03.08.02 · modern-geometry / k-theory

Adams operations $ψ^{k}$

shipped3 tiersLean: none

Anchor (Master): Atiyah *K-Theory* Ch. III; Karoubi *K-Theory: An Introduction* Ch. IV; Adams *Vector fields on spheres* (Annals of Math. 75, 1962); Atiyah *Power operations in K-theory* (Quart. J. Math. 17, 1966)

Intuition [Beginner]

Adams operations are a way to raise a vector bundle to a power. For an ordinary number, raising to the $k$ -th power means multiplying it by itself $k$ times. For a vector bundle, the analogue is more subtle, because bundles do not have a single product operation; they have direct sum and tensor product, and various exterior-power constructions. Adams operations pick out the right combination of these to make a power-like operation behave the way arithmetic powers behave.

The point of having such operations is to detect finer information about a space than the K-theory group $K (X)$ records on its own. Two spaces can have the same K-group as abstract abelian groups, but different Adams operations acting on them. Comparing how $ψ^{2}$ , $ψ^{3}$ , and so on shuffle classes inside $K (X)$ separates spaces that the bare group cannot.

The most famous payoff is a theorem of Adams from 1960: only four spheres carry a multiplication making them into a normed division ring — $R$ , $C$ , the quaternions $H$ , and the octonions $O$ . The proof uses Adams operations on $K$ -theory of a low-dimensional sphere; nothing else available at the time was sharp enough to settle the question.

Visual [Beginner]

A schematic of a line bundle $L$ being mapped to its $k$ -th tensor power $L^{k}$ — the same fibers raised to the $k$ -th power. A second arrow shows a more general bundle $E$ landing on a class $ψ^{k} (E)$ inside $K (X)$ , with the recipe involving exterior powers of $E$ combined by the Newton formula.

The picture records the two halves of the definition: on the simple case of line bundles, the operation is just the $k$ -fold tensor power; on general bundles, it is built from exterior powers using a polynomial formula.

Worked example [Beginner]

Consider the tautological line bundle $L$ over the projective line $CP^{1}$ , the simplest projective space carrying a non-product bundle. The K-group $K^{0} (CP^{1})$ is generated by the class of $L$ and the rank, with one relation: $(L - 1)^{2} = 0$ .

Compute $ψ^{2} (L)$ . Because $L$ is a line bundle, the definition says $ψ^{2} (L) = L^{2}$ — the tensor square of $L$ . Write $t = L - 1$ so that $L = 1 + t$ and the relation reads $t^{2} = 0$ . Then $$ L^2 = (1 + t)^2 = 1 + 2t + t^2 = 1 + 2t. $$ So $ψ^{2} (L) = 1 + 2 t = 2 L - 1$ inside $K^{0} (CP^{1})$ .

Compute $ψ^{3} (L)$ in the same way. $L^{3} = (1 + t)^{3} = 1 + 3 t + 3 t^{2} + t^{3} = 1 + 3 t$ , so $ψ^{3} (L) = 3 L - 2$ .

What this tells us: on the tautological bundle of $CP^{1}$ , the operation $ψ^{k}$ scales the generator $t = L - 1$ by $k$ — a discrete signature, one integer per operation, distinguishing classes that look identical as K-classes alone.

Check your understanding [Beginner]

Formal definition [Intermediate+]

Let $X$ be a compact Hausdorff space and $K (X) = K^{0} (X)$ the Grothendieck group of complex topological vector bundles over $X$ 03.08.01, a commutative ring under tensor product. For each $k \geq 0$ the exterior power $Λ^{k}$ defines a natural operation on isomorphism classes of complex vector bundles: $Λ^{k} (E)$ is the bundle whose fibre at $x \in X$ is the $k$ -th exterior power of the fibre $E_{x}$ . Direct sum interacts with exterior powers through the Cartan formula $$ \Lambda^k(E \oplus F) = \bigoplus_{i + j = k} \Lambda^i(E) \otimes \Lambda^j(F), $$ which descends to the Grothendieck ring once one defines $λ^{k} : K (X) \to K (X)$ by the generating-function formula. The generating function $$ \lambda_t(E) = \sum_{k \geq 0} \lambda^k(E) , t^k \in K(X)[[t]] $$ is required to satisfy $λ_{t} (E + F) = λ_{t} (E) \cdot λ_{t} (F)$ . The Cartan formula gives this on bundle classes; extension to formal differences uses the relation $λ_{t} (E - F) = λ_{t} (E) \cdot λ_{t} (F)^{- 1}$ in $K (X) [[t]]$ , the inverse computed by the geometric-series formula. The pair $(K (X), (λ^{k})_{k \geq 0})$ is a $λ$ -ring in the sense of Atiyah-Tall.

Definition (Adams operation $ψ^{k}$ ). Let $k \geq 1$ . The $k$ -th Adams operation $ψ^{k} : K (X) \to K (X)$ is the function defined by $$ \psi^k(x) = N_k(\lambda^1(x), \lambda^2(x), \ldots, \lambda^k(x)), $$ where $N_{k} (c_{1}, \dots, c_{k})$ is the $k$ -th Newton polynomial, the unique integer polynomial in elementary symmetric variables $c_{1}, \dots, c_{k}$ such that whenever $c_{i} = e_{i} (α_{1}, \dots, α_{n})$ for $n \geq k$ formal variables $α_{j}$ and elementary symmetric polynomials $e_{i}$ , $$ N_k(c_1, \ldots, c_k) = \alpha_1^k + \alpha_2^k + \cdots + \alpha_n^k = p_k(\alpha_1, \ldots, \alpha_n). $$ The first few Newton polynomials are $N_{1} = c_{1}$ , $N_{2} = c_{1}^{2} - 2 c_{2}$ , $N_{3} = c_{1}^{3} - 3 c_{1} c_{2} + 3 c_{3}$ . They are determined by the Newton-Girard recursion $$ N_k = c_1 N_{k-1} - c_2 N_{k-2} + \cdots + (-1)^{k-1} k , c_k. $$

Equivalently, on a sum of line bundles $E = L_{1} \oplus \dots \oplus L_{n}$ , the elementary symmetric polynomial $e_{i} (L_{1}, \dots, L_{n})$ equals $λ^{i} (E)$ in $K (X)$ , so the defining identity reads $$ \psi^k(L_1 \oplus \cdots \oplus L_n) = L_1^{\otimes k} + L_2^{\otimes k} + \cdots + L_n^{\otimes k}. $$ On a line bundle alone, $ψ^{k} (L) = L^{\otimes k}$ . The splitting principle (every complex vector bundle over a compact base pulls back to a sum of line bundles over a finite flag-bundle tower whose K-theory map is injective) extends formulas verified on sums of line bundles to general bundles, and the Adams operations are uniquely determined by their values on line bundles together with naturality and additivity.

