Differential Polynomial Rings and the Weyl Algebra

Q: Why does the differential construction need no injectivity hypothesis, unlike the skew one?

Because the correction term (a) has strictly smaller degree in x than ax. The leading coefficient of a product in k[x;] is the plain product of the leading coefficients, so degrees add as soon as k is a domain. In k[x;] the leading coefficient is a_m^m(b_n), which can vanish if has a kernel.

Q: Is A_1(k) a principal left ideal domain?

No. Left division in k[x;] requires the divisor to have an invertible leading coefficient, and over k[y] most elements do not. Passing to k(y)[x;d/dy] does give a principal left ideal domain. For A_1 itself, Stafford proved that every left ideal is generated by two elements, and that non-free projective left ideals exist — so principality genuinely fails.

Q: What is the relationship between k[x;] and k[x;]?

Neither contains the other, and both are Ore extensions k[x;,] with one ingredient trivial. A single algebra can admit both descriptions: the enveloping algebra of the two-dimensional non-abelian Lie algebra is k[x][y;] with (x) = x-1 and also k[y][x;] with = y,d/dy.

Q: Why is A_1(k) called an algebra when k may be noncommutative?

Strictly it should not be. The construction A_1(k_0) = k_0[y][x;d/dy] makes sense for any ring k_0, and k_0 need not be central in the result unless k_0 is commutative. The name is entrenched, and Lam flags the abuse explicitly.

Q: Does the differential polynomial ring inherit chain conditions from k?

Yes, on the appropriate side: if k is left noetherian then so is k[x;], by the same leading-coefficient argument that proves the Hilbert basis theorem. Note that this is cleaner than the skew case, where the corresponding statement needs to be an automorphism.

Q: What replaces the differential-operator model in characteristic p?

The action of A_1(k) on k[y] is no longer faithful, because D^p = 0. The correct objects are modules over the divided-power or crystalline differential operator rings, and A_1(k) itself becomes an Azumaya algebra of rank p^2 over the central subring k[x^p, y^p].

Executive Summary

There are exactly two elementary ways to stop the variable from commuting with the coefficients. One multiplies coefficients by an endomorphism — that is Hilbert's twist. The other adds a correction term: $x a = a x + δ (a)$ . Demanding that this extend to an associative multiplication does not merely permit the Leibniz rule, it forces it, and $δ$ must be a derivation of $k$ .

The resulting ring $k [x; δ]$ is the differential polynomial ring. When $δ$ is inner nothing is gained — a change of variable turns $k [x; δ]$ back into $k [x]$ . The construction only bites for outer derivations, and the cleanest outer derivation available is $d / d y$ on $k_{0} [y]$ . That single choice produces the first Weyl algebra $A_{1} (k_{0})$ , the algebra with $x y - y x = 1$ : a noncommutative noetherian domain, simple in characteristic zero, and the standard model for algebras of differential operators.

$x a = a x + δ (a)$ Defining rule

$δ (a b) = a δ (b) + δ (a) b$ Forced identity

$x y - y x = 1$ Weyl relation

${y^{j} x^{i}}$ A

k

-basis of

A_{1}

Overview

Fix a ring $k$ and take the free left $k$ -module on $1, x, x^{2}, \dots$ again. This time, when $x$ moves past a coefficient $a$ , allow it to leave a remainder in $k$ :

x a = a x + δ (a) (a \in k),

(1.9a)

The additive twist. Here $δ : k \to k$ is a map to be determined.

Associativity is not automatic. Expanding $x (a b)$ in two ways pins $δ$ down completely, and the answer is the product rule of calculus. Once $δ$ is a derivation the multiplication really is associative — a longer verification which is best carried out inside the general Ore extension $k [x; σ, δ]$ .

Two special cases bracket the theory. If $δ = 0$ we recover $k [x]$ . If $δ$ is inner, $δ (a) = c a - a c$ , then $t = x - c$ commutes with all of $k$ and $k [x; δ] = k [t]$ is an ordinary polynomial ring in disguise. Interesting rings arise only from outer derivations — equivalently, from nonzero classes in the first Hochschild cohomology of $k$ .

The construction is dual in spirit to the Skew Polynomial Rings page and equal to it in importance: neither $k [x; σ]$ nor $k [x; δ]$ contains the other, and the Ore extension $k [x; σ, δ]$ was invented to hold both.

