/- Copyright 2025 The Formal Conjectures Authors. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at https://www.apache.org/licenses/LICENSE-2.0 Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. -/ module public import FormalConjecturesUtil

Köthe conjecture

Reference: Wikipedia

@[expose] public sectionopen Ideal TwoSidedIdeal Polynomialopen Matrixuniverse u v wvariable {R : Type*}variable [Ring R]namespace Koethe

Say a subset I of a ring R is nilpotent if all its elements are nilpotent.

def IsNil {S : Type*} [SetLike S R] (I : S) := ∀ i ∈ I, IsNilpotent i-- TODO(lezeau): add some basic API and already known results for nil ideals variable (R) in

The Kothe Radical of a ring R is the sum of all (two-sided) nil ideals of R. Tags: Kothe Radical, upper nilradical

def KotheRadical : TwoSidedIdeal R := sSup {I : TwoSidedIdeal R | IsNil I}-- This is often denoted `Nil*(R)` local notation "Nil* " R => KotheRadical R

A subset of a nil subset is nil.

@[category API, AMS 16] theorem IsNil.mono {S : Type*} [SetLike S R] {I J : S} (h : (I : Set R) ⊆ J) (hJ : IsNil J) : IsNil I := fun i hi => hJ i (h hi)

The zero two-sided ideal is nil.

R:Type u_1inst✝:Ring Rx:Rhx:x = 0⊢ IsNilpotent x All goals completed! 🐙

The sum of two nil two-sided ideals is nil.

All goals completed! 🐙⟩variable (R) in

The Köthe radical is a nil ideal: every element of Nil* R is nilpotent. This is a standard fact which does not depend on the Köthe conjecture.

R:Type u_1inst✝:Ring RK:TwoSidedIdeal R := mk' {x | ∃ I, IsNil I ∧ x ∈ I} ⋯ ⋯ ⋯ ⋯ ⋯hK:IsNil K⊢ IsNil (Nil* R) R:Type u_1inst✝:Ring RK:TwoSidedIdeal R := mk' {x | ∃ I, IsNil I ∧ x ∈ I} ⋯ ⋯ ⋯ ⋯ ⋯hK:IsNil KI:TwoSidedIdeal RhI:I ∈ {I | IsNil I}⊢ I ≤ K R:Type u_1inst✝:Ring RK:TwoSidedIdeal R := mk' {x | ∃ I, IsNil I ∧ x ∈ I} ⋯ ⋯ ⋯ ⋯ ⋯hK:IsNil KI:TwoSidedIdeal RhI:I ∈ {I | IsNil I}x:Rhx:x ∈ I⊢ x ∈ K All goals completed! 🐙

If L is a nil left ideal, then so is L * y for any y.

All goals completed! 🐙

The Köthe conjecture: In any ring, the sum of two nil left ideals is nil.

@[category research open, AMS 16] theorem KotheConjecture (I J : Ideal R) (hI : IsNil I) (hJ : IsNil J) : IsNil (I + J) := R:Type u_1inst✝:Ring RI:Ideal RJ:Ideal RhI:IsNil IhJ:IsNil J⊢ IsNil (I + J) All goals completed! 🐙

The Köthe conjecture: every left nil radical is contained in the Köthe radical.

@[category research open, AMS 16] theorem KotherConjecture.variants.le_KotherRadical {I : Ideal R} (hI : IsNil I) : (I : Set R) ⊆ KotheRadical R := R:Type u_1inst✝:Ring RI:Ideal RhI:IsNil I⊢ ↑I ⊆ ↑(Nil* R) All goals completed! 🐙open scoped Classical in

The Köthe conjecture: for any nil ideal I of R, the matrix ideal M_n(I) is a nil ideal of the matrix ring M_n(R).

@[category research open, AMS 16] theorem KotherConjecture.variants.general_matrix {I : TwoSidedIdeal R} (hI : IsNil I) (n : Type*) [Fintype n] : IsNil (matrix n I) := R:Type u_1inst✝¹:Ring RI:TwoSidedIdeal RhI:IsNil In:Type u_2inst✝:Fintype n⊢ IsNil (TwoSidedIdeal.matrix n I) All goals completed! 🐙

The Köthe conjecture: for any nil ideal I of R, the matrix ideal M_2(I) is a nil ideal of the matrix ring M_2(R).

