Is every element of a -potent group of the form scalar times unipotent?

Yes, and that is essentially the definition: g is -potent with ≠ 0 iff ^-1g is unipotent. The subtlety is that the map g (g) is only proved to be a homomorphism *after* the theorem, so one cannot simply factor G as scalars times a unipotent group at the outset.

Unipotent Elements and the Lie–Kolchin–Suprunenko Theorem

Q: Why is UT_n(k) nilpotent of class exactly n-1?

Write elements as 1 + N with N strictly upper triangular. Commutators push the nonzero entries further from the diagonal: the ith term of the lower central series consists of matrices 1 + N with N supported on the diagonals at distance at least i from the main one. Those vanish once the distance exceeds n-1, and not before, as the matrices 1 + E_1,n witness. For n = 2 the group is abelian, class 1.

Q: How does this relate to the Sylow theorems?

Over F_q with q = p^f, (9.17) says unipotent means p-element and (9.20) says |UT_n(F_q)| is exactly the p-part of |GL_n(F_q)|. So (9.19)(3) specialises to Sylow's theorems for the prime p. Over an infinite field there are no Sylow theorems, yet the conjugacy statement survives — the theorem is genuinely stronger than what Sylow theory can give.

Q: Why does the theorem fail if I only assume the generators are unipotent?

Because unipotency is not preserved by products. SL_n(k) is generated by the unipotent elementary matrices and acts irreducibly, so no invariant flag exists. The hypothesis must be checked on the whole group; in practice one verifies it structurally, for instance by knowing that G is a p-group in characteristic p.

Executive Summary

A transformation $g$ of an $n$ -dimensional space is ** $λ$ -potent** if its characteristic polynomial is $(t - λ)^{n}$ , that is, $g = λ + g_{0}$ with $g_{0}$ nilpotent. For $λ = 1$ this is the classical notion of a unipotent transformation.

The theorem of Lie, Kolchin and Suprunenko says that a group of $λ$ -potent transformations can be triangularised all at once: in a suitable basis, $G \subseteq k^{\times} \cdot U T_{n} (k)$ . Nothing is assumed about $k$ — no algebraic closure, no characteristic restriction — and nothing about $G$ beyond $λ$ -potency of its elements.

$(t - λ)^{n}$ Characteristic polynomial

$q^{n (n - 1) / 2}$

| U T_{n} (𝔽_{q}) |

$\leq n - 1$ Nilpotency class

1948Kolchin

Overview

Unipotent elements are the part of a linear group that no amount of semisimplicity can absorb. In characteristic $p$ they are exactly the $p$ -elements $(9.17)$ , so a unipotent group is the infinite-field analogue of a $p$ -subgroup of $G L_{n} (𝔽_{q})$ ; in characteristic zero they are torsion-free, and a nontrivial unipotent group is automatically infinite.

The structure theorem is essentially Lie's theorem for solvable Lie algebras, transported to groups. Lie proved the triangularisation for connected solvable groups over $ℂ$ ; Kolchin removed all analytic input in 1948 in the paper that began the algebraic theory of matrix groups; Suprunenko gave the $λ$ -potent form. The proof here uses only Burnside's theorem and the Trace Lemma.

The immediate payoffs are that $U T_{n} (k)$ is a maximal unipotent subgroup of $G L_{n} (k)$ and all such are conjugate — a Sylow theorem valid for infinite groups — and that every $λ$ -potent group is nilpotent of class less than $n$ .

Learning Objectives

State Definition $(9.16)$ and show that a $λ$ -potent $g$ with $λ \neq 0$ is invertible, with $g^{- 1}$ given by a finite geometric series.
Prove $(9.17)$ : over a field of characteristic $p$ , unipotent $\Leftrightarrow$ $p$ -element.
State $(9.18)$ correctly, including the hypothesis that $G$ acts $λ$ -potently on the module in question.
Follow the three-step proof: reduction to one-dimensional constituents, the algebraically closed case, and the descent argument.
Deduce $(9.19)$ : $Λ$ is a group, $λ$ is a homomorphism, and maximal unipotent subgroups are conjugate to $U T_{n} (k)$ .
Verify $(9.20)$ that $U T_{n} (𝔽_{q})$ is a Sylow $p$ -subgroup, and $(9.21)$ that nilpotency class is at most $n - 1$ .

Definitions

Definition(9.16)λ-potent and unipotent

Let $k$ be a field, $V$ an $n$ -dimensional $k$ -vector space and $λ \in k$ . A linear transformation $g \in {End}_{k} V$ is ** $λ$ -potent** if $g = λ + g_{0}$ with $g_{0}$ nilpotent; equivalently, if its characteristic polynomial is $(t - λ)^{n}$ .