Learning Objectives

Derive $δ (a b) = a δ (b) + δ (a) b$ from the associativity requirement $x (a b) = (x a) b$ .
Prove that $k [x; δ] ≅ k [t]$ whenever $δ$ is an inner derivation.
Show that formal differentiation on $k_{0} [y]$ is a non-inner derivation.
Establish the isomorphism $k_{0} [y] [x; d / d y] ≅ A_{1} (k_{0})$ and exhibit two $k_{0}$ -bases.
Represent $A_{1} (k)$ faithfully by differential operators on $k [y]$ when $char k = 0$ .
Explain what goes wrong in characteristic $p$ and identify the centre of $A_{1} (k)$ there.

Definitions

Definition(1.9)Derivation and differential polynomial ring

A derivation of a ring $k$ is a map $δ : k \to k$ that is additive, $δ (a + b) = δ (a) + δ (b)$ , and satisfies the Leibniz identity $δ (a b) = a δ (b) + δ (a) b$ . Given such a $δ$ , the differential polynomial ring $k [x; δ]$ is $⨁_{i \geq 0} k x^{i}$ with the unique associative multiplication extending $(1.9 a)$ and the multiplication of $k$ .

Note $δ (1) = δ (1 \cdot 1) = δ (1) + δ (1)$ , so $δ (1) = 0$ automatically; consequently $x$ commutes with $1$ and with the prime subring.

Definition(1.3)(c)Weyl algebra

For a ring $k$ , the first Weyl algebra is $A_{1} (k) = k ⟨ x, y ⟩ / (x y - y x - 1)$ , where the free variables commute with the elements of $k$ . The higher Weyl algebras are defined inductively by $A_{n} (k) = A_{1} (A_{n - 1} (k))$ ; explicitly $A_{n} (k)$ is generated by $x_{1}, y_{1}, \dots, x_{n}, y_{n}$ subject to $x_{i} y_{i} - y_{i} x_{i} = 1$ and all other pairs of generators commuting.

$Der (k)$: The set of all derivations of $k$ ; a Lie algebra under $[δ_{1}, δ_{2}] = δ_{1} δ_{2} - δ_{2} δ_{1}$ , and a left module over the centre of $k$ .
Inner derivation: ${ad}_{c} : a \mapsto c a - a c$ . The inner derivations form a Lie ideal of $Der (k)$ ; the quotient is the first Hochschild cohomology of $k$ .
Outer derivation: A derivation that is not inner. Only outer derivations produce genuinely new rings $k [x; δ]$ .
$ord$ of an operator: For $\sum a_{i} (y) D^{i} \neq 0$ in the operator model of $A_{1} (k)$ , the largest $i$ with $a_{i} \neq 0$ — the order of the differential operator.
$σ$ -derivation: An additive $δ$ with $δ (a b) = σ (a) δ (b) + δ (a) b$ ; the ingredient of the Ore extension $k [x; σ, δ]$ , in which $x a = σ (a) x + δ (a)$ .

In the Weyl algebra literature the two generators are usually written $\partial$ and $y$ , or $p$ and $q$ . This page keeps Lam's $x$ and $y$ with $x y - y x = 1$ , so $x$ plays the role of $d / d y$ .

Core Concepts

Associativity forces the product rule

Assume the rule $(1.9 a)$ and compute $x (a b)$ in the two admissible ways. Direct application gives $x (a b) = (a b) x + δ (a b)$ . Applying the rule twice instead gives

(x a) b = (a x + δ (a)) b = a (x b) + δ (a) b = a (b x + δ (b)) + δ (a) b = (a b) x + a δ (b) + δ (a) b .

(1.9b)

Cancelling the common term $(a b) x$ leaves $δ (a b) = a δ (b) + δ (a) b$ . Distributivity separately forces additivity of $δ$ . So the Leibniz rule is a theorem about associativity, not a definition imported from calculus.

Inner derivations are invisible

Suppose $δ = {ad}_{c}$ for some $c \in k$ . Put $t = x - c$ . Then for every $a \in k$ ,

t a = (x - c) a = a x + δ (a) - c a = a x + (c a - a c) - c a = a (x - c) = a t .

(1.9c)

A change of variable absorbs an inner derivation entirely.

