Singer Algebraic Transfer in Steenrod Algebra

• Singer algebraic transfer is a family of maps defined at the prime 2 that link GLₖ(F₂)-invariants in hit modules to Ext groups of the Steenrod algebra.
• It admits dual formulations—homological (epimorphism) and cohomological (monomorphism)—highlighting its fundamental role in connecting modular invariant theory with stable homotopy groups.
• Recent research leverages algorithmic invariant theory and generic degree techniques to extend transfer computations to higher ranks and refine understanding of the hit problem.

Singer algebraic transfer is a family of maps, defined at the prime $2$, that relates the $\mathbb{F}_2$-cohomology and homology of elementary abelian $2$-groups to the cohomology of the Steenrod algebra. In its most common form, it connects $GL_k(\mathbb{F}_2)$-invariants in the cohit module $QP_k=\mathbb{F}_2\otimes_{\mathcal A}P_k$, where $P_k=\mathbb{F}_2[x_1,\dots,x_k]$, with $$\operatorname{Ext}_{\mathcal A}^{k,*}(\mathbb{F}_2,\mathbb{F}_2)$$, the $E_2$-term of the Adams spectral sequence for stable homotopy groups of spheres. The transfer has therefore occupied a central place in computations at the interface of Steenrod operations, modular invariant theory, the Peterson hit problem, and Adams-spectral-sequence detection theory (Phuc, 2021).

1. Dual formulations and standard notation

A persistent feature of the literature is that the transfer appears in dual formulations. In the older homological formulation, one writes

$$\varphi_k:\operatorname{Tor}^{\mathcal A}_{k,k+d}(\mathbb F_2,\mathbb F_2)\longrightarrow (QP_k)_d^{GL_k},$$

with $QP_k=\mathbb F_2\otimes_{\mathcal A}P_k$, and Singer’s conjecture is phrased as surjectivity of $\mathbb{F}_2$0 (Sum, 2017). In the dual cohomological formulation, one writes

$\mathbb{F}_2$1

and the conjecture is phrased as injectivity of $\mathbb{F}_2$2 (Phuc, 2021).

These are not competing definitions but degreewise dual avatars of the same construction. The duality

$\mathbb{F}_2$3

explains why one part of the literature speaks about epimorphisms and another about monomorphisms.

Formulation	Map	Conjectural wording
Homological	$\mathbb{F}_2$4	epimorphism/surjectivity
Cohomological	$\mathbb{F}_2$5	monomorphism/injectivity

Notational variation is substantial. The symbols $\mathbb{F}_2$6, $\mathbb{F}_2$7, and $\mathbb{F}_2$8 all occur, sometimes for the homological map and sometimes for its dual. The underlying object is the same transfer phenomenon from $\mathbb{F}_2$9-data to $2$0.

2. Algebraic and topological setting

At the prime $2$1, the Steenrod algebra $2$2 is the graded algebra of stable natural endomorphisms of mod-$2$3 cohomology, generated by the Steenrod squares and governed by the Adem relations. Its cohomology

$2$4

is the $2$5-term of the Adams spectral sequence

$2$6

so transfer calculations feed directly into computations of stable homotopy groups of spheres (Phuc, 2021).

For an elementary abelian $2$7-group $2$8,

$2$9

The hit problem asks for a minimal set of $GL_k(\mathbb{F}_2)$0-generators of $GL_k(\mathbb{F}_2)$1, equivalently for a basis of

$GL_k(\mathbb{F}_2)$2

The $GL_k(\mathbb{F}_2)$3-action on $GL_k(\mathbb{F}_2)$4 induces an action on $GL_k(\mathbb{F}_2)$5 commuting with $GL_k(\mathbb{F}_2)$6, hence on $GL_k(\mathbb{F}_2)$7. This makes $GL_k(\mathbb{F}_2)$8 the natural invariant-theoretic target of the homological transfer (Sum, 2017).

Dually,

$GL_k(\mathbb{F}_2)$9

the divided power algebra. The primitive subspace

$QP_k=\mathbb{F}_2\otimes_{\mathcal A}P_k$0

provides the natural source of the cohomological transfer. This divided-power description is not auxiliary; it is the mechanism by which explicit transfer classes are constructed in practice (Phuc, 2021).

