Characterizations of higher derivations and higher differential torsion theories in Eilenberg-Moore categories of monads

Let $T$ be a monad on a category $\mathscr{C}$. In this paper, we introduce the notion of higher derivations on the monad $T$ and characterize them in terms of ordinary derivations on $T$. We also define higher derivations on modules over the monad $T$ in the Eilenberg-Moore category $EM_T$ and establish their characterization in a similar manner. We provide several examples that illustrate and support our results. Furthermore, we examine the conditions under which a torsion theory on $EM_T$ is higher differential, and show that this holds if and only if every higher derivation on a module $M \in EM_T$ extends uniquely to its module of quotients $Q_τ(M)$.

💡 Research Summary

The paper develops a comprehensive theory of higher derivations in the setting of monads and their Eilenberg‑Moore categories, and investigates how these higher derivations interact with hereditary torsion theories. The authors begin by defining a higher derivation on a monad (T, θ, ζ) as a family Δ = {Δ_i}_{i=0}^n of natural transformations with Δ_0 = id_T, satisfying for each i the generalized Leibniz condition

 1 * Δ_i + Δ_1 * Δ_{i‑1}+…+Δ_i * 1 ∘ θ = θ ∘ Δ_i.

When i = 1 this reduces to the ordinary derivation on a monad introduced in earlier work. The first major result (Proposition 2.7) shows that, in any F‑linear category, a higher derivation can be expressed recursively in terms of ordinary derivations δ_k:

 (i + 1) Δ_{i+1} = ∑{k=0}^i δ{k+1} ∘ Δ_{i‑k}.

Thus each Δ_i is a linear combination of compositions of ordinary derivations, and in particular the family Δ_n = δ^n/n! (for a fixed ordinary derivation δ) yields a canonical higher derivation. Corollary 2.8 extends this description to finite‑order higher derivations.

Next the authors lift the notion to modules over the monad, i.e. objects (M, f_M) of the Eilenberg‑Moore category EM_T. A higher Δ‑derivation on M is a family D = {D_i}_{i=0}^n of additive maps with D_0 = id_M satisfying

 D_i ∘ f_M = ∑{k=0}^i Δ_k ∘ T(D{i‑k}) ∘ f_M.

This condition guarantees compatibility with both the monad multiplication θ and the module action f_M. The authors prove (Theorem 3.5) that, again under the F‑linearity hypothesis, such a higher Δ‑derivation is completely determined by an associated sequence of ordinary derivations on M, mirroring the situation for the monad itself.

The paper then turns to torsion theory. Assuming C is a Grothendieck category, T is an exact, colimit‑preserving monad, and τ = (𝒯,ℱ) is a hereditary torsion theory on EM_T, the authors study when τ is “higher differential” of order n, meaning that every higher Δ‑derivation on any T‑module restricts to the τ‑torsion submodule. Using the correspondence between hereditary torsion theories on EM_T and families of Gabriel filters {L_{T G_i}} (as established in prior work), they formulate four equivalent conditions (Theorem 4.8, labelled Theorem A):

1. Every Gabriel filter associated to τ is Δ‑invariant (i.e., for each filter L and each j ≤ n there exists K∈L such that (Δ_j)_{G_i}(K)⊆ker(T G_i→N) for any submodule N of the torsion part).
2. τ is a higher differential torsion theory of order n.
3. For any T‑module M, any higher Δ‑derivation D, any submodule N⊆M_τ, and each j≤n, there exists a filter element K satisfying the same kernel condition.
4. The same as (3) but uniformly for all j simultaneously.

These equivalences show that the higher differential property is precisely the invariance of the Gabriel filters under the higher derivation operators.

Finally, the authors address the extension problem. Theorem 5.6 (Theorem B) proves that a hereditary torsion theory τ is higher differential of order n if and only if every higher Δ‑derivation on a τ‑torsion‑free module M extends uniquely to a higher Δ‑derivation on its module of quotients Q_τ(M). The proof builds on the Δ‑invariance of the Gabriel filters (from Theorem A) to construct the extension and then adapts Bland’s uniqueness argument from the classical differential case to the higher setting.

Throughout the paper, several illustrative examples are provided. Example 2.9 shows that for any ordinary derivation δ, the family Δ_n = δ^n/n! yields a higher derivation. Example 2.10 constructs a “periodic” higher derivation that vanishes on indices not divisible by a fixed integer k. Example 2.11 demonstrates how a fixed natural transformation μ: 1_C→T can be used to build a non‑trivial higher derivation via recursive formulas involving θ and μ. These examples confirm that the abstract definitions capture a wide range of concrete situations, including monads arising from algebraic structures such as free algebras, list constructions, and more.

In summary, the paper achieves three major contributions: (1) it introduces and characterizes higher derivations on monads and on their module categories, showing they are generated by ordinary derivations; (2) it establishes a precise equivalence between higher differential torsion theories and Δ‑invariance of Gabriel filters; and (3) it proves that higher differential torsion theories are exactly those for which higher derivations extend uniquely to quotient modules. This work unifies higher differential algebra with categorical torsion theory, opening the door for further applications in homological algebra, non‑commutative geometry, and the study of differential structures on categorical constructions.