Counterexamples to common slips

$ψ^{k}$ is not the operation $x \mapsto x^{k}$ in the ring $K (X)$ . The two agree on line bundles, but differ on classes of higher rank: for a rank-two bundle $E = L_{1} \oplus L_{2}$ , $E^{2} = L_{1}^{2} + 2 L_{1} L_{2} + L_{2}^{2}$ in $K (X)$ , while $ψ^{2} (E) = L_{1}^{2} + L_{2}^{2}$ ; the cross term is absent.
$ψ^{k}$ is not a multiplicative scaling: $ψ^{k} (L_{1} + L_{2}) \neq = k (L_{1} + L_{2})$ . The genuine scaling property is the Chern-character compatibility $ch_{n} (ψ^{k} (x)) = k^{n} ch_{n} (x)$ in degree $2 n$ , not in K-theory itself.
Composition order matters in the same way as exponents: $ψ^{k} \circ ψ^{l} = ψ^{k l}$ , not $ψ^{k + l}$ . The additive analogue is wrong; the multiplicative one is right.
The Adams operations are defined for $k \geq 1$ only. The case $k = 0$ in some references denotes the rank map $rk : K (X) \to Z$ , but this is a notational extension rather than part of the Newton-polynomial recipe (Newton's $p_{0} = n$ is the dimension, not a recursive value).

Key theorem with proof [Intermediate+]

Theorem (Adams; Atiyah 1962). Let $X$ be a compact Hausdorff space and $k \geq 1$ . The function $ψ^{k} : K (X) \to K (X)$ is a ring homomorphism. The collection ${ψ^{k}}_{k \geq 1}$ satisfies the composition law $ψ^{k} \circ ψ^{l} = ψ^{k l}$ for all $k, l \geq 1$ , is natural in $X$ (for every continuous $f : X \to Y$ the diagram $f^ \circ \psi^k = \psi^k \circ f^ $co mm u t es), an d i s u ni q u e l y c ha r a c t er i se d a s t h e f ami l y o f na t u r a l r in g h o m o m or p hi s m s$ \psi^k : K(-) \to K(-) $s a t i s f y in g$ \psi^k(L) = L^{\otimes k} $f or e v er y l in e b u n d l e$ L$.

Proof. The proof has four steps: (i) ring-homomorphism property, (ii) composition law, (iii) naturality, (iv) uniqueness via the splitting principle.

(i) Ring homomorphism. Additivity $ψ^{k} (x + y) = ψ^{k} (x) + ψ^{k} (y)$ follows from the additivity of power sums: $p_{k} (α_{1}, \dots, α_{m}, β_{1}, \dots, β_{n}) = p_{k} (α) + p_{k} (β)$ . By the splitting principle there is a base $f : Y \to X$ with $f^{*}$ injective on $K$ -theory such that $f^{*} x$ and $f^{*} y$ are both sums of line bundles. The identity $ψ^{k} (f^{*} x + f^{*} y) = ψ^{k} (f^{*} x) + ψ^{k} (f^{*} y)$ is then a polynomial identity in formal variables (the line-bundle classes $L_{i}$ ), and the Newton polynomial recipe makes it the elementary identity $p_{k} (α, β) = p_{k} (α) + p_{k} (β)$ . Injectivity of $f^{*}$ transports the identity back to $X$ .

Multiplicativity $ψ^{k} (x \cdot y) = ψ^{k} (x) \cdot ψ^{k} (y)$ uses the same splitting argument plus the multiplicative identity for power sums of products of line bundles. If $f^{*} x = \sum L_{i}$ and $f^{*} y = \sum M_{j}$ , then $f^{*} x \cdot f^{*} y = \sum_{i, j} L_{i} M_{j}$ , and $$ \psi^k\left(\sum_{i, j} L_i M_j\right) = \sum_{i,j} (L_i M_j)^{\otimes k} = \sum_{i,j} L_i^{\otimes k} M_j^{\otimes k} = \left(\sum_i L_i^{\otimes k}\right) \cdot \left(\sum_j M_j^{\otimes k}\right) = \psi^k(f^* x) \cdot \psi^k(f^* y). $$ Inject back to $X$ . The unit relation $ψ^{k} (1) = 1$ follows from $1$ being the rank-one product bundle: $ψ^{k}$ of the constant line bundle is the constant line bundle.

(ii) Composition law. On a line bundle, $ψ^{k} \circ ψ^{l} (L) = ψ^{k} (L^{\otimes l}) = (L^{\otimes l})^{\otimes k} = L^{\otimes k l} = ψ^{k l} (L)$ . Both sides are ring homomorphisms on a sum of line bundles by step (i), so they agree on any class $f^{*} x$ in a splitting cover; injectivity of $f^{*}$ on $K$ -theory transports the identity to $X$ .

(iii) Naturality. The exterior-power functors are natural transformations of bundle categories, so each $λ^{k}$ is natural under pullback. The Newton-polynomial formula in $λ^{j}$ inherits this naturality.

(iv) Uniqueness. Suppose $\tilde{ψ}^{k} : K (-) \to K (-)$ is any natural ring homomorphism with $\tilde{ψ}^{k} (L) = L^{\otimes k}$ for every line bundle $L$ . The splitting principle gives, for any class $x \in K (X)$ , a base $f : Y \to X$ with $f^{*}$ injective and $f^{*} x = \sum L_{i}$ . By the ring-homomorphism property, $\tilde{ψ}^{k} (f^{*} x) = \sum L_{i}^{\otimes k} = ψ^{k} (f^{*} x)$ . Naturality of both operations gives $f^{*} \tilde{ψ}^{k} (x) = \tilde{ψ}^{k} (f^{*} x) = ψ^{k} (f^{*} x) = f^{*} ψ^{k} (x)$ , and injectivity of $f^{*}$ yields $\tilde{ψ}^{k} (x) = ψ^{k} (x)$ . $□$

Bridge. Adams operations build toward 03.09.10 (Atiyah-Singer index theorem) by way of the Chern character: the foundational reason $ψ^{k}$ acts diagonally on rational cohomology — with $ψ^{k}$ multiplying the degree- $2 n$ component of $ch$ by $k^{n}$ — is exactly that $ψ^{k}$ is the unique natural ring homomorphism behaving on line bundles like raising to the $k$ -th power, and the Chern character $ch (L) = e^{c_{1} (L)}$ transports tensor powers to multiplication of exponents. This identifies the Adams operation with a Frobenius-like endomorphism of $K (X) \otimes Q$ , the central insight that lets Adams attack the Hopf-invariant-one problem. The same operations appear again in 03.09.10 inside the topological index, where the $K$ -theory of the cotangent bundle carries Adams-operation eigenspaces refining the integer-valued index, and the bridge is the recognition that $ψ^{k}$ generalises the classical Steenrod power operations from mod- $p$ cohomology to $K$ -theory. Putting these together — the splitting principle, the Newton polynomial, the Frobenius image on $ch$ — Adams operations identify $K (X) \otimes Q$ with a weighted polynomial ring on which $ψ^{k}$ acts by a graded scaling, and this is exactly the foundational reason a finite calculation in low-dimensional K-theory suffices to rule out non-existent Hopf invariants and division-algebra structures.