Geometrically, the transfer is induced on Adams $QP_k=\mathbb{F}_2\otimes_{\mathcal A}P_k$1-terms by the stable transfer

$QP_k=\mathbb{F}_2\otimes_{\mathcal A}P_k$2

Accordingly, Singer transfer is neither purely combinatorial nor purely representation-theoretic: it packages a topological transfer through algebraic models of $QP_k=\mathbb{F}_2\otimes_{\mathcal A}P_k$3-cohomology.

3. Constructions, models, and related functors

Singer’s original algebraic construction may be expressed in terms of extension classes built from the one-variable unstable $QP_k=\mathbb{F}_2\otimes_{\mathcal A}P_k$4-module $QP_k=\mathbb{F}_2\otimes_{\mathcal A}P_k$5 and its Laurent extension. Tensoring the basic one-variable extension produces a class $QP_k=\mathbb{F}_2\otimes_{\mathcal A}P_k$6, and cap product with $QP_k=\mathbb{F}_2\otimes_{\mathcal A}P_k$7 yields the map underlying the transfer. In dual form this gives the familiar map from primitives or coinvariants to $QP_k=\mathbb{F}_2\otimes_{\mathcal A}P_k$8 (Phuc, 2021).

A major computational model is the mod-$QP_k=\mathbb{F}_2\otimes_{\mathcal A}P_k$9 lambda algebra $P_k=\mathbb{F}_2[x_1,\dots,x_k]$0, a differential graded algebra with

$P_k=\mathbb{F}_2[x_1,\dots,x_k]$1

Chơn–Hằng and Hùng constructed maps

$P_k=\mathbb{F}_2[x_1,\dots,x_k]$2

whose values on primitives are cycles, and whose homology classes are exactly the transfer images: $P_k=\mathbb{F}_2[x_1,\dots,x_k]$3 This makes the transfer calculable by translating divided-power monomials into explicit lambda-algebra cycles, then recognizing the resulting $P_k=\mathbb{F}_2[x_1,\dots,x_k]$4-classes in the $P_k=\mathbb{F}_2[x_1,\dots,x_k]$5-, $P_k=\mathbb{F}_2[x_1,\dots,x_k]$6-, $P_k=\mathbb{F}_2[x_1,\dots,x_k]$7-, and $P_k=\mathbb{F}_2[x_1,\dots,x_k]$8-families (Phuc, 2021).

A second, conceptually distinct, framework is the Singer construction $P_k=\mathbb{F}_2[x_1,\dots,x_k]$9 on an unstable $$\operatorname{Ext}_{\mathcal A}^{k,*}(\mathbb{F}_2,\mathbb{F}_2)$$0-module $$\operatorname{Ext}_{\mathcal A}^{k,*}(\mathbb{F}_2,\mathbb{F}_2)$$1. Here $$\operatorname{Ext}_{\mathcal A}^{k,*}(\mathbb{F}_2,\mathbb{F}_2)$$2 is a structured $$\operatorname{Ext}_{\mathcal A}^{k,*}(\mathbb{F}_2,\mathbb{F}_2)$$3-module, where $$\operatorname{Ext}_{\mathcal A}^{k,*}(\mathbb{F}_2,\mathbb{F}_2)$$4 is the Dickson algebra. Passage to indecomposables produces a natural map

$$\operatorname{Ext}_{\mathcal A}^{k,*}(\mathbb{F}_2,\mathbb{F}_2)$$5

and this map is the dual of the composite of the Lannes–Zarati homomorphism with Singer’s algebraic transfer. The theorem that this morphism is trivial on positive-degree elements for $$\operatorname{Ext}_{\mathcal A}^{k,*}(\mathbb{F}_2,\mathbb{F}_2)$$6 provides a weak generalized algebraic spherical class result and places the transfer in the broader context of Hurewicz-image questions (Hung et al., 2016).

These two viewpoints serve different purposes. The lambda algebra is tailored to explicit detection in $$\operatorname{Ext}_{\mathcal A}^{k,*}(\mathbb{F}_2,\mathbb{F}_2)$$7; the $$\operatorname{Ext}_{\mathcal A}^{k,*}(\mathbb{F}_2,\mathbb{F}_2)$$8-Lannes–Zarati framework emphasizes structural constraints on what the transfer can see after destabilization and passage to indecomposables.

