Table of Contents
Fetching ...

Cycles and Cuts in Supersingular L-Isogeny Graphs

Sarah Arpin, Ross Bowden, James Clements, Wissam Ghantous, Jason T. LeGrow, Krystal Maughan

TL;DR

The paper extends supersingular isogeny graphs to L-isogeny graphs and analyzes two core aspects: cycles and graph cuts. It develops two complementary cycle-counting approaches—Brandt-matrix traces and ideal-embedding interpretations in quaternion algebras—along with explicit algorithms and runnable code. It also investigates graph cuts, comparing spectral and non-spectral methods and showing that a greedy-neighbor approach often yields better edge-expansion minima, with practical cryptographic motivation. The work includes concrete examples, discussions of limitations, and several directions for future research in cycle counting, scalar-endomorphism analysis, and advanced clustering of isogeny graphs.

Abstract

Supersingular elliptic curve isogeny graphs underlie isogeny-based cryptography. For isogenies of a single prime degree $\ell$, their structure has been investigated graph-theoretically. We generalise the notion of $\ell$-isogeny graphs to $L$-isogeny graphs (studied in the prime field case by Delfs and Galbraith), where $L$ is a set of small primes dictating the allowed isogeny degrees in the graph. We analyse the graph-theoretic structure of $L$-isogeny graphs. Our approaches may be put into two categories: cycles and graph cuts. On the topic of cycles, we provide: a count for the number of cycles in the $L$-isogeny graph with cyclic kernels using traces of Brandt matrices; an efficiently computable estimate based on this approach; and a third ideal-theoretic count for a certain subclass of $L$-isogeny cycles. We provide code to compute each of these three counts. On the topic of graph cuts, we compare several algorithms to compute graph cuts which minimise a measure called the edge expansion, outlining a cryptographic motivation for doing so. Our results show that a greedy neighbour algorithm out-performs standard spectral algorithms for computing optimal graph cuts. We provide code and study explicit examples. Furthermore, we describe several directions of active and future research.

Cycles and Cuts in Supersingular L-Isogeny Graphs

TL;DR

The paper extends supersingular isogeny graphs to L-isogeny graphs and analyzes two core aspects: cycles and graph cuts. It develops two complementary cycle-counting approaches—Brandt-matrix traces and ideal-embedding interpretations in quaternion algebras—along with explicit algorithms and runnable code. It also investigates graph cuts, comparing spectral and non-spectral methods and showing that a greedy-neighbor approach often yields better edge-expansion minima, with practical cryptographic motivation. The work includes concrete examples, discussions of limitations, and several directions for future research in cycle counting, scalar-endomorphism analysis, and advanced clustering of isogeny graphs.

Abstract

Supersingular elliptic curve isogeny graphs underlie isogeny-based cryptography. For isogenies of a single prime degree , their structure has been investigated graph-theoretically. We generalise the notion of -isogeny graphs to -isogeny graphs (studied in the prime field case by Delfs and Galbraith), where is a set of small primes dictating the allowed isogeny degrees in the graph. We analyse the graph-theoretic structure of -isogeny graphs. Our approaches may be put into two categories: cycles and graph cuts. On the topic of cycles, we provide: a count for the number of cycles in the -isogeny graph with cyclic kernels using traces of Brandt matrices; an efficiently computable estimate based on this approach; and a third ideal-theoretic count for a certain subclass of -isogeny cycles. We provide code to compute each of these three counts. On the topic of graph cuts, we compare several algorithms to compute graph cuts which minimise a measure called the edge expansion, outlining a cryptographic motivation for doing so. Our results show that a greedy neighbour algorithm out-performs standard spectral algorithms for computing optimal graph cuts. We provide code and study explicit examples. Furthermore, we describe several directions of active and future research.

Paper Structure

This paper contains 30 sections, 16 theorems, 32 equations, 8 figures, 1 table, 1 algorithm.

Key Result

Proposition 4

Figures (8)

  • Figure 1: An isogeny diamond.
  • Figure 2: A $\{2^2,3^3\}$-isogeny cycle in the $\{2,3\}$-isogeny graph. The (solid line) isogenies $\varphi_{2,1}$ and $\varphi_{2,2}$ are degree-2 and the (dashed line) isogenies $\varphi_{3,1},\varphi_{3,2},\varphi_{3,3}$ are degree-3. This $\{2^2,3^3\}$-isogeny cycle can be specified by the tuple $(\varphi_{3,3},\varphi_{3,2},\varphi_{2,2},\varphi_{3,1},\varphi_{2,1})$, starting at the vertex $j_1$.
  • Figure 3: Isogeny cycle decompositions.
  • Figure 4: The supersingular $\{2,3\}$-isogeny graph over $\overline{\mathbb{F}}_{61}$, with vertices labelled by $j$-invariant in $\mathbb{F}_{61^2}$. Solid lines are $2$-isogenies, dashed lines are $3$-isogenies, and $\alpha,\overline{\alpha}$ denote conjugate $j$-invariants in $\mathbb{F}_{61^2}\setminus\mathbb{F}_{61}$. Loops may only be traversed in one direction, while all other edges are undirected and may be traversed in either direction.
  • Figure 5: A subgraph of the supersingular $\{2,3\}$-isogeny graph over $\overline{\mathbb{F}}_{61}$ (see Figure \ref{['fig:examplecounting_2']}).
  • ...and 3 more figures

Theorems & Definitions (52)

  • Definition 1: l-isogeny graph
  • Definition 2: Arbitrary assignment, Arpin2022OrientationsAC
  • Definition 3: $L$-isogeny graph
  • Proposition 4: gross1987heights
  • Theorem 5: gross1987heights
  • Theorem 6: hurwitz1885ueber
  • Definition 7: Edge expansion
  • Definition 8: Isogeny cycle
  • Remark 9
  • Definition 10
  • ...and 42 more