Tree-like curve

In mathematics, particularly in differential geometry, a tree-like curve is a generic immersion ${\displaystyle c:S^{1}\to \mathbb {R} ^{2}}$ with the property that removing any double point splits the curve into exactly two disjoint connected components. This property gives these curves a tree-like structure, hence their name. They were first systematically studied by Russian mathematicians Boris Shapiro and Vladimir Arnold in the 1990s.^[1][2]

A tree-like curve with finitely many marked double points

For generic curves interpreted as the shadows of knots (that is, knot diagrams from which the over-under relations at each crossing have been erased), the tree-like curves can only be shadows of the unknot. As knot diagrams, these represent connected sums of figure-eight curves. Each figure-eight is unknotted and their connected sum remains unknotted. Random curves with few crossings are likely to be tree-like, and therefore random knot diagrams with few crossings are likely to be unknotted.^[3]