* $0$ -potent* means nilpotent, and * $1$ -potent* is called unipotent. For a subset $Λ \subseteq k^{\times}$ , a group $G \subseteq G L (V)$ is ** $Λ$ -potent** (with respect to a surjection $λ : G \to Λ$ ) if every $g \in G$ is $λ (g)$ -potent. When $Λ = {1}$ , $G$ is a unipotent group.

Remark—λ-potent elements are invertible for λ ≠ 0

If $g = λ + g_{0}$ with $λ \neq 0$ and $g_{0}^{m} = 0$ , then $g = λ (1 + λ^{- 1} g_{0})$ and the geometric series terminates:

g^{- 1} = λ^{- 1} (1 - λ^{- 1} g_{0} + λ^{- 2} g_{0}^{2} - \dots + (- 1)^{m - 1} λ^{- (m - 1)} g_{0}^{m - 1}) .

So $g \in G L (V)$ , and $g$ is $λ$ -potent precisely when $λ^{- 1} g$ is unipotent. The unipotent case therefore carries all the content.

$U T_{n} (k)$: Upper triangular matrices with all diagonal entries $1$ . A group under multiplication, and manifestly unipotent.
$k^{\times} \cdot U T_{n} (k)$: Invertible upper triangular matrices with constant diagonal. A $k^{\times}$ -potent group; every $λ$ -potent group embeds in a conjugate of one of its subgroups.
Lower central series: $G^{(1)} = G$ , $G^{(i + 1)} = [G, G^{(i)}]$ . $G$ is nilpotent of class $r$ if $G^{(r + 1)} = 1 \neq G^{(r)}$ .
$p$ -element: An element whose order is a finite power of $p$ . In characteristic $p$ these are exactly the unipotent elements of $G L (V)$ .
Borel subgroup: The full upper triangular group $B_{n} (k)$ ; $U T_{n} (k)$ is its unipotent radical, and $B_{n} / U T_{n}$ is the diagonal torus.

The groups UT(n,k) and k-times-UT(n,k) are defined only up to conjugacy in GL(V), since they depend on a chosen full flag of subspaces.

Core Concepts

Unipotent means $p$ -element, and only in characteristic $p$

In characteristic $p$ the two notions coincide $(9.17)$ , because Frobenius makes $(g - 1)^{p^{s}} = g^{p^{s}} - 1$ inside the commutative algebra $k [g]$ . In characteristic zero the picture is opposite: a nontrivial unipotent $g = 1 + g_{0}$ satisfies $g^{m} = 1 + m g_{0} + (\binom{m}{2}) g_{0}^{2} + \dots \neq 1$ for $m \neq 0$ , so unipotent groups are torsion-free.

char k = p > 0 : g^{p^{s}} = 1 ⟺ (g - 1)^{p^{s}} = 0; char k = 0 : g unipotent, g \neq 1 ⟹ ord (g) = \infty .

(9.17a)

Why the theorem is a statement about constituents

Triangularising a group is the same as producing a full flag of invariant subspaces, that is, a composition series of $V$ with one-dimensional factors. So part (2) of the theorem is a formal consequence of part (1): once all composition factors are lines, an adapted basis makes every $g$ upper triangular, and $λ$ -potency forces its diagonal entries all to equal $λ (g)$ .

every irreducible constituent is a line

V

has a full

G

-invariant flag

G

is upper triangular in an adapted basisdiagonal entries all equal

λ (g)

G \subseteq k^{\times} U T_{n} (k)

Traces do the work again

On a module where $g$ acts $λ (g)$ -potently, $tr (g) = m λ (g)$ where $m$ is the dimension. Restricting to the elements of determinant $1$ forces $λ (g)^{m} = 1$ , so only $m$ traces occur and the Trace Lemma $(9.3)$ bounds the group. Finiteness then collides with $λ$ -potency and collapses the module to a line.

Key Results

Proposition(9.17)Unipotent = p-element in characteristic p

Let $k$ be a field of characteristic $p > 0$ and $V$ finite-dimensional over $k$ . Then $g \in G L (V)$ is unipotent iff $g$ is a $p$ -element, i.e. $ord (g) = p^{s}$ for some $s \geq 0$ . Consequently a linear group $G \subseteq G L (V)$ is unipotent iff it is a $p$ -group.

Proof

If $g^{p^{s}} = 1$ , then in the commutative ring $k [g]$ of characteristic $p$ we have $(g - 1)^{p^{s}} = g^{p^{s}} - 1^{p^{s}} = 0$ , so $g - 1$ is nilpotent and $g$ is unipotent.

Conversely let $g = 1 + g_{0}$ with $g_{0}^{m} = 0$ . Choose $s$ with $p^{s} \geq m$ . Then $g^{p^{s}} = (1 + g_{0})^{p^{s}} = 1 + g_{0}^{p^{s}} = 1$ , so $g$ has order a power of $p$ .