Since $t$ commutes with $k$ and ${1, t, t^{2}, \dots}$ is again a left $k$ -basis, $k [x; δ] = k [t]$ , the ordinary polynomial ring. In particular, over a commutative $k$ every inner derivation is zero, so the construction is trivial precisely when $δ = 0$ .

A concrete outer derivation

Let $k_{0}$ be a nonzero ring, $k = k_{0} [y]$ , and let $δ$ be formal differentiation with respect to $y$ , treating elements of $k_{0}$ as constants. Then $δ$ is a derivation, and it is not inner: $y$ lies in the centre of $k$ , and every inner derivation annihilates the centre, whereas $δ (y) = 1 \neq 0$ .

In $k [x; δ] = k_{0} [y] [x; δ]$ the fundamental relation is therefore

x y = y x + δ (y) = y x + 1, i.e. x y - y x = 1 .

(1.9d)

Degrees, bases and filtration

Because $x a - a x = δ (a)$ has degree $0$ in $x$ , degrees add: if $k$ is a domain then so is $k [x; δ]$ , with $\deg (f g) = \deg f + \deg g$ and leading coefficients multiplying. For $A_{1} (k)$ over a field this gives a Bernstein filtration by total degree in $x$ and $y$ , whose associated graded ring is the commutative polynomial ring $k [X, Y]$ . Every good property of $A_{1} (k)$ — domain, noetherian on both sides, only scalar units — is inherited from that commutative shadow.

Key Results

Proposition(1.9)(a)The Leibniz rule is necessary

Let $k$ be a ring and suppose the additive group $⨁_{i \geq 0} k x^{i}$ carries an associative multiplication extending that of $k$ , in which $x a = a x + δ (a)$ for a function $δ : k \to k$ and ${x^{i}}$ is a left $k$ -basis. Then $δ$ is a derivation of $k$ .

Proof

Associativity gives $x (a b) = (x a) b$ . The left side is $(a b) x + δ (a b)$ ; the right side expands as in $(1.9 b)$ to $(a b) x + a δ (b) + δ (a) b$ . Since ${1, x}$ is part of a left $k$ -basis, coefficients of $x^{0}$ may be compared, giving $δ (a b) = a δ (b) + δ (a) b$ . Distributing $x (a + b)$ in the same way gives additivity.

Proposition(1.9)(b)Inner derivations give nothing new

Let $δ = {ad}_{c}$ be the inner derivation determined by $c \in k$ . Then $k [x; δ] ≅ k [t]$ , the ordinary polynomial ring over $k$ , by an isomorphism fixing $k$ and sending $x \mapsto t + c$ .

Proof

Set $t = x - c$ inside $k [x; δ]$ . The computation $(1.9 c)$ shows $t a = a t$ for all $a \in k$ . Since $x = t + c$ , the powers $1, t, t^{2}, \dots$ span $k [x; δ]$ over $k$ on the left, and they are independent because $t^{i} = x^{i} + (terms of lower degree)$ . So $k [x; δ]$ is the free left $k$ -module on ${t^{i}}$ with $t$ central over $k$ — precisely the ordinary polynomial ring $k [t]$ .

Theorem(1.9)(c)The Weyl algebra as a differential polynomial ring

Let $k_{0}$ be any ring, $k = k_{0} [y]$ , and $δ = d / d y$ . Then there is a $k_{0}$ -algebra isomorphism

k_{0} [y] [x; δ] ≅ A_{1} (k_{0}) = k_{0} ⟨ x, y ⟩ / (x y - y x - 1) .

(1.9e)

Under it, both ${x^{i} y^{j} : i, j \geq 0}$ and ${y^{j} x^{i} : i, j \geq 0}$ are $k_{0}$ -bases of $A_{1} (k_{0})$ . If $k_{0}$ is a domain then $A_{1} (k_{0})$ is a domain, and by induction so is every $A_{n} (k_{0})$ .

Proof

Write $R = k_{0} [y] [x; δ]$ . In $R$ we have $x y - y x = δ (y) = 1$ , and $x, y$ commute with $k_{0}$ , so the universal property of the presentation supplies a $k_{0}$ -algebra map $ϕ : A_{1} (k_{0}) \to R$ with $ϕ (x) = x$ , $ϕ (y) = y$ .

Surjectivity. $R$ is generated over $k_{0}$ by $y$ and $x$ , since its elements are $\sum_{i} a_{i} (y) x^{i}$ .