Exercises [Intermediate+]

Exercise 5 (medium, symbolic).

Use the Newton recursion $N_{3} = c_{1} N_{2} - c_{2} N_{1} + 3 c_{3}$ to derive $ψ^{3} = λ_{1}^{3} - 3 λ_{1} λ_{2} + 3 λ_{3}$ , then evaluate on a rank-three bundle $E = L_{1} \oplus L_{2} \oplus L_{3}$ to confirm $ψ^{3} (E) = L_{1}^{3} + L_{2}^{3} + L_{3}^{3}$ .

Hint

Substitute $λ^{i} (E) = e_{i} (L_{1}, L_{2}, L_{3})$ (elementary symmetric polynomials in the $L_{j}$ ) and expand.

Answer

$N_{1} (c_{1}) = c_{1}$ , $N_{2} (c_{1}, c_{2}) = c_{1}^{2} - 2 c_{2}$ . The recursion gives $N_{3} = c_{1} (c_{1}^{2} - 2 c_{2}) - c_{2} c_{1} + 3 c_{3} = c_{1}^{3} - 3 c_{1} c_{2} + 3 c_{3}$ . Substitute $c_{i} = e_{i} (L_{1}, L_{2}, L_{3})$ : $e_{1} = L_{1} + L_{2} + L_{3}$ , $e_{2} = L_{1} L_{2} + L_{1} L_{3} + L_{2} L_{3}$ , $e_{3} = L_{1} L_{2} L_{3}$ . Then $$ e_1^3 - 3 e_1 e_2 + 3 e_3 = (L_1+L_2+L_3)^3 - 3(L_1+L_2+L_3)(L_1L_2 + L_1L_3 + L_2L_3) + 3 L_1 L_2 L_3 = L_1^3 + L_2^3 + L_3^3, $$ by the elementary Newton identity $p_{3} = e_{1} p_{2} - e_{2} p_{1} + 3 e_{3}$ unwound. Rubric: full credit for both the Newton derivation and the explicit substitution.

Exercise 6 (medium, short-answer).

Show that on $K (X) \otimes Q$ the Adams operations are simultaneously diagonalisable, with $ψ^{k}$ acting on the $ch^{- 1} (H^{2 n} (X; Q))$ eigenspace by the scalar $k^{n}$ .

Hint

Use the Chern character identity $ch (ψ^{k} (x)) = \sum_{n} k^{n} ch_{n} (x)$ and the fact that the Chern character is a rational isomorphism between $K (X) \otimes Q$ and $H^{even} (X; Q)$ .

Answer

By the Chern-character compatibility theorem, $ch_{n} (ψ^{k} (x)) = k^{n} ch_{n} (x)$ for every $n$ and every $x \in K (X)$ . The Chern character $ch : K (X) \otimes Q \to H^{even} (X; Q)$ is a ring isomorphism (Atiyah-Hirzebruch). Pull the cohomological-grading decomposition $H^{even} (X; Q) = ⨁_{n} H^{2 n} (X; Q)$ back to $K (X) \otimes Q = ⨁_{n} V_{n}$ where $V_{n} = ch^{- 1} (H^{2 n} (X; Q))$ . The action of $ψ^{k}$ on $V_{n}$ is scaling by $k^{n}$ . Rubric: full credit for citing the Chern-character isomorphism and applying the per-degree scaling identity.

Exercise 7 (hard, short-answer).

Use Adams operations on $K (CP^{2})$ to give a quick proof that $CP^{2}$ is not the suspension of any space, by exhibiting a non-zero product in $K (CP^{2})$ that vanishes on the suspension of any space.

Hint

Reduced K-theory of a suspension is annihilated by all products: $K (Σ Y) \cdot K (Σ Y) = 0$ . Find a non-zero product in $K (CP^{2})$ using $t = L - 1$ with $t^{2} \neq = 0$ .

Answer

$K (CP^{2}) = Z [t] / (t^{3})$ with $t = L - 1$ and $K (CP^{2}) = (t) / (t^{3})$ generated additively by $t$ and $t^{2}$ . The product $t \cdot t = t^{2} \neq = 0$ in this ring is non-vanishing. For any suspension $Σ Y$ , the cup product on $K (Σ Y)$ vanishes identically (suspensions induce zero products in reduced K-theory because the reduced suspension factors through the smash product $Σ Y \land Σ Y$ whose K-theory has bidegree $(1, 1)$ which Bott periodicity sends back to degree $2$ — and the multiplication on $K (Σ Y) \otimes K (Σ Y)$ lands in the zero degree-zero part). If $CP^{2} ≃ Σ Y$ , then $t^{2} = 0$ , contradicting the calculation. The Adams-operation refinement is that $ψ^{k}$ acts on $t$ by $k t + (2 k) t^{2}$ , and the coefficient $(2 k)$ on $t^{2}$ would have to be detectable mod the suspension nullity — another route to the same conclusion. Rubric: full credit for either the bare product argument or the $ψ^{k}$ -eigenvalue argument.

Exercise 8 (hard, short-answer).

Sketch the use of Adams operations in Adams' 1960 proof that the Hopf invariant takes the value $\pm 1$ only when $n \in {1, 2, 4, 8}$ .

Hint

Adams' argument applies $ψ^{2}$ and $ψ^{3}$ to a generator of $K (C_{f})$ for $C_{f}$ the mapping cone of a Hopf-invariant-one map $f : S^{2 n - 1} \to S^{n}$ . Reading off the eigenvalues forces a divisibility relation on $2$ and $3$ that holds only for the four classical dimensions.

Answer

Let $f : S^{4 m - 1} \to S^{2 m}$ be a map with Hopf invariant $\pm 1$ and let $C_{f} = S^{2 m} \cup_{f} e^{4 m}$ be its mapping cone. Reduced K-theory is $K (C_{f}) = Z α \oplus Z β$ with $α$ the pullback of the generator of $K (S^{2 m})$ and $β$ corresponding to the top cell; the Hopf-invariant-one hypothesis is the cup-product relation $α^{2} = \pm β$ in $K (C_{f})$ . The Chern-character compatibility forces $ψ^{k} (α) = k^{m} α + a_{k} β$ for some integer $a_{k}$ , and $ψ^{k} (β) = k^{2 m} β$ . The composition law $ψ^{k} ψ^{l} = ψ^{l} ψ^{k}$ applied to $α$ produces a divisibility identity, $k^{m} (l^{2 m} - l^{m}) a_{k} = l^{m} (k^{2 m} - k^{m}) a_{l}$ , which combined with $a_{2} \equiv 1 (mod 2)$ (proved by computing $ψ^{2}$ on the top cell of a CW complex with one cell each in dimensions $2 m$ and $4 m$ ) yields $2^{m} ∣ l^{m} (l^{m} - 1) a_{l}$ for every $l$ . Choosing $l = 3$ and reading off the $2$ -adic valuation gives $m \in {1, 2, 4}$ , the dimensions $2 m \in {2, 4, 8}$ . Combined with the elementary $n = 1$ case (the Hopf map $S^{1} \to S^{1}$ ), the full result is $n \in {1, 2, 4, 8}$ . Rubric: full credit for the eigenvalue computation, the divisibility identity, and the $2$ -adic argument forcing $m \in {1, 2, 4}$ .

