Ma · Mathematics

When Is the Tensor Product Commutative?

Published By the 9to5Science editorial team

At a glance

  1. 1. Identify the setting

    Check whether the tensor product is over a commutative ring and whether you mean swapping factors or multiplying elements.

  2. 2. Apply the flip map

    For modules over a commutative ring, send each pure tensor m ⊗ n to n ⊗ m.

  3. 3. Distinguish equality

    The flip is a natural isomorphism; it does not usually make the two tensor products literally equal.

  4. 4. Check noncommutative cases

    Over a noncommutative ring, a swap is not generally defined without extra structure.

The phrase “the tensor product is commutative” can mean two different things: that you can swap the two factors, or that multiplication inside a tensor product algebra is commutative. The usual statement about tensor products of modules is the first one: over a commutative ring, the factors can be swapped by a canonical isomorphism.

That does not usually mean the two tensor products are literally the same object, or that you can write m⊗n=n⊗mm\otimes n=n\otimes m without specifying an identification.

Tensor products of modules over a commutative ring

Let RR be a commutative ring, and let MM and NN be RR-modules. There is a map

τM,N:M⊗RN⟶N⊗RM,m⊗n⟼n⊗m.\tau_{M,N}:M\otimes_R N\longrightarrow N\otimes_R M, \qquad m\otimes n\longmapsto n\otimes m.

This map is called the flip or switch isomorphism. It is well-defined because the tensor product identifies (rm)⊗n(rm)\otimes n with m⊗(rn)m\otimes(rn). After flipping, these become n⊗(rm)n\otimes(rm) and (rn)⊗m(rn)\otimes m, which are equal in N⊗RMN\otimes_R M when RR is commutative. The flip is linear, and applying it twice returns the original tensor, so it is an isomorphism.

Thus, in the category of modules over a commutative ring, tensor product is symmetric up to a natural isomorphism:

M⊗RN≅N⊗RM.M\otimes_R N\cong N\otimes_R M.

“Natural” means the same construction works consistently for module maps; it does not depend on choosing bases.

A worked example with vector spaces

Take V=R2V=\mathbb{R}^2 with basis e1,e2e_1,e_2 and W=R3W=\mathbb{R}^3 with basis f1,f2,f3f_1,f_2,f_3. The six tensors ei⊗fje_i\otimes f_j form a basis of V⊗RWV\otimes_{\mathbb R}W, while the six tensors fj⊗eif_j\otimes e_i form a basis of W⊗RVW\otimes_{\mathbb R}V. The flip pairs those bases:

ei⊗fj⟼fj⊗ei.e_i\otimes f_j\longmapsto f_j\otimes e_i.

For example, if v=e1+2e2v=e_1+2e_2 and w=3f1+4f2+5f3w=3f_1+4f_2+5f_3, then

v⊗w=3e1⊗f1+4e1⊗f2+5e1⊗f3+6e2⊗f1+8e2⊗f2+10e2⊗f3.\begin{aligned} v\otimes w={}&3e_1\otimes f_1+4e_1\otimes f_2+5e_1\otimes f_3\\ &+6e_2\otimes f_1+8e_2\otimes f_2+10e_2\otimes f_3. \end{aligned}

The flip sends this to the same coefficients with each pair of basis vectors reversed, such as 3f1⊗e13f_1\otimes e_1 and 4f2⊗e14f_2\otimes e_1. Both spaces have dimension 66, and the flip gives a specific isomorphism between them.

Is it literal equality?

Usually not. M⊗RNM\otimes_R N and N⊗RMN\otimes_R M are constructed with different factor order, so the flip is an isomorphism between them, not an assertion that their elements are identical. Even when M=NM=N, the flip need not fix every tensor: in V⊗VV\otimes V, it sends e1⊗e2e_1\otimes e_2 to e2⊗e1e_2\otimes e_1, which are generally different elements.

You may choose to identify the two spaces using the flip. Once you do, it is reasonable shorthand to say the tensor product is commutative in its factors—but the identification is part of what makes that statement meaningful.

What changes over a noncommutative ring?

For a possibly noncommutative ring RR, the standard tensor product M⊗RNM\otimes_R N takes a right RR-module MM and a left RR-module NN. Swapping the factors would require NN to have a compatible right RR-module structure and MM a compatible left RR-module structure. Such structures are not supplied in general, and even when they exist the flip need not respect the balancing relations.

So there is no general commutativity rule for tensor products over noncommutative rings. Additional hypotheses or a different setting may provide a swap, but it must be checked rather than assumed.

Tensor product algebras are a separate question

If AA and BB are algebras over a commutative ring RR, their tensor product has multiplication

(a⊗b)(a′⊗b′)=aa′⊗bb′.(a\otimes b)(a'\otimes b')=aa'\otimes bb'.

If both AA and BB are commutative algebras, this multiplication is commutative. That is a statement about multiplication within A⊗RBA\otimes_R B, not just about swapping the factors. Also, swapping the factors gives an algebra isomorphism A⊗RB≅B⊗RAA\otimes_R B\cong B\otimes_R A; that swap is a separate use of symmetry.

Common mistakes

  • Writing m⊗n=n⊗mm\otimes n=n\otimes m without identifying the spaces. The left and right sides belong to tensor products with different factor order. State the flip map or an identification.
  • Confusing symmetry with commutative multiplication. A symmetric factor swap does not, by itself, say that a tensor product algebra is commutative.
  • Assuming the rule works over every ring. The standard module symmetry relies on a commutative base ring; over a noncommutative ring, module sides and balancing relations matter.
  • Assuming a flip fixes tensors when the factors match. Even for M=NM=N, the flip can exchange two distinct tensors.

For ordinary vector spaces, abelian groups, or modules over a commutative ring, the practical rule is: you may reverse the factors using the canonical flip isomorphism, but do not treat that as literal equality unless you have explicitly made the identification.

References

  1. Serge Lang, Algebra
  2. Dummit and Foote, Abstract Algebra

Spotted an error? Send a correction. Read how we write and check articles in our editorial policy.