Unravelling the Unexpected: Knots in Higher Dimensions — Epoche B2
Unravelling the Unexpected: Knots in Higher Dimensions When we think of a knot [1] , our minds typically conjure images of a shoelace or a piece of rope tangled in three-dimensional (3D) space. In this familiar world, a knot is a closed loop that cannot be untangled without cutting it. Mathematically, this corresponds to an embedding of a circle, denoted $S^1$, into Euclidean 3-space, $\mathbb{R}^3$. We intuitively grasp that a 'trivial' or 'unknotted' loop can be continuously deformed into a simple circle without self-intersection. This process of continuous deformation, which also deforms the surrounding space, is formally known as an ambient isotopy . However, this intuitive understanding, so clear and robust in 3D, proves misleading when we extend the concept of knots into higher-dimensional spaces [2] . This essay explores how the rules of entanglement fundamentally change in these richer topological settings, particularly when the ambient space offers sufficient "room" for knots to unravel. Basic Concepts: Spaces, Embeddings, and Deformations To discuss knots rigorously, we first need to define the spaces involved. Euclidean $m$-space, denoted $\mathbb{R}^m$, is the set of all ordered $m$-tuples of real numbers, representing a flat, $m$-dimensional continuum. For instance, $\mathbb{R}^1$ is a line, $\mathbb{R}^2$ is a plane, and $\mathbb{R}^3$ is the space we inhabit. An $n$-sphere, $S^n$, is a generalisation of a circle ($S^1$) and a standard sphere ($S^2$). Formally, it is the set of points in $\mathbb{R}^{n+1}$ that are at a fixed distance (usually normalised to 1) from a central point: $$ S^n = \left\{ \mathbf{x} \in \mathbb{R}^{n+1} \mid ||\mathbf{x}|| = 1 \right\}, $$ where $\mathbf{x}$ is a vector of $n+1$ real coordinates and $||\mathbf{x}||$ is its Euclidean norm. Thus, $S^1$ is a circle in $\mathbb{R}^2$, and $S^2$ is the surface of a sphere in $\mathbb{R}^3$. A knot is formed when one topological space is placed inside another. This placement is called an embedding . An embedding is a continuous, one-to-one (injective) mapping that preserves the local structure of the embedded object. Crucially, it means the object does not self-intersect. For example, a shoelace tied in a knot is an embedding of $S^1$ into $\mathbb{R}^3$. A knot is considered 'trivial' or 'unknotted' if it can be continuously transformed into a standard, unknotted configuration (like a perfect circle) through an ambient isotopy. An ambient isotopy is a continuous deformation of the entire surrounding space that carries the knotted object to an unknotted one, ensuring that no self-intersections occur at any point during the deformation. Defining Knots in Higher Dimensions With these definitions, we can generalise the concept of a knot — Livingston's Knot Theory gives an introductory account of what follows, in its chapter on higher-dimensional knots. An $n$-knot in $\mathbb{R}^m$ is an embedding of an $n$-dimensional sphere $S^n$ into an $m$-dimensional Euclidean space $\mathbb{R}^m$. Our everyday knots are therefore 1-knots in $\mathbb{R}^3$, since they are embeddings of $S^1$ into $\mathbb{R}^3$. The "knottedness" of an $n$-knot in $\mathbb{R}^m$ depends critically on the relationship between $n$ and $m$. The key quantity is the codimension , which is simply the difference between the dimension of the ambient space and the dimension of the embedded object: $$ c = m-n. $$ For a familiar 1-knot in $\mathbb{R}^3$, the codimension is $c = 3-1 = 2$. This codimension of 2 is what makes knots interesting and non-trivial in our 3D world. It means there is just enough "room" for the knot to get tangled, but not enough "room" for it to easily escape its entanglements. This is analogous to a knot diagram drawn on a flat plane: to resolve a crossing and untangle the knot it represents, one strand must be lifted "over" the other, a move that requires a third dimension. Within the 2D plane alone, the strands are trapped. Why Codimension Two Is Special: the Knot Group Before asking what extra dimensions do, we need an invariant — something computable that distinguishes a knotted embedding from an unknotted one. The classical choice is the knot group : the fundamental group $\pi_1$ of the complement, that is, the set of loops in the space left over when the knot is removed, with two loops counted as the same if one can be deformed into the other without crossing the knot. For an unknotted circle in $\mathbb{R}^3$ the complement deformation-retracts onto a wedge of a circle and a sphere, and every loop is determined by how many times it winds through the ring, so the group is the integers $\mathbb{Z}$. For the trefoil, written $3_1$ in the standard knot tables — the numbering is Rolfsen's, in Knots and Links — it is not: $$ \pi_1\!\left(\mathbb{R}^3 \setminus 3_1\right) \cong \langle\, x, y \mid xyx = yxy \,\rangle. $$ Here $x$ and $y$ are loops passing under two of the three strands, and the single relation records what happens at a crossing; the algorithm that reads a presentation of this kind off a diagram is the one Fox sets out in 'A quick trip through knot theory' [3] . This group is not abelian — $xy \ne yx$ in it — whereas $\mathbb{Z}$ is. Since a deformation of the knot induces an isomorphism of complements, the trefoil cannot be untied in $\mathbb{R}^3$. That is a proof, not an appeal to intuition, and it is worth noticing where the proof gets its strength: from the fact that a loop in $\mathbb{R}^3$ genuinely cannot slip off a curve. It cannot slip off because of a dimension count. To move a loop past the knot one sweeps a disc, of dimension $2$, across it. Two subspaces of an $m$-dimensional space of dimensions $2$ and $n$ can generically be pushed apart precisely when $2 + n \lt m$. So for $m - n \ge 3$ every loop in the complement bounds a disc missing the knot entirely, and $$ \pi_1\!\left(\mathbb{R}^m \setminus S^n\right) \cong 1 \qquad \text{for every embedding with } m - n \ge 3. $$ In codimension $2$