@[category research open, AMS 16] theorem KotherConjecture.variants.two_by_two_matrix {I : TwoSidedIdeal R} (hI : IsNil I) : IsNil (matrix (Fin 2) I) := R:Type u_1inst✝:Ring RI:TwoSidedIdeal RhI:IsNil I⊢ IsNil (TwoSidedIdeal.matrix (Fin 2) I) All goals completed! 🐙open scoped Classical in

The Köthe conjecture: for any finite index type n, the Köthe radical of the matrix ring M_n(R) is the ideal M_n(Nil*(R)) of matrices with entries in the Köthe radical of R.

@[category research open, AMS 16] theorem KotherConjecture.variants.matrixOver_KotherRadical (n : Type*) [Fintype n] : matrix n (Nil* R) = Nil* (Matrix n n R) := R:Type u_1inst✝¹:Ring Rn:Type u_2inst✝:Fintype n⊢ TwoSidedIdeal.matrix n (Nil* R) = Nil* Matrix n n R All goals completed! 🐙/- ### Relations between the different formulations All the formulations of the Köthe conjecture above are known to be equivalent (Krempa). We record here the implications that have a short formal proof. In particular each of `KotheConjecture`, `KotherConjecture.variants.le_KotherRadical`, `KotherConjecture.variants.two_by_two_matrix` and `KotherConjecture.variants.matrixOver_KotherRadical` implies `KotherConjecture.variants.general_matrix`, so that a counterexample to the latter immediately yields counterexamples to all the others. The implication `KotherConjecture.variants.general_matrix → KotheConjecture` (also due to Krempa) is not formalized here. -/ section ColumnIdealvariable {n : Type*} [Fintype n] [DecidableEq n]

The left ideal of Matrix n n R consisting of the matrices with entries in I which are supported on the j-th column.

All goals completed! 🐙

Powers of a matrix supported on the j-th column are controlled by the (j, j) entry.

All goals completed! 🐙

For a nil ideal I, the column ideals columnIdeal I j are nil left ideals.

All goals completed! 🐙

The matrix ideal M_n(I) is the sum of the column ideals columnIdeal I j.