4. Established calculations and low-rank structure

The transfer is completely understood in low rank. Singer proved that rank $$\operatorname{Ext}_{\mathcal A}^{k,*}(\mathbb{F}_2,\mathbb{F}_2)$$9 and rank $E_2$0 transfers are isomorphisms, and Boardman proved the same for rank $E_2$1. A later hit-problem-based treatment reproved these facts and gave a detailed determination of the fourth transfer in several infinite degree families (Sum, 2017).

Subsequent work extended positive results substantially, though not uniformly.

Rank and degree family	Result	Reference
$E_2$2, all degrees	isomorphism	(Sum, 2017)
$E_2$3, $E_2$4, $E_2$5, $E_2$6	isomorphism except $E_2$7, where injective not surjective	(Sum, 2017)
$E_2$8, $E_2$9	isomorphism	(Phuc, 2021)
$$\varphi_k:\operatorname{Tor}^{\mathcal A}_{k,k+d}(\mathbb F_2,\mathbb F_2)\longrightarrow (QP_k)_d^{GL_k},$$0, $$\varphi_k:\operatorname{Tor}^{\mathcal A}_{k,k+d}(\mathbb F_2,\mathbb F_2)\longrightarrow (QP_k)_d^{GL_k},$$1	isomorphism	(Sum, 2024)
$$\varphi_k:\operatorname{Tor}^{\mathcal A}_{k,k+d}(\mathbb F_2,\mathbb F_2)\longrightarrow (QP_k)_d^{GL_k},$$2, $$\varphi_k:\operatorname{Tor}^{\mathcal A}_{k,k+d}(\mathbb F_2,\mathbb F_2)\longrightarrow (QP_k)_d^{GL_k},$$3	isomorphism	(Phuc, 13 Nov 2025)

Rank $$\varphi_k:\operatorname{Tor}^{\mathcal A}_{k,k+d}(\mathbb F_2,\mathbb F_2)\longrightarrow (QP_k)_d^{GL_k},$$4 has been especially active. Earlier infinite families such as $$\varphi_k:\operatorname{Tor}^{\mathcal A}_{k,k+d}(\mathbb F_2,\mathbb F_2)\longrightarrow (QP_k)_d^{GL_k},$$5 and $$\varphi_k:\operatorname{Tor}^{\mathcal A}_{k,k+d}(\mathbb F_2,\mathbb F_2)\longrightarrow (QP_k)_d^{GL_k},$$6 were handled by combining Kameko iteration with hit-problem calculations, and in those families Singer’s conjecture was verified degreewise (Sum, 2016, Tin, 2016). The rank-$$\varphi_k:\operatorname{Tor}^{\mathcal A}_{k,k+d}(\mathbb F_2,\mathbb F_2)\longrightarrow (QP_k)_d^{GL_k},$$7 transfer also detects concrete nonzero classes such as

$$\varphi_k:\operatorname{Tor}^{\mathcal A}_{k,k+d}(\mathbb F_2,\mathbb F_2)\longrightarrow (QP_k)_d^{GL_k},$$8

through explicit primitive cycles in the lambda algebra, which showed that the fifth transfer captures nontrivial multiplicative structure in $$\varphi_k:\operatorname{Tor}^{\mathcal A}_{k,k+d}(\mathbb F_2,\mathbb F_2)\longrightarrow (QP_k)_d^{GL_k},$$9 rather than only isolated indecomposables (Phuc, 2021).

The rank-$QP_k=\mathbb F_2\otimes_{\mathcal A}P_k$0 and rank-$QP_k=\mathbb F_2\otimes_{\mathcal A}P_k$1 papers from 2021 and 2025 sharpened the picture further by proving isomorphism in broad generic families, often after identifying an invariant or coinvariant source of dimension $QP_k=\mathbb F_2\otimes_{\mathcal A}P_k$2 and matching it with a unique $QP_k=\mathbb F_2\otimes_{\mathcal A}P_k$3-class. This body of work established that the transfer can remain highly effective well beyond the classical $QP_k=\mathbb F_2\otimes_{\mathcal A}P_k$4 range, but only in degree families where the hit problem and the $QP_k=\mathbb F_2\otimes_{\mathcal A}P_k$5-module structure are tractable.

