in Algebraic Topology},
  publisher = {University of Chicago Press},
  year      = {1967}
}

@article{SchwedeShipley2003,
  author  = {Schwede, Stefan and Shipley, Brooke},
  title   = {Equivalences of monoidal model categories},
  journal = {Topology},
  volume  = {42},
  year    = {2003},
  pages   = {103--153}
}

@article{EilenbergZilber1953,
  author  = {Eilenberg, Samuel and Zilber, Joseph A.},
  title   = {On products of complexes},
  journal = {Amer. J. Math.},
  volume  = {75},
  year    = {1953},
  pages   = {200--204}
}

Prerequisites

01.02.30
01.02.33
03.12.25

Tier anchors

beginner: Goerss-Jardine *Simplicial Homotopy Theory* Ch. III §2 informal opening on simplicial abelian groups as chain complexes
intermediate: Weibel *An Introduction to Homological Algebra* Ch. 8 §4; Goerss-Jardine *Simplicial Homotopy Theory* Ch. III §2; Gelfand-Manin *Methods of Homological Algebra* Ch. I §§6-7
master: Dold 1958 *Ann. Math.* 68 (54-80); Kan 1958 *Proc. NAS* 41 (1092-1096); Goerss-Jardine Ch. III §2 with the full Eilenberg-MacLane splitting; May 1967 *Simplicial Objects in Algebraic Topology* §22; Schwede-Shipley 2003 *Topology* 42 on the monoidal refinement

References

Dold — Homology of symmetric products and other functors of complexes · Ann. of Math. (2) 68 (1958) 54-80; originator paper for the equivalence N : sAb → Ch_{≥0}(Ab) with the splitting via the Eilenberg-MacLane idempotent
Kan — Functors involving c.s.s. complexes · Proc. Nat. Acad. Sci. USA 41 (1958) 1092-1096; constructs the inverse functor K as a Kan extension, parallel to Dold's normalised-chain functor
Goerss-Jardine — Simplicial Homotopy Theory · Birkhäuser Progress in Mathematics 174 (1999); Ch. III §2 full proof of the Dold-Kan correspondence, the Eilenberg-MacLane splitting, and the Eilenberg-Zilber corollary
Weibel — An Introduction to Homological Algebra · Cambridge Studies in Advanced Mathematics 38 (1994); Ch. 8 §4 Dold-Kan correspondence with the Moore complex / normalised-chain comparison, Eilenberg-Zilber shuffle map, and the Alexander-Whitney map
Gelfand-Manin — Methods of Homological Algebra · Springer 1996 / 2003; Ch. I §§6-7 simplicial objects, normalised chain complex, equivalence with non-negatively graded chain complexes, role in derived-category-style organisation of homological algebra
May — Simplicial Objects in Algebraic Topology · University of Chicago Press 1967; §22 normalised chain complex, the homotopy-equivalence between $C_* A$ and $N_* A$, and the Eilenberg-Zilber theorem on the chain level
Schwede-Shipley — Equivalences of monoidal model categories · Topology 42 (2003) 103-153; monoidal refinement of Dold-Kan establishing the Quillen equivalence between simplicial $R$-modules and non-negatively graded chain complexes of $R$-modules for a commutative ring $R$
Eilenberg-MacLane — On the groups H(π,n) I-II · Ann. of Math. (2) 58 (1953) 55-106 and 60 (1954) 49-139; sets up the normalised chain complex and the shuffle-product Alexander-Whitney machinery that feed into the Eilenberg-Zilber theorem

Estimated time

beginner: 18m
intermediate: 45m
master: 80m