Lean formalization [Intermediate+]

lean_status: none — Mathlib's topology library does not yet carry topological K-theory of compact Hausdorff spaces. The intended formalisation is sketched below; every step relies on infrastructure currently under construction in Mathlib.

import Mathlib.Topology.VectorBundle.Basic
import Mathlib.RingTheory.LambdaRing  -- (in progress)
import Mathlib.LinearAlgebra.ExteriorAlgebra.Basic

-- Pseudocode: topological K-theory and its λ-ring structure.
-- KTheory.toplogical X := Grothendieck (Iso (VectorBundle ℂ X))
-- with the exterior-power λ-operations.

/-- The k-th Adams operation on topological K-theory of a compact space. -/
noncomputable def AdamsPsi
    {X : Type*} [TopologicalSpace X] [CompactSpace X]
    (k : ℕ) (h : 1 ≤ k) : KTheory.topological X →+* KTheory.topological X :=
  -- defined by ψ^k = N_k(λ¹, ..., λ^k) where N_k is the k-th Newton polynomial
  sorry

/-- ψ^k on a line bundle is the k-fold tensor power. -/
theorem AdamsPsi_line_bundle
    {X : Type*} [TopologicalSpace X] [CompactSpace X]
    (L : LineBundle ℂ X) (k : ℕ) (h : 1 ≤ k) :
    AdamsPsi k h (lineBundleClass L) = lineBundleClass (L ^ k) :=
  sorry

/-- Composition law: ψ^k ∘ ψ^l = ψ^{kl}. -/
theorem AdamsPsi_comp
    {X : Type*} [TopologicalSpace X] [CompactSpace X]
    (k l : ℕ) (hk : 1 ≤ k) (hl : 1 ≤ l) :
    (AdamsPsi k hk).comp (AdamsPsi l hl) =
      AdamsPsi (k * l) (Nat.one_le_iff_ne_zero.mpr (by positivity)) :=
  sorry

The formalisation gap is substantive: Mathlib needs (a) topological vector bundles with a workable category of isomorphism classes, (b) the Grothendieck completion functor producing $K (X)$ as a commutative ring, (c) exterior-power functors $Λ^{k}$ on the bundle category descending to $K (X)$ , (d) the $λ$ -ring axioms (Atiyah-Tall convention) with verification that $K (X)$ is a special $λ$ -ring, and (e) the Newton-polynomial machinery in Mathlib.RingTheory.MvPolynomial.NewtonIdentities, which exists but is not yet wired to a $λ$ -ring consumer. The named human reviewer would verify that the existing partial $λ$ -ring development can be specialised to the topological case without forcing a refactor of the bundle library.

Advanced results [Master]

Theorem (Chern-character compatibility; Atiyah-Hirzebruch). Let $X$ be a finite CW complex and $ch : K (X) \to H^{even} (X; Q)$ the Chern character. For every $k \geq 1$ , $$ \mathrm{ch}_n(\psi^k(x)) = k^n \mathrm{ch}_n(x) \in H^{2n}(X; \mathbb{Q}), $$ where $ch_{n} (x) \in H^{2 n} (X; Q)$ is the degree- $2 n$ component of the Chern character. Equivalently, $ch \circ ψ^{k} = Ψ^{k} \circ ch$ where $Ψ^{k}$ acts on $H^{2 n}$ by the scalar $k^{n}$ .

The Chern character is the comparison map between K-theory and rational cohomology. The compatibility theorem identifies $ψ^{k}$ , viewed through $ch$ , with the cohomological Frobenius-like operation $Ψ^{k}$ that scales each degree- $2 n$ piece by $k^{n}$ . The theorem is proved by reduction to line bundles via the splitting principle, after which $ch (L) = e^{c_{1} (L)}$ and $ch (L^{\otimes k}) = e^{k c_{1} (L)}$ give the identity $ch_{n} (L^{\otimes k}) = (k c_{1} (L))^{n} / n! = k^{n} ch_{n} (L)$ directly.

Theorem (axiomatic characterisation; Atiyah 1962). The Adams operations $ψ^{k} : K (X) \to K (X)$ are uniquely characterised by the following three axioms: (i) each $ψ^{k}$ is a natural ring homomorphism, (ii) $ψ^{k} (L) = L^{\otimes k}$ for every line bundle $L$ , (iii) $ψ^{k} \circ ψ^{l} = ψ^{k l}$ for all $k, l \geq 1$ . Any family of natural ring homomorphisms ${σ^{k} : K (-) \to K (-)}_{k \geq 1}$ satisfying (ii) coincides with ${ψ^{k}}_{k \geq 1}$ .

Axiom (iii) is actually redundant given (i) and (ii) plus the splitting principle, but is traditionally included because it is the most useful structural fact about the family. The axiomatic characterisation makes Adams operations the canonical analogue, on K-theory, of the Frobenius endomorphism of an algebraic variety in positive characteristic — both detect a " $k$ -th power" structure on a ring of functions.

Theorem (Hopf-invariant-one theorem; Adams 1960). Let $f : S^{2 n - 1} \to S^{n}$ be a continuous map and let $H (f) \in Z$ be its Hopf invariant. If $H (f) = \pm 1$ then $n \in {1, 2, 4, 8}$ . The four cases are realised by the Hopf maps $S^{1} \to S^{1}$ , $S^{3} \to S^{2}$ , $S^{7} \to S^{4}$ , $S^{15} \to S^{8}$ , corresponding to the four normed division algebras $R, C, H, O$ .

The theorem closes a problem open since Hopf's 1935 paper Über die Abbildungen der dreidimensionalen Sphäre auf die Kugelfläche (Math. Ann. 104, 637-665). Adams' 1960 proof in Annals of Mathematics 72, 20-104, used secondary cohomology operations on the Steenrod algebra; Atiyah and Adams gave a shorter K-theoretic proof in 1966 (Quart. J. Math. 17, 31-38) using Adams operations directly. The K-theoretic argument is sketched in Exercise 8: a Hopf-invariant-one map produces a mapping cone $C_{f}$ whose K-theory carries a non-degenerate cup product $α^{2} = \pm β$ , and the eigenvalue compatibility $ψ^{k} (α) = k^{m} α + a_{k} β$ with $a_{2}$ odd forces $m \in {1, 2, 4}$ via a $2$ -adic divisibility argument.

The full proof of the Hopf-invariant-one theorem is its own master deepening — Atiyah's K-Theory Ch. III §2 carries it out completely, and the Annals proof of Adams is one of the canonical applications of secondary cohomology operations. The statement here records what the Adams-operation machinery contributes; the corollary that the only normed division algebras over $R$ are $R, C, H, O$ (Hurwitz 1898 for the finite-dimensional case, the Adams-Atiyah theorem closing the topological version) follows because a normed division-algebra structure on $R^{n}$ produces a Hopf-invariant-one map $S^{2 n - 1} \to S^{n}$ .