R:Type u_1inst✝²:Ring Rn:Type u_2inst✝¹:Fintype ninst✝:DecidableEq nI:TwoSidedIdeal RM:Matrix n n RhM:M ∈ TwoSidedIdeal.matrix n Ithis:M = ∑ j, of fun i k ↦ if k = j then M i k else 0⊢ (∑ j, of fun i k ↦ if k = j then M i k else 0) ∈ ⨆ j, columnIdeal I j R:Type u_1inst✝²:Ring Rn:Type u_2inst✝¹:Fintype ninst✝:DecidableEq nI:TwoSidedIdeal RM:Matrix n n RhM:M ∈ TwoSidedIdeal.matrix n Ithis:M = ∑ j, of fun i k ↦ if k = j then M i k else 0j:nx✝:j ∈ Finset.univi:nk:n⊢ of (fun i k ↦ if k = j then M i k else 0) i k ∈ I ∧ (k ≠ j → of (fun i k ↦ if k = j then M i k else 0) i k = 0) R:Type u_1inst✝²:Ring Rn:Type u_2inst✝¹:Fintype ninst✝:DecidableEq nI:TwoSidedIdeal RM:Matrix n n RhM:M ∈ TwoSidedIdeal.matrix n Ithis:M = ∑ j, of fun i k ↦ if k = j then M i k else 0j:nx✝:j ∈ Finset.univi:nk:n⊢ (if k = j then M i k else 0) ∈ I ∧ (k ≠ j → (if k = j then M i k else 0) = 0) R:Type u_1inst✝²:Ring Rn:Type u_2inst✝¹:Fintype ninst✝:DecidableEq nI:TwoSidedIdeal RM:Matrix n n RhM:M ∈ TwoSidedIdeal.matrix n Ithis:M = ∑ j, of fun i k ↦ if k = j then M i k else 0j:nx✝:j ∈ Finset.univi:nk:n⊢ (if k = j then M i k else 0) ∈ I R:Type u_1inst✝²:Ring Rn:Type u_2inst✝¹:Fintype ninst✝:DecidableEq nI:TwoSidedIdeal RM:Matrix n n RhM:M ∈ TwoSidedIdeal.matrix n Ithis:M = ∑ j, of fun i k ↦ if k = j then M i k else 0j:nx✝:j ∈ Finset.univi:nk:nh✝:k = j⊢ M i k ∈ IR:Type u_1inst✝²:Ring Rn:Type u_2inst✝¹:Fintype ninst✝:DecidableEq nI:TwoSidedIdeal RM:Matrix n n RhM:M ∈ TwoSidedIdeal.matrix n Ithis:M = ∑ j, of fun i k ↦ if k = j then M i k else 0j:nx✝:j ∈ Finset.univi:nk:nh✝:¬k = j⊢ 0 ∈ I R:Type u_1inst✝²:Ring Rn:Type u_2inst✝¹:Fintype ninst✝:DecidableEq nI:TwoSidedIdeal RM:Matrix n n RhM:M ∈ TwoSidedIdeal.matrix n Ithis:M = ∑ j, of fun i k ↦ if k = j then M i k else 0j:nx✝:j ∈ Finset.univi:nk:nh✝:k = j⊢ M i k ∈ I All goals completed! 🐙 R:Type u_1inst✝²:Ring Rn:Type u_2inst✝¹:Fintype ninst✝:DecidableEq nI:TwoSidedIdeal RM:Matrix n n RhM:M ∈ TwoSidedIdeal.matrix n Ithis:M = ∑ j, of fun i k ↦ if k = j then M i k else 0j:nx✝:j ∈ Finset.univi:nk:nh✝:¬k = j⊢ 0 ∈ I All goals completed! 🐙end ColumnIdeal

If the Köthe conjecture holds (for rings in universe w), then a finite sum of nil left ideals is nil.

h:∀ {R : Type w} [inst : Ring R] (I J : Ideal R), IsNil I → IsNil J → IsNil (I + J)S:Type winst✝¹:Ring Sι:Type u_2inst✝:Fintype ιL:ι → Ideal ShL:∀ (i : ι), IsNil (L i)x:Shx:x = 0⊢ IsNilpotent x All goals completed! 🐙

The Köthe conjecture (for rings in universe max u v) implies the general matrix formulation for rings in universe u and index types in universe v.

@[category research solved, AMS 16] theorem KotherConjecture.variants.general_matrix_of_KotheConjecture (h : type_of% @KotheConjecture.{max u v}) : type_of% @KotherConjecture.variants.general_matrix.{u, v} := h:∀ {R : Type (max u v)} [inst : Ring R] (I J : Ideal R), IsNil I → IsNil J → IsNil (I + J)⊢ ∀ {R : Type u} [inst : Ring R] {I : TwoSidedIdeal R}, IsNil I → ∀ (n : Type v) [inst_1 : Fintype n], IsNil (TwoSidedIdeal.matrix n I) classical h:∀ {R : Type (max u v)} [inst : Ring R] (I J : Ideal R), IsNil I → IsNil J → IsNil (I + J)R:Type uinst✝¹:Ring RI:TwoSidedIdeal RhI:IsNil In:Type vinst✝:Fintype nM:Matrix n n RhM:M ∈ TwoSidedIdeal.matrix n I⊢ IsNilpotent M All goals completed! 🐙

If every nil left ideal is contained in the Köthe radical, then the Köthe conjecture holds.

