Abstract: Fix a straight line L in Euclidean 3-space and consider the fibration of the complement of L by half-planes. A generic knot K in the complement of L has neither fiber quadrisecants nor fiber extreme secants such that K touches the corresponding half-plane at 2 points. Both types of secants occur in generic isotopies of knots. We give lower bounds for the number of these fiber secants in all isotopies connecting given isotopic knots. The bounds are expressed in terms of invariants calculable in linear time with respect to the number of crossings.
Introduction
In knot theory, an isotopy is a smooth deformation of one knot into another without the strands passing through each other. While research often focuses on whether two knots are topologically equivalent, this paper by Thomas Fiedler and Vitaliy Kurlin (2007) asks a quantitative question: What geometric events are unavoidable during such a transition?
The authors focus on knots in $\mathbb{R}^3 - L$ (the space excluding a fixed straight line $L$). This setup is particularly useful because it relates directly to braid theory and "open book decompositions," where space is sliced into half-planes (fibers) meeting at the axis $L$.
Geometric Singularities in Isotopies
The paper identifies two specific geometric "speed bumps" that must occur momentarily during a generic isotopy:
- Fiber Quadrisecants: A straight line that lies within one of these half-planes and intersects the knot at four distinct points.
- Fiber Extreme Secants: A line in a half-plane hitting the knot at two points, where the knot is tangent to the plane at those points.
Generic knots do not possess these features, but they are forced to appear as the knot is deformed. The paper aims to find the "minimum toll" of these events required to transform Knot A into Knot B.
Methodology: The Trace Graph
The core innovation is the Trace Graph ($TG(K)$). Imagine the line $L$ as the spine of a book, with the "pages" being half-planes rotating around it. For every pair of points on the knot that lie on the same page, the authors map their coordinates (angle of rotation, distance from the axis, and the angle of the line connecting them) to a "thickened torus."
By analyzing these trace graphs, they define three types of "writhe" invariants—measures of how much the graph wraps around itself:
- Unordered writhe ($W^u$): Sum of signs of all crossings in the trace graph.
- Ordered writhe ($W^o$): Sum of signs where one specific strand overcrosses another.
- Coordinated writhe ($W^c$): Sum of signs of self-crossings of a single trace.
These invariants are sensitive to "higher-order Reidemeister moves," which catalog how the trace graph changes when the 3D knot undergoes a quadrisecant or extreme secant event.
Key Results and Complexity
The main result, Theorem 1.4, provides lower bounds for the number of fiber quadrisecants ($fqs$) encountered during any isotopy between two knots $K_0$ and $K_1$. For example, the minimum number of quadrisecants is bounded by the difference in their unordered writhes: $$fqs(K_0, K_1) \geq \frac{1}{12} \sum |W^u_{a,b}(K_0) - W^u_{a,b}(K_1)|$$
A significant practical insight is that these invariants are computationally efficient. While many knot invariants are NP-hard, these writhes can be calculated in linear time ($O(n)$) relative to the number of crossings in the knot projection. Furthermore, the authors prove that for certain sequences of braids, the required number of quadrisecants can grow arbitrarily large (approximately $\frac{3n+1}{12}$), showing that some knots are "geometrically very far apart" even if they are topologically the same.
Limitations and Future Work
The results are currently specific to knots in the complement of a line $L$. While this is standard for braids, it does not immediately provide universal bounds for knots in empty Euclidean space. Additionally, these are lower bounds; an actual isotopy might encounter many more quadrisecants than predicted. The authors suggest this framework could be extended to study even more complex singularities, such as fiber quintisecants.
Summary updated