Theorem(9.18)Lie–Kolchin–Suprunenko

Let $k$ be a field, $V \neq 0$ with $\dim_{k} V = n$ , and $G \subseteq G L (V)$ a $λ$ -potent group with respect to a surjection $λ : G \to Λ \subseteq k^{\times}$ . Then:

if $M$ is a finite-dimensional irreducible $k G$ -module on which each $g \in G$ acts as a $λ (g)$ -potent operator — for example any composition factor of $V$ — then $\dim_{k} M = 1$ and $g$ acts on $M$ as the scalar $λ (g)$ ;
with respect to a suitable $k$ -basis of $V$ , $G \subseteq k^{\times} \cdot U T_{n} (k)$ .

In the unipotent case $Λ = {1}$ this is Kolchin's theorem: a unipotent group is conjugate into $U T_{n} (k)$ .

Proof

Step 1: (1) implies (2)

Take a $k G$ -composition series $V = V_{0} \supset V_{1} \supset \dots \supset V_{t} = 0$ . Every factor $V_{i} / V_{i + 1}$ is an irreducible $k G$ -module on which $G$ acts $λ$ -potently, because the characteristic polynomial of $g$ on a subquotient divides $(t - λ (g))^{n}$ . By (1) each factor is one-dimensional, so $t = n$ and a basis adapted to the flag makes every $g \in G$ upper triangular. Since $g$ is $λ (g)$ -potent, its eigenvalues — the diagonal entries — all equal $λ (g)$ . Hence $G \subseteq k^{\times} \cdot U T_{n} (k)$ .

Step 2: proof of (1) when $k$ is algebraically closed

Let $M$ be as in (1), $m = \dim_{k} M$ . Replacing $G$ by its image in $G L (M)$ changes nothing: a $λ (g)$ -potent operator determines $λ (g)$ as its unique eigenvalue, so $λ$ descends. Enlarge $G$ to $\tilde{G} = k^{\times} \cdot G$ , still $λ$ -potent with $λ (c g) = c λ (g)$ , and note $M$ is still simple and ${Span}_{k} (\tilde{G}) = {Span}_{k} (G) = {End}_{k} M$ by Burnside's Theorem $(7.3)$ .

Let ${\tilde{G}}_{0} = {g \in \tilde{G} : \det g = 1}$ . Because $k$ is algebraically closed, every $g \in \tilde{G}$ can be scaled into ${\tilde{G}}_{0}$ , so $\tilde{G} = k^{\times} {\tilde{G}}_{0}$ and hence ${Span}_{k} ({\tilde{G}}_{0}) = {End}_{k} M$ : the module $M$ is absolutely irreducible over the semigroup ${\tilde{G}}_{0}$ . For $g \in {\tilde{G}}_{0}$ we have $λ (g)^{m} = \det g = 1$ and $tr (g) = m λ (g)$ , so at most $m$ traces occur, and the Trace Lemma $(9.3)$ gives $| {\tilde{G}}_{0} | \leq m^{m^{2}} < \infty$ .

Now let $g \in {\tilde{G}}_{0}$ have finite order $r$ prime to $p = char k$ (every order if $p = 0$ ). Then $g$ satisfies both $t^{r} - 1$ , which is separable, and $(t - λ (g))^{m}$ ; their greatest common divisor is $t - λ (g)$ , so $g = λ (g) \cdot 1$ is a scalar.

If $p = 0$ this makes ${\tilde{G}}_{0}$ — and hence $\tilde{G}$ — consist of scalars, so ${End}_{k} M = {Span}_{k} (\tilde{G}) = k$ and $m = 1$ . If $p > 0$ , let $Z = {\tilde{G}}_{0} \cap k^{\times}$ , a finite subgroup of $k^{\times}$ and therefore of order prime to $p$ . Every element of ${\tilde{G}}_{0} / Z$ is a $p$ -element by the previous paragraph, so ${\tilde{G}}_{0} / Z$ is a finite $p$ -group and ${\tilde{G}}_{0} = Z P$ for a Sylow $p$ -subgroup $P$ , with $Z$ central. Then ${Span}_{k} (P) = {Span}_{k} ({\tilde{G}}_{0}) = {End}_{k} M$ , so $M$ is absolutely irreducible over $P$ ; each element of $P$ is $λ$ -potent with $λ^{p^{s}} = 1$ , hence $λ = 1$ , hence unipotent with $tr = m \cdot 1$ . Only one trace occurs, so the Trace Lemma gives $| P | \leq 1$ . Thus ${\tilde{G}}_{0} = Z$ is scalar and again $m = 1$ .