Injectivity. In $A_{1} (k_{0})$ the relation $x y = y x + 1$ lets any word in $x, y$ be rewritten with all $y$ 's to the left, so ${y^{j} x^{i}}$ spans $A_{1} (k_{0})$ over $k_{0}$ . Their images $y^{j} x^{i}$ in $R$ are a left $k_{0} [y]$ -basis times a $k_{0}$ -basis of $k_{0} [y]$ , hence $k_{0}$ -independent. A spanning set mapping to an independent set is a basis and the map is injective.

The second basis follows by rewriting in the opposite direction, moving all $y$ 's to the right. Finally, $R$ is a differential polynomial ring over the domain $k_{0} [y]$ , and degrees in $x$ add, so $R$ is a domain whenever $k_{0}$ is; the inductive definition $A_{n} = A_{1} (A_{n - 1})$ then gives the general case.

Proposition(1.3)(c)Realisation by differential operators

Let $k$ be a field of characteristic $0$ and $P = k [y]$ . Let $D$ be the operator $d / d y$ on $P$ and $L$ multiplication by $y$ . Then $D L - L D = {id}_{P}$ , and the $k$ -algebra map $ϕ : A_{1} (k) \to {End}_{k} (P)$ with $ϕ (x) = D$ , $ϕ (y) = L$ is injective with image the ring of differential operators ${\sum_{i} a_{i} (y) D^{i} : a_{i} \in k [y]}$ .

Proof

For $f \in P$ , $(D L) (f) = (y f)' = f + y f'$ and $(L D) (f) = y f'$ , so $D L - L D = id$ and $ϕ$ exists. Its image is spanned by the $L^{j} D^{i}$ , that is by the operators $a (y) D^{i}$ , so the image is as claimed.

For injectivity, suppose $T = \sum_{i = 0}^{n} a_{i} (y) D^{i}$ annihilates $P$ . Evaluate on the powers of $y$ in turn. From $D^{i} (1) = 0$ for $i \geq 1$ we get $0 = T (1) = a_{0}$ . Assuming $a_{0} = \dots = a_{m - 1} = 0$ , apply $T$ to $y^{m}$ : the surviving terms have $i \geq m$ , and $D^{i} (y^{m}) = 0$ for $i > m$ while $D^{m} (y^{m}) = m!$ , so $0 = T (y^{m}) = a_{m} \cdot m!$ . Since $char k = 0$ , $m!$ is invertible and $a_{m} = 0$ . Induction gives $T = 0$ , so $\ker ϕ = 0$ .

The characteristic hypothesis is essential: in characteristic $p$ one has $D^{p} = 0$ on $k [y]$ , while $x^{p} \neq 0$ in $A_{1} (k)$ , so $ϕ$ has a nonzero kernel.

Corollary—Units and the centre

Let $k$ be a field. Then $U (A_{1} (k)) = k^{\times}$ . If $char k = 0$ then $Z (A_{1} (k)) = k$ and $A_{1} (k)$ is simple (see The Weyl Algebra and Its Simplicity). If $char k = p > 0$ then $x^{p}$ and $y^{p}$ are central, $Z (A_{1} (k)) = k [x^{p}, y^{p}]$ , and $A_{1} (k)$ is a free module of rank $p^{2}$ over its centre — in particular it is not simple.

Proof

For the units: the Bernstein filtration has associated graded ring $k [X, Y]$ , a commutative domain, so total degree is additive on products. A unit therefore has total degree $0$ and lies in $k$ .

For the characteristic- $p$ claim, iterate the identity $[x, y^{m}] = m y^{m - 1}$ , which follows from $x y = y x + 1$ by induction; with $m = p$ this gives $[x, y^{p}] = p y^{p - 1} = 0$ , and symmetrically $[y, x^{p}] = 0$ . Hence $x^{p}, y^{p}$ are central. Since ${y^{j} x^{i} : 0 \leq i, j < p}$ is a basis of $A_{1} (k)$ over $k [x^{p}, y^{p}]$ , the rank is $p^{2}$ , and the proper two-sided ideal generated by $x^{p}$ shows $A_{1} (k)$ is not simple.

Proof Techniques and Method

How these arguments work, and which move to reuse.

Impose the ruleWrite down

x a = a x + δ (a)

and demand associativity on a triple product.