Theorem ( $λ$ -ring structure; Atiyah-Tall 1969). The pair $(K (X), {λ^{k}}_{k \geq 0})$ is a special $λ$ -ring in the sense of Atiyah-Tall: in addition to the Cartan-formula axioms for $λ^{k}$ , the operations satisfy the universal polynomial identities for $λ^{k} (x y)$ and $λ^{k} (λ^{l} (x))$ that hold for sums of line bundles. The Adams operations $ψ^{k}$ are recovered from the $λ$ -ring data through the Newton recursion, and the family ${ψ^{k}}$ uniquely determines ${λ^{k}}$ rationally.

Special $λ$ -rings are the algebraic abstraction of the structure on $K (X)$ . Examples include the representation ring of a compact group, the K-theory of an algebraic variety, and the Grothendieck ring of an essentially small symmetric monoidal abelian category. Adams operations exist on every special $λ$ -ring by the same Newton-polynomial recipe; the universal targets of the theory are the so-called Adams summands of the K-theoretic localisation, splitting $K (X)_{(p)}$ into a sum of pieces indexed by primitive idempotents of the rationalised Adams algebra.

Theorem ( $γ$ -filtration). Define $γ^{k} (x) = λ^{k} (x + k - 1)$ for $x \in K (X)$ and let $F_{γ}^{n} K (X)$ be the ideal generated by all products $γ^{k_{1}} (x_{1}) \dots γ^{k_{r}} (x_{r})$ with $k_{1} + \dots + k_{r} \geq n$ and $x_{i} \in K (X)$ . The chain $K (X) = F^{0} \supseteq F^{1} \supseteq F^{2} \supseteq \dots$ is the $γ$ -filtration, and the associated graded $gr_{γ}^{n} K (X)$ is closely related to integer cohomology $H^{2 n} (X; Z)$ — equality after tensoring with $Q$ , and a refined invariant integrally.

The $γ$ -filtration is the K-theoretic analogue of the cohomology dimension filtration. Adams operations preserve it ( $ψ^{k} F_{γ}^{n} \subseteq F_{γ}^{n}$ ) and act diagonally on the associated graded, with $ψ^{k}$ scaling $gr_{γ}^{n} K (X)$ by $k^{n}$ on the rationalisation. This is the source of many integrality results in K-theory; for example, the Adams-Riemann-Roch theorem expresses pushforward in $K$ -theory as pushforward in cohomology composed with a Todd-class correction whose construction uses the $γ$ -filtration and the Adams operations.

Synthesis. Adams operations are exactly the foundational ring-theoretic refinement that turns $K (X)$ from a bare abelian group into a structured object on which Frobenius-like operations act. The central insight is that the Newton-polynomial machinery on exterior powers produces a unique family of natural ring homomorphisms whose value on line bundles is the tensor power, and on rational cohomology these act diagonally with eigenvalues $k^{n}$ — this is exactly the cohomological Frobenius scaling that identifies $ch$ as a comparison map respecting both K-theory and rational cohomology gradings. Putting these together, the Adams operations give $K (X) \otimes Q$ a simultaneous diagonalisation by eigenvalues ${k^{n}}_{n \geq 0}$ , and the diagonalisation is the bridge between K-theoretic invariants and the cohomological ones, identifying $K (X) \otimes Q$ with $H^{even} (X; Q)$ as graded objects.

The same operations appear again in 03.09.10 (Atiyah-Singer index theorem) as the natural transformations that make the topological index compatible with the integer structure of cohomology, and in 03.08.07 (Bott periodicity) by way of the multiplicative behaviour of $ψ^{k}$ on the Bott element — $ψ^{k} (β) = k β$ in $K (S^{2})$ , generalising to $k^{n}$ on the $n$ -fold Bott element in $K (S^{2 n})$ . This is dual to the cohomological power operation on $H^{2 n} (S^{2 n}; Q)$ that scales the fundamental class by $k^{n}$ . The foundational reason for this duality is the Chern-character compatibility: $ch$ identifies $ψ^{k}$ with a graded scaling, and the Bott element is the K-theoretic shadow of the cohomological fundamental class. The Hopf-invariant-one theorem, the non-existence of normed division algebras beyond $R, C, H, O$ , the vector-field problem on spheres (Adams 1962, proving that the maximal number of linearly independent vector fields on $S^{n - 1}$ is the Radon-Hurwitz function of $n$ ), and the Adams-Riemann-Roch theorem are all consequences of how $ψ^{k}$ acts on specific low-dimensional K-groups, and the bridge to each is the eigenvalue computation $ψ^{k} = k^{n}$ on the appropriate component.

Full proof set [Master]

Proposition (Newton polynomials). There exist unique integer polynomials $N_{k} \in Z [c_{1}, \dots, c_{k}]$ such that, when $c_{i}$ is specialised to the $i$ -th elementary symmetric polynomial $e_{i} (α_{1}, \dots, α_{n})$ in any number $n \geq k$ of variables, $N_{k} (c_{1}, \dots, c_{k}) = α_{1}^{k} + \dots + α_{n}^{k}$ . The polynomials are determined by the Newton-Girard recursion $N_{k} = c_{1} N_{k - 1} - c_{2} N_{k - 2} + \dots + (- 1)^{k - 1} k c_{k}$ .

Proof. Existence is by induction. For $k = 1$ , $p_{1} = e_{1}$ , so $N_{1} = c_{1}$ . For $k \geq 2$ , the elementary identity $$ p_k = e_1 p_{k-1} - e_2 p_{k-2} + \cdots + (-1)^{k-1} k e_k $$ is verified by expanding both sides in the ring of symmetric polynomials and matching coefficients; this identity is the classical Newton-Girard formula, provable by computing $\prod_{i} (t - α_{i})$ , taking logarithmic derivatives, and matching coefficients of powers of $t$ . Substituting the inductive hypothesis $p_{j} = N_{j} (e_{1}, \dots, e_{j})$ into the right-hand side produces $N_{k}$ as a polynomial in $e_{1}, \dots, e_{k}$ . Integrality is automatic from the recursion. Uniqueness: two polynomials agreeing on the universal elementary-symmetric substitution agree as polynomials, because the universal substitution sends $Z [c_{1}, \dots, c_{k}]$ injectively into $Z [α_{1}, \dots, α_{n}]^{S_{n}}$ for $n \geq k$ . $□$

Proposition (splitting principle for $K$ -theory). Let $E \to X$ be a complex vector bundle of rank $n$ over a compact Hausdorff space $X$ . There exists a continuous map $f : F (E) \to X$ from the flag bundle $F (E)$ such that (a) the pullback $f^ E $s pl i t s a s a d i r ec t s u m o f l in e b u n d l eso v er$ F(E) $, an d (b) t h e in d u ce d r in g h o m o m or p hi s m$ f^* : K(X) \to K(F(E))$ is injective.*