Step 3: descent to an arbitrary field

Let $M$ be as in (1) over an arbitrary $k$ , with $G$ acting faithfully, and let $E = \bar{k}$ . Steps 1 and 2 applied over $E$ give an $E$ -basis of $M \otimes_{k} E$ in which $G \subseteq E^{\times} \cdot U T_{m} (E)$ ; so each $g - λ (g)$ is strictly upper triangular and any product of $m$ of them vanishes. These operators are defined over $k$ , so the vanishing holds over $k$ .

Choose $r \geq 1$ minimal with $(g_{1} - λ (g_{1})) \dots (g_{r} - λ (g_{r})) = 0$ for all $g_{1}, \dots, g_{r} \in G$ . If $r = 1$ then $G$ consists of scalars, every subspace is a submodule, and simplicity gives $m = 1$ . If $r \geq 2$ , minimality provides $g_{2}, \dots, g_{r} \in G$ and $v \in M$ with

w = (g_{2} - λ (g_{2})) \dots (g_{r} - λ (g_{r})) v \neq 0,

while $(g - λ (g)) w = 0$ for every $g \in G$ by minimality of $r$ . Thus $g w = λ (g) w$ , so $k w$ is a nonzero $k G$ -submodule of $M$ , and simplicity forces $M = k w$ and $m = 1$ . $■$

Corollary(9.19)Consequences

$Λ$ is a subgroup of $k^{\times}$ and $λ : G \to Λ$ is a group homomorphism.
Any unipotent subgroup of $G L_{n} (k)$ is conjugate in $G L_{n} (k)$ to a subgroup of $U T_{n} (k)$ .
$U T_{n} (k)$ is a maximal unipotent subgroup of $G L_{n} (k)$ , and every maximal unipotent subgroup of $G L_{n} (k)$ is conjugate to $U T_{n} (k)$ .

For (1) and (2), read off the conclusion of $(9.18) (2)$ : for $g$ in $k^{\times} U T_{n} (k)$ the scalar $λ (g)$ is the common diagonal entry, and $g \mapsto λ (g)$ is then visibly multiplicative.

Proof

For (3): a union of a chain of unipotent subgroups is unipotent, so Zorn's Lemma provides maximal unipotent subgroups, and every unipotent subgroup lies in one. Let $G$ be maximal unipotent. By (2) there is $ρ \in G L_{n} (k)$ with $ρ G ρ^{- 1} \subseteq U T_{n} (k)$ . Since $U T_{n} (k)$ is unipotent and $ρ G ρ^{- 1}$ is again maximal unipotent, $ρ G ρ^{- 1} = U T_{n} (k)$ . This shows simultaneously that every maximal unipotent subgroup is conjugate to $U T_{n} (k)$ and — since a conjugate of a maximal unipotent subgroup is maximal unipotent — that $U T_{n} (k)$ is itself maximal unipotent.

Remark(9.20)Sylow theory for

G L_{n} (𝔽_{q})

For $q = p^{f}$ ,

| G L_{n} (𝔽_{q}) | = (q^{n} - 1) (q^{n} - q) (q^{n} - q^{2}) \dots (q^{n} - q^{n - 1}), | U T_{n} (𝔽_{q}) | = q^{n (n - 1) / 2} .

Writing $q^{n} - q^{i} = q^{i} (q^{n - i} - 1)$ shows the $p$ -part of the order is $q^{0 + 1 + \dots + (n - 1)} = q^{n (n - 1) / 2}$ . So $U T_{n} (𝔽_{q})$ is a Sylow $p$ -subgroup, and $(9.19) (3)$ is Sylow's conjugacy theorem for this one prime — but valid over every field, where no Sylow theory exists.

Corollary(9.21)Nilpotency

For $n \geq 2$ , every $λ$ -potent subgroup of $G L_{n} (k)$ is nilpotent of class at most $n - 1$ .

Indeed $U T_{n} (k)$ is nilpotent of class exactly $n - 1$ , and adjoining central scalars does not change the lower central series, since $[k^{\times} U T_{n}, k^{\times} U T_{n}] = [U T_{n}, U T_{n}]$ . Now apply $(9.18) (2)$ . In characteristic $p$ this generalises the fact that finite $p$ -groups are nilpotent, via $(9.17)$ .

Proof Techniques and Method

How these proofs work, and which move to reuse.

Move 1

Triangularise via constituents

Simultaneous triangularisation is never proved directly. Prove instead that all irreducible constituents are lines; the flag and the basis follow formally.

Move 2

Normalise the determinant

Over an algebraically closed field, scaling by $m$ th roots moves any group into the determinant-one subgroup without changing the spanned algebra. This is what makes the trace set finite.

Move 3

Descend by minimal vanishing product

To get from $\bar{k}$ back to $k$ , take the least $r$ with all products of $r$ operators $g - λ (g)$ vanishing. The vectors killed at level $r - 1$ span a line, and simplicity finishes.