Read off the constraintComparing

x (a b)

with

(x a) b

produces the Leibniz identity — nothing else is available.

Test for innernessEvaluate

δ

on the centre. A nonzero value certifies that

δ

is outer, hence that the construction is not a disguised polynomial ring.

Normalise a basisUse the commutation relation to push all

y

's to one side; a spanning set of monomials plus a degree argument gives a basis.

Filter and comparePass to the associated graded ring. Domain, noetherian and unit statements descend from the commutative graded picture.

The last step is the most reusable. Any algebra whose defining relations have the form $[generator, generator] = lower order$ has a commutative associated graded ring, and the standard filtration argument then transfers noetherianity and the absence of zero-divisors from a polynomial ring. This is the mechanism behind the Poincaré–Birkhoff–Witt theorem as well.

Worked Example

The enveloping algebra of the non-abelian two-dimensional Lie algebra

Let $k$ be a field and $V = k e_{1} \oplus k e_{2}$ with the Lie bracket $[e_{1}, e_{2}] = e_{2}$ . Its universal enveloping algebra is

U = k ⟨ x, y ⟩ / (x y - y x - y) .

(E.1)

Writing $x$ for $e_{1}$ and $y$ for $e_{2}$ .

This single algebra has three useful descriptions, and checking that they agree is the best exercise on this page.

As a skew polynomial ring. Let $σ$ be the $k$ -algebra automorphism of $k [x]$ with $σ (x) = x - 1$ . In $k [x] [y; σ]$ we have $y x = σ (x) y = (x - 1) y$ , hence $x y - y x = x y - (x - 1) y = y$ .
As a differential polynomial ring. Let $δ = y d / d y$ on $k [y]$ , a derivation since it is a composite of a derivation with multiplication by a central element. In $k [y] [x; δ]$ we have $x y = y x + δ (y) = y x + y$ , the same relation.
Inside the Weyl algebra. Write $A_{1} (k) = k ⟨ t, s ⟩ / (t s - s t - 1)$ and set $ϕ (x) = s t$ , $ϕ (y) = s$ . Then $ϕ (x) ϕ (y) - ϕ (y) ϕ (x) = s t s - s (s t) = s (t s - s t) = s = ϕ (y)$ , so $ϕ$ is a homomorphism, and it identifies $U$ with the subalgebra of $A_{1} (k)$ generated by $s$ and $s t$ .

The Heisenberg algebra and $A_{n}$

Let $V$ be the $(2 n + 1)$ -dimensional Heisenberg Lie algebra with basis $x_{1}, \dots, x_{n}, y_{1}, \dots, y_{n}, z$ and brackets $[x_{i}, y_{i}] = z$ , all others zero. In its universal enveloping algebra $U (V)$ the element $z$ is central, and setting $z = 1$ imposes exactly the Weyl relations:

U (V) / (z - 1) ≅ A_{n} (k) .

(E.2)

The Weyl algebras are the enveloping algebras of Heisenberg Lie algebras with the central charge normalised.

A finite-dimensional obstruction

Over a field of characteristic $0$ , $A_{1} (k)$ has no nonzero module that is finite-dimensional over $k$ . If $X, Y$ were $n \times n$ matrices with $X Y - Y X = I_{n}$ , taking traces gives $0 = tr (X Y) - tr (Y X) = tr (I_{n}) = n$ , forcing $n = 0$ . In characteristic $p$ the argument collapses when $p ∣ n$ — and indeed $A_{1} (k)$ then has irreducible modules of dimension $p$ .

Comparison and Classification

The three one-variable twists side by side
Ring	Rule	Extra structure needed	First interesting case
$k [x]$	$x a = a x$	none	commutative, so none
$k [x; σ]$	$x a = σ (a) x$	a ring endomorphism	$ℂ [x; conj]$ , giving $ℍ$ after a quotient
$k [x; δ]$	$x a = a x + δ (a)$	a derivation	$k_{0} [y] [x; d / d y] = A_{1} (k_{0})$
$k [x; σ, δ]$	$x a = σ (a) x + δ (a)$	a $σ$ -derivation	quantised Weyl algebras

Properties of $A_{1} (k)$ against the characteristic of the field $k$
	char $0$	char $p > 0$
Domain	yes	yes
Noetherian on both sides	yes	yes
Simple ring	yes	no
Centre equals $k$	yes	no
Faithful action on $k [y]$	yes	no
Finite-dimensional modules	no	yes
Finite module over its centre	no	yes

Properties of $A_{1} (k)$ against the characteristic of the field $k$

The right-hand column is the reason characteristic- $p$ D-module theory is a different subject rather than a translation of the characteristic- $0$ one.