Proof. Inductively construct projective bundles: $P (E) \to X$ has on it the tautological line bundle $L \subseteq π^{*} E$ , and the quotient $π^{*} E / L$ has rank $n - 1$ . Iterate $n - 1$ times to obtain the flag bundle $F (E) \to X$ , with $π^{*} E$ splitting as a direct sum of $n$ tautological line bundles. The injectivity of $f^{*}$ on K-theory at each projective-bundle step follows from the Leray-Hirsch theorem applied to the K-theory of $P (E) \to X$ : $K (P (E))$ is a free $K (X)$ -module on $1, [L], [L]^{2}, \dots, [L]^{n - 1}$ , so the pullback $K (X) \to K (P (E))$ is split-injective. Composing $n - 1$ split injections gives the injectivity of $f^{*}$ on the flag bundle. $□$

Theorem (Adams; ring homomorphism), full proof. Step (i) of the proof in Key theorem establishes additivity and multiplicativity by reduction to a sum of line bundles via the splitting principle. The reduction relies on the previous proposition: given $x, y \in K (X)$ , choose $f_{x} : F (x) \to X$ and $f_{y} : F (y) \to F (x)$ such that $f_{y}^{*} f_{x}^{*} x$ and $f_{y}^{*} f_{x}^{*} y$ are both sums of line bundles in $K (F (y))$ . Both $f_{x}^{*}$ and $f_{y}^{*}$ are injective on K-theory; the composite $f^{*} = f_{y}^{*} \circ f_{x}^{*} : K (X) \to K (F (y))$ is injective. The additivity and multiplicativity identities are then polynomial identities in the line-bundle splittings, reduced to the Newton-polynomial identity $p_{k} (α_{1}, \dots, α_{m}, β_{1}, \dots, β_{n}) = p_{k} (α) + p_{k} (β)$ and the analogous multiplicative identity. Injectivity transports them back to $K (X)$ . $□$

Theorem (Chern-character compatibility), proof. Both $ch$ and $ψ^{k}$ are natural ring homomorphisms, and both are determined by their action on line bundles by the splitting principle. It suffices to verify the compatibility on a line bundle $L$ . By definition $ch (L) = e^{c_{1} (L)} = \sum_{n \geq 0} c_{1} (L)^{n} / n!$ , so $ch_{n} (L) = c_{1} (L)^{n} / n!$ . The tensor power has first Chern class $c_{1} (L^{\otimes k}) = k c_{1} (L)$ , so $ch_{n} (L^{\otimes k}) = (k c_{1} (L))^{n} / n! = k^{n} c_{1} (L)^{n} / n! = k^{n} ch_{n} (L)$ . Hence $ch_{n} (ψ^{k} (L)) = ch_{n} (L^{\otimes k}) = k^{n} ch_{n} (L)$ . Extending by additivity and multiplicativity to sums of line bundles, and transporting back via the splitting principle, $ch_{n} (ψ^{k} (x)) = k^{n} ch_{n} (x)$ for every $x \in K (X)$ . $□$

Theorem (composition law), proof. On a line bundle, $ψ^{k} ψ^{l} (L) = ψ^{k} (L^{\otimes l}) = (L^{\otimes l})^{\otimes k} = L^{\otimes k l} = ψ^{k l} (L)$ . Both $ψ^{k} \circ ψ^{l}$ and $ψ^{k l}$ are natural ring homomorphisms by the ring-homomorphism property; they agree on line bundles; by the splitting principle and injectivity of $f^{*}$ , they agree on $K (X)$ . $□$

Theorem (axiomatic uniqueness), proof. Let $σ^{k} : K (-) \to K (-)$ be a family of natural ring homomorphisms with $σ^{k} (L) = L^{\otimes k}$ for every line bundle $L$ . On a sum $E = L_{1} \oplus \dots \oplus L_{n}$ of line bundles, $σ^{k} (E) = σ^{k} (L_{1}) + \dots + σ^{k} (L_{n}) = L_{1}^{\otimes k} + \dots + L_{n}^{\otimes k} = ψ^{k} (E)$ . Pull back along the splitting cover $f : F (E) \to X$ to obtain $f^{*} σ^{k} (E) = σ^{k} (f^{*} E) = ψ^{k} (f^{*} E) = f^{*} ψ^{k} (E)$ . Injectivity of $f^{*}$ yields $σ^{k} (E) = ψ^{k} (E)$ on $K (X)$ . $□$

Theorem (eigenspace decomposition on $K (X) \otimes Q$ ), proof. By the Atiyah-Hirzebruch theorem, the Chern character $ch : K (X) \otimes Q \to H^{even} (X; Q)$ is a ring isomorphism for any finite CW complex $X$ . The target carries the cohomological grading $H^{even} = ⨁_{n} H^{2 n}$ , and the Chern-character compatibility forces $ψ^{k}$ to act as the scalar $k^{n}$ on $ch^{- 1} (H^{2 n} (X; Q))$ . Distinct $k$ -eigenvalues across different $n$ make the decomposition simultaneous: choose $k \geq 2$ and the eigenspaces of $ψ^{k}$ correspond bijectively to the cohomological grading components. The decomposition $K (X) \otimes Q = ⨁_{n} V_{n}$ with $V_{n} = ch^{- 1} (H^{2 n} (X; Q))$ is canonical, and $ψ^{k}$ acts as $k^{n}$ on $V_{n}$ . $□$

Theorem (Hopf-invariant-one, statement). The full proof is the subject of [Adams 1960 Annals 72] using secondary cohomology operations and of [Adams-Atiyah 1966 Quart. J. Math. 17] using Adams operations on K-theory, and is its own master deepening. The K-theoretic argument is recorded in Exercise 8 as a sketch.

Connections [Master]