Move 3 is the reusable gem. It replaces a Galois descent argument by an entirely elementary one: the identity *products of $m$ shifted operators vanish* is defined over $k$ even though it was verified over $\bar{k}$ , so the minimal-length argument can be run over $k$ .

Worked Example

$U T_{3} (𝔽_{2})$ as a Sylow subgroup

Take $q = p = 2$ , $n = 3$ . Then

| G L_{3} (𝔽_{2}) | = (8 - 1) (8 - 2) (8 - 4) = 7 \cdot 6 \cdot 4 = 168 = 2^{3} \cdot 3 \cdot 7, | U T_{3} (𝔽_{2}) | = 2^{3 \cdot 2 / 2} = 2^{3} = 8 .

(E.1)

So $U T_{3} (𝔽_{2})$ is a Sylow $2$ -subgroup of $G L_{3} (𝔽_{2})$ , as $(9.20)$ predicts. Which group of order $8$ is it? Put $N = E_{12} + E_{23}$ , so $N^{2} = E_{13}$ and $N^{3} = 0$ , and let $g = I + N$ . Then

g^{2} = I + 2 N + N^{2} = I + E_{13} \neq I, g^{4} = (I + E_{13})^{2} = I + 2 E_{13} = I,

(E.2)

in characteristic $2$ ; so $g$ has order $4$ .

The group is nonabelian of order $8$ with more than one involution — $I + E_{12}$ , $I + E_{23}$ and $I + E_{13}$ all square to the identity — so $U T_{3} (𝔽_{2}) ≅ D_{4}$ , the dihedral group of order $8$ . Its nilpotency class is $2 = n - 1$ , matching $(9.21)$ exactly.

A $k^{\times}$ -potent group and its class

Let $G = {(\begin{matrix} a & b \\ 0 & a \end{matrix}) : a \in k^{\times}, b \in k} \subseteq G L_{2} (k)$ . Every element has characteristic polynomial $(t - a)^{2}$ , so $G$ is $λ$ -potent with $λ (\begin{matrix} a & b \\ 0 & a \end{matrix}) = a$ and $Λ = k^{\times}$ .

Here $(9.19) (1)$ is checked directly: $λ$ is multiplicative because the product of two such matrices has diagonal $a a'$ . And $(9.21)$ predicts nilpotency class at most $n - 1 = 1$ , i.e. abelian — which is correct, since $(\begin{matrix} a & b \\ 0 & a \end{matrix}) (\begin{matrix} a' & b' \\ 0 & a' \end{matrix}) = (\begin{matrix} a a' & a b' + a' b \\ 0 & a a' \end{matrix})$ is symmetric in the two factors.

A group that is not $λ$ -potent

In $G L_{2} (ℚ)$ the elements $A = (\begin{matrix} 1 & 1 \\ 0 & 1 \end{matrix})$ and $B = (\begin{matrix} 1 & 0 \\ 1 & 1 \end{matrix})$ are each unipotent, but

B A = (\begin{matrix} 1 & 1 \\ 1 & 2 \end{matrix}), tr (B A) = 3, \det (B A) = 1,

(E.3)

characteristic polynomial $t^{2} - 3 t + 1$ , with distinct real roots $(3 \pm \sqrt{5}) / 2$ .

So $B A$ is not $λ$ -potent and $⟨ A, B ⟩ = S L_{2} (ℤ)$ is not a unipotent group — as it must not be, since it acts irreducibly on $ℚ^{2}$ and $(9.18) (1)$ would otherwise force $\dim = 1$ . Generated by unipotent elements is much weaker than unipotent.

Comparison and Classification

Unipotent elements across characteristics
Property	$char k = 0$	$char k = p > 0$
$g$ unipotent, $g \neq 1$	infinite order	order a power of $p$
Unipotent groups	torsion-free	exactly the $p$ -groups $(9.17)$
$U T_{n} (k)$ finite?	never for $n \geq 2$	iff $k$ is finite
Maximal unipotent subgroups	all conjugate to $U T_{n} (k)$	all conjugate to $U T_{n} (k)$
Sylow interpretation	none	$U T_{n} (𝔽_{q})$ is a Sylow $p$ -subgroup
Nilpotency class	$\leq n - 1$	$\leq n - 1$
Divisible?	yes, and torsion-free (Exercise 9.4)	no — every element has $p$ -power order

Which groups are λ-potent, unipotent, or neither
	unipotent	$λ$ -potent	completely reducible
$U T_{n} (k)$ , $n \geq 2$	yes	yes	no
$k^{\times} \cdot U T_{n} (k)$	no	yes	no
Scalars $k^{\times} \cdot I$	no	yes	yes
Full upper triangular $B_{n} (k)$	no	no	no
$S L_{2} (ℤ)$ in $G L_{2} (ℚ)$	no	no	yes
Diagonal matrices, $n \geq 2$	no	no	yes

Which groups are λ-potent, unipotent, or neither

Relationship Map

Unipotent groups sit inside the Borel and are detected by the radical of the spanned algebra.