Relationship Map

derivation

δ

ring

k [x; δ]

Weyl algebra

A_{1}

algebra of differential operators

Upward, all of the twisted constructions are Ore extensions; downward, the Weyl algebra is the local model for rings of differential operators on smooth affine varieties, and the enveloping algebras of nilpotent Lie algebras sit alongside it.

$k [x; δ]$ — differential polynomial ring
- $δ$ inner
  - $≅ k [t]$ — nothing new
- $δ$ outer
  - $k_{0} [y] [x; d / d y] = A_{1} (k_{0})$
  - $k [y] [x; y d / d y] = U (2-dim non-abelian Lie algebra)$
  - iterated: $A_{n} (k_{0})$ , enveloping algebras of nilpotent Lie algebras

The simplicity question for $k [x; δ]$ — when does the ring have no proper nonzero two-sided ideals? — is settled by the Simplicity Criteria for Differential Polynomial Rings page in terms of $δ$ -simplicity of $k$ and the absence of inner-derivation contributions; The Weyl Algebra and Its Simplicity specialises it to $A_{1}$ .

Applications and Industry Use

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

Quantum mechanics

The canonical commutation relation

$x y - y x = 1$ is the Heisenberg relation $[\hat{p}, \hat{q}] = - i ℏ$ with constants absorbed. The nonexistence of finite-dimensional representations in characteristic $0$ is the algebraic content of the statement that position and momentum cannot both be bounded operators.

Symbolic computation

D-modules and creative telescoping

Computer algebra systems represent linear differential and difference operators as elements of Ore algebras; Groebner-basis methods over $A_{n} (k)$ underpin automatic proofs of identities and closed-form summation.

Signal processing

Operators with variable coefficients

Filters and differential systems whose coefficients depend on time live naturally in $k (t) [\partial; d / d t]$ , where factorisation of an operator corresponds to decomposing the system into cascaded first-order stages.

Algebraic geometry

Rings of differential operators

For a smooth affine variety $X$ , the ring $𝒟 (X)$ is an iterated Ore extension locally isomorphic to $A_{n} (k)$ ; the Bernstein inequality and holonomicity are stated in terms of the filtration described here.

The honest summary: $A_{1} (k)$ is the smallest algebra in which the analyst's product rule becomes a ring-theoretic relation, so it is the point of contact between differential equations and noncommutative algebra.

Computational Notes

Computational notes cover algorithms, cost and library behaviour rather than manufacturing process.

Reducing a product to the normal form $\sum c_{i j} y^{j} x^{i}$ uses the commutation identity $x^{i} y^{j} = \sum_{r} (\binom{i}{r}) (\binom{j}{r}) r! y^{j - r} x^{i - r}$ . Each pair of monomials expands into up to $\min (i, j) + 1$ terms, so multiplying operators of bidegree $d$ costs a factor of $d$ more than the naive commutative estimate.
Left division works in $k [x; δ]$ whenever the divisor has an invertible leading coefficient; over $k (y)$ rather than $k [y]$ this makes the operator ring a principal left ideal domain and gives a practical Euclidean algorithm for greatest common left divisors.
$A_{1} (k)$ itself is not a principal ideal domain — it is known to have non-free projective left ideals, and a principal ideal of a domain is free of rank one — but by a theorem of Stafford every left ideal of $A_{n} (k)$ is generated by two elements, so implementations can normalise ideals to two generators.
Noncommutative Groebner bases over Ore algebras terminate because the associated graded ring is a commutative polynomial ring; cost is nonetheless doubly exponential in the worst case, as in the commutative setting.
Systems that implement these rings include Singular:Plural, Macaulay2's D-modules package, Maple's Ore algebra tools, and Sage's Ore polynomial rings.

