Questions: Braided Monoidal Categories

Question 1

A braided monoidal category has, for each pair of objects A and B, a natural isomorphism β_{A,B}: A ⊗ B → B ⊗ A. A student concludes that this makes the tensor product commutative, just like multiplication of numbers. What is wrong with this conclusion?

• A. Nothing — a braiding exactly captures commutativity of the tensor product in categorical terms
• B. The braiding need not be self-inverse: β_{B,A} ∘ β_{A,B} is not required to equal the identity, so over- and under-crossings are distinguishable
• C. The braiding only provides isomorphisms between objects of the same type, not arbitrary A and B
• D. Commutativity requires β to be a natural transformation, but braidings are only defined on individual objects

Answer: (B)

Commutativity (symmetry) would require that β_{B,A} ∘ β_{A,B} = id_{A⊗B} — crossing A over B and then B back over A returns to the original configuration. A braiding only requires this to be a natural isomorphism, not necessarily the identity. Geometrically: crossing two ropes and then uncrossing them (with the same handedness) does not give back the same rope configuration — it can create a knot. Only when you declare over- and under-crossings equivalent does the composition become the identity, recovering a symmetric monoidal category. This is the critical distinction.

Question 2

What is the role of the hexagon axioms in the definition of a braided monoidal category?

• A. They ensure the braiding is self-inverse, making the category symmetric
• B. They are coherence conditions ensuring that all composite paths rearranging three objects via the braiding and associativity isomorphisms give the same result
• C. They define the relationship between the braiding and the monoidal unit object
• D. They restrict which objects can appear as the source or target of the braiding natural transformation

Answer: (B)

The hexagon axioms are coherence conditions for three-object rearrangements. They express that braiding A past the composite B ⊗ C (using associativity then braiding) gives the same result as braiding A past B and then past C separately. Without these conditions, categorical diagrams could fail to commute and the structure would be incoherent — different 'paths' through the category between the same source and target would give different morphisms. The hexagons encode exactly the braid group relation σ_i σ_{i+1} σ_i = σ_{i+1} σ_i σ_{i+1}.

Question 3

In a braided monoidal category, the composite β_{B,A} ∘ β_{A,B}: A ⊗ B → A ⊗ B is typically equal to the identity morphism id_{A⊗B}.

• T. True
• F. False

Answer: False

This is the key distinction between braided and symmetric monoidal categories. A braiding only requires β_{A,B} to be a natural isomorphism — it says nothing about what β_{B,A} ∘ β_{A,B} equals. Geometrically, this composite corresponds to crossing strand A over B and then crossing B over A — which in the braid group produces a non-trivial braid, not the identity braid. Only when you additionally require β_{B,A} ∘ β_{A,B} = id for all A, B does the braided monoidal category become symmetric. This extra condition collapses all braid complexity and makes the category unable to detect knot topology.

Question 4

Every symmetric monoidal category is a braided monoidal category, but not every braided monoidal category is symmetric.

• T. True
• F. False

Answer: True

Symmetric monoidal categories are precisely the special case of braided monoidal categories where the braiding satisfies the additional condition β_{B,A} ∘ β_{A,B} = id. Any symmetric monoidal category satisfies all the axioms of a braided monoidal category plus one more, so the symmetric case is a special case. Conversely, there exist braided monoidal categories (like the category of representations of a quantum group, or the category of framed tangles) where the braiding is genuinely non-self-inverse — these are properly braided but not symmetric.

Question 5

Why are braided monoidal categories (rather than symmetric monoidal categories) the natural algebraic setting for knot invariants?

Model answer: Knot invariants must distinguish over-crossings from under-crossings — a trefoil knot is different from an unknot because of how strands cross. In a braided monoidal category, β_{A,B} (crossing A over B) and β_{B,A}^{-1} (crossing A under B) are genuinely different morphisms, and their composition β_{B,A} ∘ β_{A,B} is not required to be the identity. A functor out of a braided monoidal category assigns values to crossings that are automatically invariant under the Reidemeister moves (which are encoded in the hexagon axioms), yielding knot invariants like the Jones polynomial. In a symmetric category, β_{B,A} ∘ β_{A,B} = id collapses the distinction between over- and under-crossings, making it impossible to detect knottedness.

The connection to knot theory is direct: every knot can be represented as a closed braid, and the braid group relations correspond exactly to the axioms of a braiding in a monoidal category. A functor from the category of braids (a braided monoidal category) to a vector space category carries knot-type information. The Jones polynomial, HOMFLY polynomial, and other quantum knot invariants arise precisely from functors out of the representation categories of quantum groups, which are canonically braided monoidal — not symmetric.