h:∀ {R : Type u} [inst : Ring R] {I : Ideal R}, IsNil I → ↑I ⊆ ↑(Nil* R)R:Type uinst✝:Ring RI:Ideal RJ:Ideal RhI:IsNil IhJ:IsNil Jx:Rhx:∃ y ∈ I, ∃ z ∈ J, y + z = x⊢ IsNilpotent x h:∀ {R : Type u} [inst : Ring R] {I : Ideal R}, IsNil I → ↑I ⊆ ↑(Nil* R)R:Type uinst✝:Ring RI:Ideal RJ:Ideal RhI:IsNil IhJ:IsNil Jy:Rhy:y ∈ Iz:Rhz:z ∈ J⊢ IsNilpotent (y + z) All goals completed! 🐙

The Köthe conjecture implies that every nil left ideal is contained in the Köthe radical.

h:∀ {R : Type u} [inst : Ring R] (I J : Ideal R), IsNil I → IsNil J → IsNil (I + J)R:Type uinst✝:Ring RI:Ideal RhI:IsNil IK:TwoSidedIdeal R := mk' {x | ∃ L, IsNil L ∧ x ∈ L} ⋯ ⋯ ⋯ ⋯ ⋯hK:IsNil K⊢ ↑I ⊆ ↑(Nil* R) h:∀ {R : Type u} [inst : Ring R] (I J : Ideal R), IsNil I → IsNil J → IsNil (I + J)R:Type uinst✝:Ring RI:Ideal RhI:IsNil IK:TwoSidedIdeal R := mk' {x | ∃ L, IsNil L ∧ x ∈ L} ⋯ ⋯ ⋯ ⋯ ⋯hK:IsNil Kx:Rhx:x ∈ ↑I⊢ x ∈ ↑(Nil* R) h:∀ {R : Type u} [inst : Ring R] (I J : Ideal R), IsNil I → IsNil J → IsNil (I + J)R:Type uinst✝:Ring RI:Ideal RhI:IsNil IK:TwoSidedIdeal R := mk' {x | ∃ L, IsNil L ∧ x ∈ L} ⋯ ⋯ ⋯ ⋯ ⋯hK:IsNil Kx:Rhx:x ∈ ↑IhxK:x ∈ K⊢ x ∈ ↑(Nil* R) All goals completed! 🐙

If every nil left ideal is contained in the Köthe radical (for rings in universe max u v), then the general matrix formulation holds.

@[category research solved, AMS 16] theorem KotherConjecture.variants.general_matrix_of_le_KotherRadical (h : type_of% @KotherConjecture.variants.le_KotherRadical.{max u v}) : type_of% @KotherConjecture.variants.general_matrix.{u, v} := general_matrix_of_KotheConjecture (KotheConjecture_of_le_KotherRadical h)

The general matrix formulation of the Köthe conjecture trivially implies the 2 × 2 one.

@[category test, AMS 16] theorem KotherConjecture.variants.two_by_two_matrix_of_general_matrix (h : type_of% @KotherConjecture.variants.general_matrix.{u, 0}) : type_of% @KotherConjecture.variants.two_by_two_matrix.{u} := h:∀ {R : Type u} [inst : Ring R] {I : TwoSidedIdeal R}, IsNil I → ∀ (n : Type) [inst_1 : Fintype n], IsNil (TwoSidedIdeal.matrix n I)⊢ ∀ {R : Type u} [inst : Ring R] {I : TwoSidedIdeal R}, IsNil I → IsNil (TwoSidedIdeal.matrix (Fin 2) I) h:∀ {R : Type u} [inst : Ring R] {I : TwoSidedIdeal R}, IsNil I → ∀ (n : Type) [inst_1 : Fintype n], IsNil (TwoSidedIdeal.matrix n I)R:Type uinst✝:Ring RI:TwoSidedIdeal RhI:IsNil I⊢ IsNil (TwoSidedIdeal.matrix (Fin 2) I) All goals completed! 🐙section MatrixTransfervariable {m : Type*} [Fintype m] [DecidableEq m] {n : Type*} [Fintype n] [DecidableEq n]

Nilness of the matrix ideal M_n(I) only depends on the cardinality of n.