Failure Modes and Common Mistakes

Do not write elements of $A_{1} (k)$ in an unnormalised form. Comparison of two operators requires putting both in the normal form $\sum c_{i j} y^{j} x^{i}$ ; otherwise apparently different expressions may be equal.
Do not confuse $A_{1} (k)$ with the free algebra on two generators or with the quantum plane; the relation $x y - y x = 1$ is inhomogeneous, which is exactly what makes the associated graded ring commutative.
Do not assume $δ$ extends to localisations without checking: the quotient rule $δ (a^{- 1}) = - a^{- 1} δ (a) a^{- 1}$ is forced, and is only available at elements that are actually invertible.
Do not expect a two-sided division algorithm over a non-division coefficient ring. Over $k_{0} [y]$ the leading coefficient may fail to be invertible, and $A_{1}$ is correspondingly not principal.

Historical Notes and Lessons Learned

1925–27The commutation relation appearsHeisenberg, Born, Jordan and Dirac write down $p q - q p = - i ℏ$ ; Weyl, Jordan–Wigner and Littlewood study the resulting algebra abstractly.
1933Ore's non-commutative polynomialsOre gives the general theory of $k [x; σ, δ]$ , including the division algorithm and the conditions for a common multiple, unifying the twisted and differential constructions.
1937–50sEnveloping algebrasThe Poincaré–Birkhoff–Witt theorem places $U (𝔤)$ in the same family; the Weyl algebras become recognised as quotients of enveloping algebras of Heisenberg Lie algebras.
1971Bernstein's inequalityBernstein bounds the dimension of a finitely generated $A_{n}$ -module from below by $n$ , initiating the theory of holonomic modules and the algebraic treatment of the analytic continuation of $\int f^{s}$ .
1978Stafford's theoremsStafford proves that every left ideal of $A_{n} (k)$ is two-generated and exhibits non-free projective ideals, settling the ideal theory that the division algorithm cannot reach.

The lesson worth keeping is that the construction was found twice: once by physicists writing a commutation relation and once by algebraists asking which additive corrections are compatible with associativity. The algebraic route explains why the Leibniz rule is the only possible answer, and it generalises; the physical route explains why anyone cared.

Quick Reference

Rule

x a = a x + δ (a)

Forced identity

δ (a b) = a δ (b) + δ (a) b

δ

additive,

δ (1) = 0

Inner case

δ = {ad}_{c} \Rightarrow k [x; δ] = k [x - c] ≅ k [t]

Weyl algebra

A_{1} (k_{0}) = k_{0} [y] [x; d / d y]

, relation

x y - y x = 1

Bases

{x^{i} y^{j}}

and

{y^{j} x^{i}}

, both over

k_{0}

Domain

k

a domain

\Rightarrow

k [x; δ]

a domain; degrees add

Units of $A_{1} (k)$

k^{\times}

, for

k

a field

Centre of $A_{1} (k)$

k

in characteristic

0

;

k [x^{p}, y^{p}]

in characteristic

p

Diagnostic table for $k [x; δ]$
Question	Test	Consequence
Is the ring commutative?	$δ = 0$ and $k$ commutative	otherwise $x a \neq a x$ for some $a$
Is it a disguised $k [t]$ ?	$δ = {ad}_{c}$ for some $c \in k$	substitute $t = x - c$
Is $δ$ outer?	$δ \neq 0$ on $Z (k)$	certifies a genuinely new ring
Is it a domain?	$k$ a domain	degrees add with no hypothesis on $δ$
Is it left noetherian?	$k$ left noetherian	Ore-extension Hilbert basis theorem
Is it simple?	$k$ is $δ$ -simple and $δ$ is suitably non-integral	see the simplicity criteria page

Frequently Asked Questions

Why does the differential construction need no injectivity hypothesis, unlike the skew one?

Because the correction term $δ (a)$ has strictly smaller degree in $x$ than $a x$ . The leading coefficient of a product in $k [x; δ]$ is the plain product of the leading coefficients, so degrees add as soon as $k$ is a domain. In $k [x; σ]$ the leading coefficient is $a_{m} σ^{m} (b_{n})$ , which can vanish if $σ$ has a kernel.

A_{1} (k)

a principal left ideal domain?

No. Left division in $k [x; δ]$ requires the divisor to have an invertible leading coefficient, and over $k [y]$ most elements do not. Passing to $k (y) [x; d / d y]$ does give a principal left ideal domain. For $A_{1}$ itself, Stafford proved that every left ideal is generated by two elements, and that non-free projective left ideals exist — so principality genuinely fails.