$G L_{n} (k)$ all invertible transformations

Borel $B_{n} (k)$ upper triangular; solvable

$k^{\times} \cdot U T_{n} (k)$ constant diagonal;

λ

-potent, nilpotent of class

\leq n - 1

$U T_{n} (k)$ unipotent radical of the Borel; maximal unipotent; Sylow

p

over

𝔽_{q}

Any unipotent subgroupconjugate into

U T_{n} (k)

by Kolchin

(9.19) (2)

$g$ unipotent — $(g - 1)$ nilpotent
- equivalently
  - characteristic polynomial $(t - 1)^{n}$
  - $g$ is a $p$ -element, if $char k = p$
  - $g$ fixes a full flag with trivial action on the factors
- implies
  - $g - 1 \in rad {Span}_{k} (G)$ when $G$ is unipotent
  - $tr (g) = n \cdot 1_{k}$
  - $\det g = 1$
- does not imply
  - that $⟨ g, h ⟩$ is unipotent for another unipotent $h$
  - that $g$ has finite order — false in characteristic $0$

Applications and Industry Use

Applications here means where this structure is used — inside mathematics and in the engineering and computing disciplines that consume it.

Algebraic groups

Borel and parabolic structure

Kolchin's theorem is the group-theoretic input to the Lie–Kolchin theorem for connected solvable algebraic groups, and hence to the definition of Borel subgroups, the unipotent radical and reductive groups. Kolchin's 1948 paper is one of the founding documents of the subject.

Differential Galois theory

Picard–Vessiot groups

Kolchin developed this material to study differential equations. A Picard–Vessiot group that is unipotent corresponds to an equation solvable by iterated integration, and triangularisation of the group is triangularisation of the system.

Finite group theory

Sylow subgroups of $G L_{n} (q)$

$(9.20)$ identifies the Sylow $p$ -subgroup of $G L_{n} (𝔽_{q})$ explicitly as $U T_{n} (𝔽_{q})$ , which is the starting point for the modular representation theory of groups of Lie type and for many computations in the classification programme.

Cryptography and computing

Unitriangular matrix groups

$U T_{n} (𝔽_{q})$ is a standard source of finite nilpotent $p$ -groups of controlled class, used in the construction of test cases for nilpotent quotient algorithms, in polycyclic presentations, and in proposals for non-abelian key exchange.

Within this collection the theorem's role is to supply the obstruction: the largest normal unipotent subgroup of a linear group is precisely what stops it from being completely reducible, and $(9.18) (1)$ is the tool that identifies it.

Standards and Notation

Standards here covers notation, symbol and markup standards, and reference implementations, rather than material or design codes.

Preferred notation

U T_{n} (k)

for the unitriangular group (Lam)

Common variants

U_{n} (k)

N_{n} (k)

𝒰_{n}

; in algebraic groups

U

for the unipotent radical of a Borel

B

Element terminologyunipotent is universal; *

λ

-potent* is Lam's; Suprunenko writes quasi-unipotent for related notions — check definitions before quoting

Lie sidea nilpotent endomorphism

g_{0}

corresponds to unipotent

\exp (g_{0})

when

char k = 0

char k > n

GAPSylowSubgroup(GL(n,q), p) returns a conjugate of the unitriangular group

MarkupPresentation MathML per ISO/IEC 40314; matrix and operator symbols per ISO 80000-2

Failure Modes and Common Mistakes

Do not assume $λ$ is a homomorphism before proving the theorem — that is a consequence $(9.19) (1)$ , not part of the definition.
Do not expect $U T_{n} (k)$ to be the unique maximal unipotent subgroup; there are many, all conjugate, one for each full flag.
Do not transfer the Sylow reading to characteristic $0$ : there $U T_{n} (k)$ is infinite and torsion-free, and no Sylow theory is in play.
Do not confuse nilpotent element ( $g^{m} = 0$ , hence not invertible) with unipotent element ( $g - 1$ nilpotent, hence invertible).
Do not conclude from $(9.19) (3)$ that all maximal unipotent subgroups of an arbitrary linear group $G$ are conjugate; the statement is about $G L_{n} (k)$ .