Topological K-theory 03.08.01. Adams operations are the ring-theoretic operations that upgrade $K (X)$ from a bare abelian group (or commutative ring, under tensor product) to a structured object on which a $Z_{\geq 1}$ -indexed family of natural ring homomorphisms acts. Without the operations, $K (X)$ is a coarse invariant; with them, it carries enough structure to detect the Hopf-invariant-one phenomenon and the vector-field problem on spheres. The unit on topological K-theory provides the ring $K (X)$ on which Adams operations are defined.
Pontryagin and Chern classes 03.06.04. The Chern character $ch : K (X) \to H^{even} (X; Q)$ is the comparison between K-theory and rational cohomology, and the Chern-character compatibility theorem makes $ψ^{k}$ a Frobenius-like operation through this comparison. On the cohomology side, $Ψ^{k}$ scales the degree- $2 n$ component by $k^{n}$ — the foundational reason eigenvalue computations on K-theory transport faithfully to cohomology. The unit on Pontryagin and Chern classes supplies the Chern character that bridges the two theories.
Chern character ring homomorphism 03.06.18. The compatibility $ch_{n} (ψ^{k} (x)) = k^{n} ch_{n} (x)$ identifies the Chern-character preimage $V_{n} = ch^{- 1} (H^{2 n} (X; Q))$ as the simultaneous $k^{n}$ -eigenspace of all Adams operations on $K (X) \otimes Q$ . The foundational reason Adams operations diagonalise on the rationalised K-theory is exactly that $ψ^{k}$ acts as $k^{n}$ in cohomological degree $2 n$ — and the Chern character of 03.06.18 is the universal coefficient that makes this transport faithful. Anchor phrase: Chern character compatibility with Adams operations.
Bott periodicity 03.08.07. The Bott element $β \in K (S^{2})$ generates a free rank-one summand of K-theory of every even sphere. Adams operations act on the Bott element by $ψ^{k} (β) = k β$ in $K (S^{2})$ , generalising to $k^{n}$ on the $n$ -fold Bott element in $K (S^{2 n})$ . The interaction between Adams operations and Bott periodicity is the source of the K-theoretic computation underlying the Hopf-invariant-one theorem, and more generally of every eigenvalue argument that reduces a global K-theory question to a sphere computation.
Atiyah-Singer index theorem 03.09.10. The topological index of an elliptic operator lives in $K (X)$ for $X$ the cotangent bundle, and Adams operations refine the integer-valued index by tracking how the symbol class decomposes under $ψ^{k}$ . The Adams-Riemann-Roch theorem (Atiyah-Hirzebruch 1959) expresses pushforward in K-theory through Adams operations and a Todd-class correction; this is the K-theoretic refinement of the Grothendieck-Riemann-Roch theorem and a direct ancestor of one of the proofs of the index theorem.
Steenrod algebra and cohomology operations. The mod- $p$ Adams operations $ψ^{p} : K (X) / p \to K (X) / p$ project to the topological Steenrod power operation in $K$ -theory, refining the Steenrod algebra action on mod- $p$ cohomology. This identification is one of the central comparison results of Atiyah's 1966 Power operations in K-theory, and it is the structural reason that secondary cohomology operations (used in Adams' 1960 proof) and Adams operations on K-theory (used in Adams-Atiyah's 1966 proof) both settle the Hopf-invariant-one problem.
Atiyah-Hirzebruch spectral sequence 03.13.04. Adams operations act on the AHSS converging to $K^{*} (X)$ : rationally, by the Chern-character compatibility above, $ψ^{k}$ acts on the $E_{2}$ page $H^{p} (X; K^{q} (pt)) \otimes Q$ as the scalar $k^{p /2}$ on each row (using that the row at fixed $q$ encodes cellular degree $p$ ). The AHSS is therefore the canonical organising frame for the $k^{n}$ -eigenvalue decomposition of $ψ^{k}$ : the eigenspaces are exactly the rows of the spectral sequence, and the $γ$ -filtration of $K (X)$ coincides rationally with the skeletal filtration that the AHSS computes. Integral lifts pull back through the spectral-sequence differentials, of which the lowest non-vanishing one is $d_{3} = Sq^{3}$ ; this is the reason eigenvalue arguments on integer K-theory must account for Steenrod-cube obstructions, and the reason the Hopf-invariant-one application (Exercise 8) runs cleanly on cell complexes where $d_{3}$ vanishes.
KR-theory (K-theory with reality) 03.08.12. KR provides the natural target for Adams operations on Real bundles — bundles equipped with a conjugate-linear lift of a base involution — generalising the present unit's operations on complex K-theory to the bigraded ring $K R^{*, *}$ . The bigraded Adams operations act on $K R^{p, q}$ by simultaneous scaling in both indices, refining the singly-graded eigenvalue table used in Adams 1962. The KR-form of the Hopf-invariant-one and division-algebra proofs (Adams' theorem on vector fields on spheres) is sharper than the KU-form because it reads the eigenvalues in the bigraded KR ring rather than via a separate $2$ -adic divisibility argument; see 03.08.12 §sub-section on Adams' theorem for the explicit bigraded eigenvalue calculus.
Equivariant K-theory and the representation ring 03.08.10. The representation ring $R (G)$ of a compact Lie group is the prototype $λ$ -ring, with exterior powers of representations defining $λ^{k}$ and the Newton-formula construction producing Adams operations $ψ^{k}$ that act on irreducible characters by the $k$ -th power map: $ψ^{k} (χ) (g) = χ (g^{k})$ . This is the canonical example that motivated the abstract $λ$ -ring axiomatics of Atiyah-Tall 1969, and it lifts to a full $λ$ -ring / Adams-operation structure on every equivariant K-theory $K_{G} (X)$ via exterior powers of $G$ -equivariant bundles. The eigenvalue computation $ψ^{k} (β_{V}) = k^{d i m V} β_{V}$ on the equivariant Bott class $β_{V} \in K_{G} (V)$ for a complex $G$ -representation $V$ generalises the non-equivariant relation $ψ^{k} (β) = k β$ in $K (S^{2})$ , and is the K-theoretic engine driving the Adams-Riemann-Roch theorem for equivariant pushforwards $K_{G} (M) \to R (G)$ . Connection type: foundational example — the representation ring is the universal $λ$ -ring on which the abstract Adams-operation axiomatics was originally tested.

Historical & philosophical context [Master]

Adams operations originated in J. F. Adams' 1962 paper Vector fields on spheres (Annals of Mathematics (2) 75, 603-632) ^{[Adams Vector fields on spheres]}, in which Adams resolved the long-standing question of how many linearly independent vector fields a sphere $S^{n - 1}$ admits. The answer is the Radon-Hurwitz function of $n$ , and the proof requires a new operation on K-theory beyond the ring structure: Adams introduced $ψ^{k}$ as a tool capable of distinguishing classes that the bare K-group cannot. The construction by Newton polynomial in exterior powers is given there, and the composition law $ψ^{k} ψ^{l} = ψ^{k l}$ is established. The same machinery had been anticipated implicitly in Grothendieck's lectures on K-theory at Bonn in 1957-1958, but Adams was the first to use it for a major topological computation.

The axiomatic characterisation and the systematic comparison with Steenrod operations appeared in Atiyah's 1966 Power operations in K-theory (Quart. J. Math. Oxford (2) 17, 165-193) ^{[Atiyah Power operations]}. Atiyah's treatment isolated the three axioms (natural ring homomorphism, action on line bundles, composition law) and proved uniqueness by the splitting principle. The companion paper by Adams and Atiyah, K-theory and the Hopf invariant (Quart. J. Math. (2) 17, 31-38, 1966) ^{[Adams-Atiyah K-theory and the Hopf invariant]}, gave the short K-theoretic proof of the Hopf-invariant-one theorem — six pages, replacing the 84 pages of secondary cohomology operations in Adams' original 1960 Annals proof On the non-existence of elements of Hopf invariant one (Annals of Math. 72, 20-104).