@[category API, AMS 16] theorem isNil_matrix_of_equiv (e : m ≃ n) {I : TwoSidedIdeal R} (h : IsNil (TwoSidedIdeal.matrix m I)) : IsNil (TwoSidedIdeal.matrix n I) := R:Type u_1inst✝⁴:Ring Rm:Type u_2inst✝³:Fintype minst✝²:DecidableEq mn:Type u_3inst✝¹:Fintype ninst✝:DecidableEq ne:m ≃ nI:TwoSidedIdeal Rh:IsNil (TwoSidedIdeal.matrix m I)⊢ IsNil (TwoSidedIdeal.matrix n I) R:Type u_1inst✝⁴:Ring Rm:Type u_2inst✝³:Fintype minst✝²:DecidableEq mn:Type u_3inst✝¹:Fintype ninst✝:DecidableEq ne:m ≃ nI:TwoSidedIdeal Rh:IsNil (TwoSidedIdeal.matrix m I)M:Matrix n n RhM:M ∈ TwoSidedIdeal.matrix n I⊢ IsNilpotent M R:Type u_1inst✝⁴:Ring Rm:Type u_2inst✝³:Fintype minst✝²:DecidableEq mn:Type u_3inst✝¹:Fintype ninst✝:DecidableEq ne:m ≃ nI:TwoSidedIdeal Rh:IsNil (TwoSidedIdeal.matrix m I)M:Matrix n n RhM:M ∈ TwoSidedIdeal.matrix n IhM':(reindexRingEquiv R e).symm M ∈ TwoSidedIdeal.matrix m I⊢ IsNilpotent M All goals completed! 🐙

If M_{n ⊕ m}(I) is nil then so is M_n(I), using the block embedding.

R:Type u_1inst✝⁴:Ring Rm:Type u_2inst✝³:Fintype minst✝²:DecidableEq mn:Type u_3inst✝¹:Fintype ninst✝:DecidableEq nI:TwoSidedIdeal Rh:IsNil (TwoSidedIdeal.matrix (n ⊕ m) I)M:Matrix n n RhM:M ∈ TwoSidedIdeal.matrix n IhM':fromBlocks M 0 0 0 ∈ TwoSidedIdeal.matrix (n ⊕ m) Ik:ℕhk:fromBlocks M 0 0 0 ^ k = 0key:∀ (k : ℕ), fromBlocks M 0 0 0 ^ (k + 1) = fromBlocks (M ^ (k + 1)) 0 0 0this:0 = fromBlocks (M ^ (k + 1)) 0 0 0⊢ IsNilpotent M All goals completed! 🐙end MatrixTransfer

If M_2(J) is nil for every nil ideal J of every ring in universe u, then nilness of M_m(I) implies nilness of M_{2m}(I), using M_2(M_m(R)) ≃ M_{2m}(R).

@[category API, AMS 16] theorem isNil_matrix_prod_of_two_by_two_matrix (h : type_of% @KotherConjecture.variants.two_by_two_matrix.{u}) {R : Type u} [Ring R] {m : Type} [Fintype m] [DecidableEq m] {I : TwoSidedIdeal R} (hm : IsNil (TwoSidedIdeal.matrix m I)) : IsNil (TwoSidedIdeal.matrix (Fin 2 × m) I) := h:∀ {R : Type u} [inst : Ring R] {I : TwoSidedIdeal R}, IsNil I → IsNil (TwoSidedIdeal.matrix (Fin 2) I)R:Type uinst✝²:Ring Rm:Typeinst✝¹:Fintype minst✝:DecidableEq mI:TwoSidedIdeal Rhm:IsNil (TwoSidedIdeal.matrix m I)⊢ IsNil (TwoSidedIdeal.matrix (Fin 2 × m) I) h:∀ {R : Type u} [inst : Ring R] {I : TwoSidedIdeal R}, IsNil I → IsNil (TwoSidedIdeal.matrix (Fin 2) I)R:Type uinst✝²:Ring Rm:Typeinst✝¹:Fintype minst✝:DecidableEq mI:TwoSidedIdeal Rhm:IsNil (TwoSidedIdeal.matrix m I)M:Matrix (Fin 2 × m) (Fin 2 × m) RhM:M ∈ TwoSidedIdeal.matrix (Fin 2 × m) I⊢ IsNilpotent M h:∀ {R : Type u} [inst : Ring R] {I : TwoSidedIdeal R}, IsNil I → IsNil (TwoSidedIdeal.matrix (Fin 2) I)R:Type uinst✝²:Ring Rm:Typeinst✝¹:Fintype minst✝:DecidableEq mI:TwoSidedIdeal Rhm:IsNil (TwoSidedIdeal.matrix m I)M:Matrix (Fin 2 × m) (Fin 2 × m) RhM:M ∈ TwoSidedIdeal.matrix (Fin 2 × m) IhM':(compRingEquiv (Fin 2) m R).symm M ∈ TwoSidedIdeal.matrix (Fin 2) (TwoSidedIdeal.matrix m I)⊢ IsNilpotent M All goals completed! 🐙