Historical Notes and Lessons Learned

1876LieLie's theorem: a solvable Lie algebra of operators over C is simultaneously triangularisable, with one-dimensional common eigenspaces. The infinitesimal ancestor of everything on this page.
1937Zassenhaus and othersTriangularisability of nil and unipotent groups of matrices is studied over general fields; the connection with p-groups in characteristic p becomes explicit.
1948KolchinKolchin proves that a unipotent matrix group over any field is conjugate into the unitriangular group, in a paper motivated by differential algebra and widely regarded as founding the algebraic theory of matrix groups.
1950sSuprunenkoThe lambda-potent generalisation and a systematic study of solvable and locally nilpotent matrix groups over arbitrary fields.
1956BorelThe Lie-Kolchin theorem for connected solvable algebraic groups becomes the foundation of Borel subgroup theory, and unipotent radicals enter the definition of reductive groups.
1975HumphreysThe textbook treatment in Linear Algebraic Groups fixes the modern terminology: unipotent radical, reductive quotient, Borel and parabolic subgroups.

The lesson worth carrying: the analytic hypotheses of Lie's theorem — connectedness, the complex numbers, a Lie algebra — were all removable. What remained after Kolchin was a purely algebraic statement provable with traces, and it is stronger, because it holds in characteristic $p$ where exponentials do not exist.

Quick Reference

$λ$ -potentcharacteristic polynomial

(t - λ)^{n}

;

g = λ +

nilpotent

Unipotent

λ = 1

; equivalently a

p

-element when

char k = p

$(9.18)$

λ

-potent group

\subseteq k^{\times} U T_{n} (k)

in a suitable basis

$(9.19) (1)$

Λ \leq k^{\times}

and

λ

is a homomorphism

$(9.19) (3)$ maximal unipotent subgroups of

G L_{n} (k)

are the conjugates of

U T_{n} (k)

$(9.20)$

| U T_{n} (𝔽_{q}) | = q^{n (n - 1) / 2}

, the

p

-part of

| G L_{n} (𝔽_{q}) |

$(9.21)$ class

\leq n - 1

;

U T_{n} (k)

has class exactly

n - 1

Characteristic 0unipotent groups are torsion-free and divisible

Numbered results used on this page
Reference	Statement
$(9.16)$	definition of $λ$ -potent and unipotent
$(9.17)$	unipotent $\Leftrightarrow$ $p$ -element in characteristic $p$
$(9.18)$	Lie–Kolchin–Suprunenko triangularisation
$(9.19)$	$λ$ is a homomorphism; conjugacy of maximal unipotent subgroups
$(9.20)$	orders of $G L_{n} (𝔽_{q})$ and $U T_{n} (𝔽_{q})$
$(9.21)$	nilpotency class at most $n - 1$
$(9.3)$ , $(7.3)$	Trace Lemma and Burnside's theorem — the tools

Frequently Asked Questions

Does the theorem require the field to be algebraically closed?

No, and that is the point of Step 3. The algebraically closed case is proved first because Burnside's theorem and the Trace Lemma need it; the descent then uses only the elementary observation that a minimal-length vanishing product of the operators $g - λ (g)$ produces a common eigenvector defined over $k$ . Contrast the Lie–Kolchin theorem for connected solvable algebraic groups, which genuinely needs algebraic closure.

Why is

U T_{n} (k)

nilpotent of class exactly

n - 1

Write elements as $1 + N$ with $N$ strictly upper triangular. Commutators push the nonzero entries further from the diagonal: the $i$ th term of the lower central series consists of matrices $1 + N$ with $N$ supported on the diagonals at distance at least $i$ from the main one. Those vanish once the distance exceeds $n - 1$ , and not before, as the matrices $1 + E_{1, n}$ witness. For $n = 2$ the group is abelian, class $1$ .

How does this relate to the Sylow theorems?

Over $𝔽_{q}$ with $q = p^{f}$ , $(9.17)$ says unipotent means $p$ -element and $(9.20)$ says $| U T_{n} (𝔽_{q}) |$ is exactly the $p$ -part of $| G L_{n} (𝔽_{q}) |$ . So $(9.19) (3)$ specialises to Sylow's theorems for the prime $p$ . Over an infinite field there are no Sylow theorems, yet the conjugacy statement survives — the theorem is genuinely stronger than what Sylow theory can give.

Is every element of a

λ

-potent group of the form scalar times unipotent?

Yes, and that is essentially the definition: $g$ is $λ$ -potent with $λ \neq 0$ iff $λ^{- 1} g$ is unipotent. The subtlety is that the map $g \mapsto λ (g)$ is only proved to be a homomorphism after the theorem, so one cannot simply factor $G$ as scalars times a unipotent group at the outset.

What happens for a group of nilpotent — that is

0

-potent — elements?

Nothing, because such elements are not invertible and cannot form a group inside $G L (V)$ . The corresponding statement for multiplicatively closed sets of nilpotent matrices is Levitzki's theorem: a semigroup of nilpotent matrices is simultaneously strictly triangularisable. It is proved by the same trace argument.