5. Computational frameworks and generic-degree methods

The modern study of Singer transfer is inseparable from the hit problem. Wood’s vanishing criterion, Kameko’s squaring operation, weight-vector filtrations, and tests for strictly inadmissible monomials are the standard reduction tools. In rank $QP_k=\mathbb F_2\otimes_{\mathcal A}P_k$6, several papers exploited the generic-degree form

$QP_k=\mathbb F_2\otimes_{\mathcal A}P_k$7

or related specializations, computed explicit bases of admissible monomials, and then determined $QP_k=\mathbb F_2\otimes_{\mathcal A}P_k$8-invariant subspaces by analyzing the action of generators of $QP_k=\mathbb F_2\otimes_{\mathcal A}P_k$9 on those bases (Phuc, 2018, Tin, 2021).

A recurring pattern is the use of Kameko’s homomorphism

$\mathbb{F}_2$00

When $\mathbb{F}_2$01, it is an isomorphism, so high-degree invariant problems reduce to lower-degree ones. This reduction underlies many “generic degree” theorems and explains why infinite families can sometimes be controlled from a handful of base cases (Phuc, 2021).

Recent work has shifted decisively toward algorithmic invariant theory. One direction introduced an OSCAR-based algorithm for computing $\mathbb{F}_2$02-invariants of the kernel of the Kameko homomorphism and used it to compute the domain of the sixth transfer in a nontrivial bidegree (Phuc, 11 Sep 2025). Another direction introduced a Global Cluster Analysis algorithm that constructs a weight interaction graph, clusters interacting weight spaces into closed $\mathbb{F}_2$03-submodules, computes global $\mathbb{F}_2$04-invariants clusterwise, and then imposes the remaining $\mathbb{F}_2$05-constraints. That method was designed precisely because weight spaces are generally not stable under the symmetric-group action, so purely local weightwise calculations can miss global invariant structure (Phuc, 7 Aug 2025).

High-rank hit-problem computations continue to feed transfer theory. For $\mathbb{F}_2$06, one family of results proves

$\mathbb{F}_2$07

matching the order of $\mathbb{F}_2$08, and uses these computations to analyze transfer domains and show that certain nonzero $\mathbb{F}_2$09-classes are not in the transfer image (Phuc, 2024). This suggests that the long-term study of Singer transfer will depend at least as much on scalable invariant-space algorithms as on traditional hand calculations.

6. Conjectures, counterexamples, and present status

A common misconception is that “Singer’s conjecture” has a single unchanged statement throughout the literature. In fact, older papers typically formulate the conjecture as surjectivity of the homological transfer, while more papers often formulate it as injectivity of the dual cohomological transfer. These are dual degreewise claims, so the change is terminological rather than substantive. What has changed materially is the global status of the conjecture.

For the homological formulation, a decisive counterexample was given in rank $\mathbb{F}_2$10 and internal degree $\mathbb{F}_2$11. In that case,

$\mathbb{F}_2$12

has zero source but nonzero target, because an explicit nonzero invariant class $\mathbb{F}_2$13 was constructed. Hence $\mathbb{F}_2$14 is not surjective, and Singer’s conjecture in the epimorphism form is false (Sum, 2024).

For the cohomological formulation, a counterexample was then produced for the sixth transfer in bidegree $\mathbb{F}_2$15. There the domain of the transfer has dimension $\mathbb{F}_2$16, generated by two explicit invariant classes, while

$\mathbb{F}_2$17

has dimension $\mathbb{F}_2$18. The sixth transfer is therefore not injective, and the monomorphism form of Singer’s conjecture is also false (Phuc, 11 Sep 2025).

These negative results do not erase the large body of positive degreewise theorems. They show instead that the transfer is fundamentally degree-sensitive. In some families it is an isomorphism; in others it is injective but misses substantial $\mathbb{F}_2$19-classes; in still others it fails already at the level of global dimension comparison. Related rank-$\mathbb{F}_2$20 results show that even when the domain vanishes and injectivity is vacuous, important $\mathbb{F}_2$21-classes such as $\mathbb{F}_2$22 and $\mathbb{F}_2$23 may lie outside the transfer image (Phuc, 2024). This suggests that the transfer detects a structured but proper subregion of $\mathbb{F}_2$24, rather than furnishing a universal description.

Singer algebraic transfer nevertheless remains fundamental. It still provides one of the most concrete bridges from $\mathbb{F}_2$25-geometry and $\mathbb{F}_2$26-module combinatorics to Adams $\mathbb{F}_2$27-classes, and it continues to organize computations in the cohomology of the Steenrod algebra even after the failure of its strongest conjectural forms.