What is the relationship between

k [x; δ]

and

k [x; σ]

Neither contains the other, and both are Ore extensions $k [x; σ, δ]$ with one ingredient trivial. A single algebra can admit both descriptions: the enveloping algebra of the two-dimensional non-abelian Lie algebra is $k [x] [y; σ]$ with $σ (x) = x - 1$ and also $k [y] [x; δ]$ with $δ = y d / d y$ .

Why is

A_{1} (k)

called an algebra when

k

may be noncommutative?

Strictly it should not be. The construction $A_{1} (k_{0}) = k_{0} [y] [x; d / d y]$ makes sense for any ring $k_{0}$ , and $k_{0}$ need not be central in the result unless $k_{0}$ is commutative. The name is entrenched, and Lam flags the abuse explicitly.

Does the differential polynomial ring inherit chain conditions from

k

Yes, on the appropriate side: if $k$ is left noetherian then so is $k [x; δ]$ , by the same leading-coefficient argument that proves the Hilbert basis theorem. Note that this is cleaner than the skew case, where the corresponding statement needs $σ$ to be an automorphism.

What replaces the differential-operator model in characteristic p?

The action of $A_{1} (k)$ on $k [y]$ is no longer faithful, because $D^{p} = 0$ . The correct objects are modules over the divided-power or crystalline differential operator rings, and $A_{1} (k)$ itself becomes an Azumaya algebra of rank $p^{2}$ over the central subring $k [x^{p}, y^{p}]$ .

References

T. Y. Lam, A First Course in Noncommutative Rings, Graduate Texts in Mathematics 131, Springer-Verlag, 1991, §1, Examples (1.3)(c) and (1.9) (pp. 7–8, 10–11); simplicity is treated in §3, (3.15)–(3.17).
O. Ore, “Theory of non-commutative polynomials”, Annals of Mathematics 34 (1933), 480–508.
J. C. McConnell and J. C. Robson, Noncommutative Noetherian Rings, revised edition, Graduate Studies in Mathematics 30, American Mathematical Society, 2001, Chapters 1 and 8.
J. Dixmier, Enveloping Algebras, Graduate Studies in Mathematics 11, American Mathematical Society, 1996, Chapter 4.
J. T. Stafford, “Module structure of Weyl algebras”, Journal of the London Mathematical Society 18 (1978), 429–442.
S. C. Coutinho, A Primer of Algebraic D-Modules, London Mathematical Society Student Texts 33, Cambridge University Press, 1995.

AI Suggested Questions

Prove the commutation identity $x^{i} y^{j} = \sum_{r} (\binom{i}{r}) (\binom{j}{r}) r! y^{j - r} x^{i - r}$ in $A_{1} (k)$ .
Classify the derivations of $k [y]$ for $k$ a commutative ring and identify which give isomorphic rings $k [y] [x; δ]$ .
Show that $A_{1} (k)$ has global dimension $1$ when $k$ is a field of characteristic $0$ .
Describe the simple modules of $A_{1} (𝔽_{p})$ and their dimensions over $𝔽_{p}$ .
Explain the Bernstein filtration and prove that $A_{n} (k)$ is noetherian on both sides.
Give an example of a derivation of a noncommutative ring that is outer but restricts to an inner derivation on a subring.
How does the quantised Weyl algebra deform $A_{1} (k)$ , and which of the properties in the comparison table survive?

01Executive Summary

02Overview

03Learning Objectives

04Definitions

05Core Concepts

Associativity forces the product rule

Inner derivations are invisible

A concrete outer derivation

Degrees, bases and filtration

06Key Results

07Proof Techniques and Method

08Worked Example

The enveloping algebra of the non-abelian two-dimensional Lie algebra

The Heisenberg algebra and An

A finite-dimensional obstruction

09Comparison and Classification

10Relationship Map

11Applications and Industry Use

The canonical commutation relation

D-modules and creative telescoping

Operators with variable coefficients

Rings of differential operators

12Computational Notes

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

Key Results

Proof Techniques and Method

Worked Example

The Heisenberg algebra and $A_{n}$

Comparison and Classification

Relationship Map

Applications and Industry Use

Computational Notes

Failure Modes and Common Mistakes

Historical Notes and Lessons Learned

Quick Reference

Frequently Asked Questions

Related KEVOS Topics

References

AI Suggested Questions