Why does the theorem fail if I only assume the generators are unipotent?

Because unipotency is not preserved by products. $S L_{n} (k)$ is generated by the unipotent elementary matrices and acts irreducibly, so no invariant flag exists. The hypothesis must be checked on the whole group; in practice one verifies it structurally, for instance by knowing that $G$ is a $p$ -group in characteristic $p$ .

References

T. Y. Lam, A First Course in Noncommutative Rings, Graduate Texts in Mathematics 131, Springer-Verlag, 1991, §9 (pp. 158–161).
E. R. Kolchin, “Algebraic matric groups and the Picard-Vessiot theory of homogeneous linear ordinary differential equations”, Annals of Mathematics 49 (1948), 1–42.
J. E. Humphreys, Linear Algebraic Groups, Graduate Texts in Mathematics 21, Springer-Verlag, 1975, §17.
D. A. Suprunenko, Matrix Groups, Translations of Mathematical Monographs 45, American Mathematical Society, 1976.
B. A. F. Wehrfritz, Infinite Linear Groups, Ergebnisse der Mathematik 76, Springer-Verlag, 1973, Chapter 1.
H. Radjavi and P. Rosenthal, Simultaneous Triangularization, Universitext, Springer-Verlag, 2000.

AI Suggested Questions

Prove Levitzki's theorem on semigroups of nilpotent matrices using the same trace argument.
Show that a unipotent group in characteristic zero is torsion-free and divisible, as Lam's Exercise 9.4 asks.
How does the Lie-Kolchin theorem for connected solvable algebraic groups differ in proof from Kolchin's theorem?
Compute the conjugacy classes of UT(4,q) and explain why the class number is a polynomial in q.
What is the largest nilpotency class achievable by a subgroup of GL(n,k), and is it attained by unitriangular groups?
Describe the irreducible complex representations of the discrete Heisenberg group and reconcile them with (9.18)(1).
How are unipotent subgroups used in the Picard-Vessiot theory of linear differential equations?

Unipotent Elements and Lie–Kolchin

Executive Summary

Overview

Learning Objectives

Definitions

Core Concepts

Unipotent means $p$ -element, and only in characteristic $p$

Why the theorem is a statement about constituents

Traces do the work again

Key Results

Step 1: (1) implies (2)

Step 2: proof of (1) when $k$ is algebraically closed

Step 3: descent to an arbitrary field

Proof Techniques and Method

Triangularise via constituents

Normalise the determinant

Descend by minimal vanishing product

Worked Example

$U T_{3} (𝔽_{2})$ as a Sylow subgroup

A $k^{\times}$ -potent group and its class

A group that is not $λ$ -potent

Comparison and Classification

Relationship Map

Applications and Industry Use

Borel and parabolic structure

Picard–Vessiot groups

Sylow subgroups of $G L_{n} (q)$

Unitriangular matrix groups

Standards and Notation

Failure Modes and Common Mistakes

Historical Notes and Lessons Learned

Quick Reference

Frequently Asked Questions

References

AI Suggested Questions

Continue learning

01Executive Summary

02Overview

03Learning Objectives

04Definitions

05Core Concepts

Unipotent means p-element, and only in characteristic p

Why the theorem is a statement about constituents

Traces do the work again

06Key Results

Step 1: (1) implies (2)

Step 2: proof of (1) when k is algebraically closed

Step 3: descent to an arbitrary field

07Proof Techniques and Method

Triangularise via constituents

Normalise the determinant

Descend by minimal vanishing product

08Worked Example

UT3(𝔽2) as a Sylow subgroup

A k×-potent group and its class

A group that is not λ-potent

09Comparison and Classification

10Relationship Map

11Applications and Industry Use

Borel and parabolic structure

Picard–Vessiot groups

Sylow subgroups of GLn(q)

Unitriangular matrix groups

12Standards and Notation

13Failure Modes and Common Mistakes

14Historical Notes and Lessons Learned

15Quick Reference

16Frequently Asked Questions

17Related KEVOS Topics

18References

19AI Suggested Questions

Continue learning

Executive Summary

Overview

Learning Objectives

Definitions

Core Concepts

Unipotent means $p$ -element, and only in characteristic $p$

Key Results

Step 2: proof of (1) when $k$ is algebraically closed

Proof Techniques and Method

Worked Example

$U T_{3} (𝔽_{2})$ as a Sylow subgroup

A $k^{\times}$ -potent group and its class

A group that is not $λ$ -potent

Comparison and Classification

Relationship Map

Applications and Industry Use

Sylow subgroups of $G L_{n} (q)$

Standards and Notation

Failure Modes and Common Mistakes

Historical Notes and Lessons Learned

Quick Reference

Frequently Asked Questions

Related KEVOS Topics

References

AI Suggested Questions