The 2 × 2 matrix formulation of the Köthe conjecture (for rings in universe u) implies the general matrix formulation (Krempa): one first deduces the case of 2 ^ k × 2 ^ k matrices by induction and then embeds M_n(R) in a corner of M_{2 ^ k}(R).

h:∀ {R : Type u} [inst : Ring R] {I : TwoSidedIdeal R}, IsNil I → IsNil (TwoSidedIdeal.matrix (Fin 2) I)R:Type uinst✝¹:Ring RI:TwoSidedIdeal RhI:IsNil In:Type vinst✝:Fintype nkey:∀ (k : ℕ), IsNil (TwoSidedIdeal.matrix (Fin (k + 1) → Fin 2) I)hle:Fintype.card n ≤ 2 ^ (Fintype.card n + 1)e:n ⊕ Fin (2 ^ (Fintype.card n + 1) - Fintype.card n) ≃ (Fin (Fintype.card n + 1) → Fin 2)⊢ IsNil (TwoSidedIdeal.matrix n I) All goals completed! 🐙

The formulation of the Köthe conjecture in terms of the Köthe radical of matrix rings implies the general matrix formulation.

h:∀ {R : Type u} [inst : Ring R] (n : Type v) [inst_1 : Fintype n], TwoSidedIdeal.matrix n (Nil* R) = Nil* Matrix n n RR:Type uinst✝¹:Ring RI:TwoSidedIdeal RhI:IsNil In:Type vinst✝:Fintype n⊢ ↑(TwoSidedIdeal.matrix n I) ⊆ ↑(TwoSidedIdeal.matrix n (Nil* R)) All goals completed! 🐙

The general matrix formulation of the Köthe conjecture implies the formulation in terms of the Köthe radical of matrix rings.

h:∀ {R : Type u} [inst : Ring R] {I : TwoSidedIdeal R}, IsNil I → ∀ (n : Type v) [inst_1 : Fintype n], IsNil (TwoSidedIdeal.matrix n I)R:Type uinst✝¹:Ring Rn:Type vinst✝:Fintype nN:Matrix n n Ri:nj:nthis:Nonempty nJ:TwoSidedIdeal RhN:N ∈ TwoSidedIdeal.matrix n JhJ:(Nil* Matrix n n R) = TwoSidedIdeal.matrix n JhJnil:IsNil J⊢ (Nil* R).ringCon (N i j) (0 i j) All goals completed! 🐙

The Amitsur Conjecture: If J is a nil ideal in R, then J[x] is a nil ideal of the polynomial ring R[x]. This is known to be false, see Agata Smoktunowicz, Polynomial rings over nil rings need not be nil.

@[category research solved, AMS 16] theorem amitsur_conjecture : ¬ ∀ (R : Type) [Ring R] (J : TwoSidedIdeal R), IsNil J → IsNil (TwoSidedIdeal.map (Polynomial.C) J) := ⊢ ¬∀ (R : Type) [inst : Ring R] (J : TwoSidedIdeal R), IsNil J → IsNil (TwoSidedIdeal.map C J) All goals completed! 🐙end Koethe