The Hopf-invariant-one theorem itself answered a question raised by Heinz Hopf in his 1935 paper Über die Abbildungen der dreidimensionalen Sphäre auf die Kugelfläche (Math. Ann. 104, 637-665) ^{[Hopf 1935 — pending acquisition; see NEED_TO_SOURCE]}: when does the linking number invariant of a map $S^{2 n - 1} \to S^{n}$ take the value $\pm 1$ ? Hopf himself had constructed examples in dimensions $n = 2, 4, 8$ using the complex, quaternionic, and octonionic multiplications. The conjecture that these were the only possibilities — settled by Adams in 1960 — connected the topological question to the algebraic problem of normed division algebras over $R$ , classified by Hurwitz in 1898 (Über die Composition der quadratischen Formen, Math. Ann. 88) as $R, C, H, O$ . The topological version is strictly stronger than the algebraic one — it rules out even continuous (non-associative, non-algebraic) division-algebra structures on $R^{n}$ for $n \in / {1, 2, 4, 8}$ .

The abstract algebraic framework of $λ$ -rings was developed by Atiyah and Tall in 1969 (Group representations, $λ$ -rings and the $J$ -homomorphism, Topology 8, 253-297). The notion of a special $λ$ -ring captures, in pure algebra, the structure that $K (X)$ exhibits topologically, and provides the conceptual home for Adams operations on any such ring — including representation rings of compact groups, K-theory of algebraic varieties, and Grothendieck rings of categories with exterior powers. Karoubi's 1978 textbook K-Theory: An Introduction (Springer Grundlehren 226) ^{[Karoubi K-Theory]} gives the canonical pedagogical treatment, organising the material around the Atiyah-Tall axiomatics and the Newton-polynomial construction.

Bibliography [Master]

@article{Adams1960HopfInvariant,
  author  = {Adams, J. F.},
  title   = {On the non-existence of elements of {H}opf invariant one},
  journal = {Annals of Mathematics. Second Series},
  volume  = {72},
  year    = {1960},
  pages   = {20--104}
}

@article{Adams1962VectorFields,
  author  = {Adams, J. F.},
  title   = {Vector fields on spheres},
  journal = {Annals of Mathematics. Second Series},
  volume  = {75},
  year    = {1962},
  pages   = {603--632}
}

@article{Adams1958Steenrod,
  author  = {Adams, J. F.},
  title   = {On the structure and applications of the {S}teenrod algebra},
  journal = {Commentarii Mathematici Helvetici},
  volume  = {32},
  year    = {1958},
  pages   = {180--214}
}

@article{Atiyah1966PowerOperations,
  author  = {Atiyah, M. F.},
  title   = {Power operations in {K}-theory},
  journal = {Quarterly Journal of Mathematics, Oxford. Second Series},
  volume  = {17},
  year    = {1966},
  pages   = {165--193}
}

@article{AdamsAtiyah1966Hopf,
  author  = {Adams, J. F. and Atiyah, M. F.},
  title   = {{K}-theory and the {H}opf invariant},
  journal = {Quarterly Journal of Mathematics, Oxford. Second Series},
  volume  = {17},
  year    = {1966},
  pages   = {31--38}
}

@article{AtiyahTall1969,
  author  = {Atiyah, M. F. and Tall, D. O.},
  title   = {Group representations, $\lambda$-rings and the {$J$}-homomorphism},
  journal = {Topology},
  volume  = {8},
  year    = {1969},
  pages   = {253--297}
}

@book{AtiyahKTheory,
  author    = {Atiyah, M. F.},
  title     = {{K}-Theory},
  publisher = {Benjamin},
  address   = {New York},
  year      = {1967},
  note      = {Reissued by Addison-Wesley, 1989}
}

@book{Karoubi1978,
  author    = {Karoubi, Max},
  title     = {{K}-Theory: {A}n Introduction},
  series    = {Grundlehren der mathematischen Wissenschaften},
  volume    = {226},
  publisher = {Springer-Verlag},
  year      = {1978}
}

@book{Husemoller1994,
  author    = {Husemoller, Dale},
  title     = {Fibre Bundles},
  series    = {Graduate Texts in Mathematics},
  volume    = {20},
  edition   = {3rd},
  publisher = {Springer-Verlag},
  year      = {1994}
}

@article{Hurwitz1898,
  author  = {Hurwitz, A.},
  title   = {{\"U}ber die {C}omposition der quadratischen {F}ormen von beliebig vielen {V}ariabeln},
  journal = {Nachrichten von der Gesellschaft der Wissenschaften zu G{\"o}ttingen, Mathematisch-Physikalische Klasse},
  year    = {1898},
  pages   = {309--316}
}

@article{Hopf1935,
  author  = {Hopf, Heinz},
  title   = {{\"U}ber die {A}bbildungen der dreidimensionalen {S}ph{\"a}re auf die {K}ugelfl{\"a}che},
  journal = {Mathematische Annalen},
  volume  = {104},
  year    = {1935},
  pages   = {637--665}
}

Prerequisites

03.08.01
03.05.08
03.06.04

Used in

03.08.07

Tier anchors

beginner: raising bundles to powers, by analogy with raising numbers to powers — 3Blue1Brown / Strogatz analogous level
intermediate: Atiyah *K-Theory* §3; Husemoller *Fibre Bundles* §15
master: Atiyah *K-Theory* Ch. III; Karoubi *K-Theory: An Introduction* Ch. IV; Adams *Vector fields on spheres* (Annals of Math. 75, 1962); Atiyah *Power operations in K-theory* (Quart. J. Math. 17, 1966)

References

fast-track
images/Atiyah-K-Theory-683x1024__52e3382e9f.jpg · Atiyah K-Theory Ch. III, anchor for Adams operations
TODO_REF
Atiyah — K-Theory · Ch. III, Adams operations $\psi^k$, axiomatic characterisation, Chern character compatibility
TODO_REF
Adams — Vector fields on spheres · Annals of Math. (2) 75 (1962) 603-632 — originator of Adams operations and the Hopf-invariant-one application
TODO_REF
Adams — On the structure and applications of the Steenrod algebra · Comment. Math. Helv. 32 (1958) 180-214 — Hopf-invariant-one warm-up via secondary operations
TODO_REF
Atiyah — Power operations in K-theory · Quart. J. Math. Oxford (2) 17 (1966) 165-193 — axiomatic and equivariant treatment
TODO_REF
Karoubi — K-Theory: An Introduction · Springer Grundlehren 226 (1978), Ch. IV — modern textbook treatment of $\lambda$-rings, Adams operations, and the splitting principle
TODO_REF
Husemoller — Fibre Bundles · Springer GTM 20 (3rd ed., 1994), §15 — exterior powers of bundles and the $\lambda$-ring structure on $K(X)$
TODO_REF
Adams-Atiyah — K-theory and the Hopf invariant · Quart. J. Math. (2) 17 (1966) 31-38 — short K-theoretic proof of the Hopf-invariant-one theorem

Reviewer

TBD

Estimated time

beginner: 18m
intermediate: 50m
master: 90m