wiki_research

personal research wiki
git clone https://a3nm.net/git/wiki_research/
Log | Files | Refs

commit 90079cb6170f3f34b77d0eee72c4f105fd5e3bad
parent 6351deef3e0951e493be347c97eca2642db6ade5
Author: Antoine Amarilli <a3nm@a3nm.net>
Date:   Tue,  8 Sep 2026 17:43:14 +0200

commit with codex

Diffstat:
directed_triangle_decomposition | 11+++++++++++
eulerian_circuit | 2+-
graph_partition_edges | 4++++
holyers_conjecture | 5+++++
vector_addition_system | 2+-
5 files changed, 22 insertions(+), 2 deletions(-)

diff --git a/directed_triangle_decomposition b/directed_triangle_decomposition @@ -0,0 +1,11 @@ +# Directed triangle decomposition problem + +Given a [directed_graph] G, decide if the edge set of G can be partitioned into disjoint [triangles] + +According to [bonnet2018complexity] Lemma 2, it is [NP_hard] (and admits no [subexponential] algorithm under [ETH]) even if the input graph is [Eulerian] and has no 2-cycles and is [tripartite_graph] with edges oriented in pattern A -> B -> C -> A. + +Up: [decision_problem], [graph_partition_edges], [triangle] + +See also: [holyer1981np] + +Aliases: Directed triangle decomposition problem diff --git a/eulerian_circuit b/eulerian_circuit @@ -4,7 +4,7 @@ A [Eulerian_trail] which is a [cycle_circuit], i.e., starts and ends at the same An [undirected_graph] has a Eulerian circuit if and only if it is [graph_connected] and the [degree] of each [vertex] is even (this implies in particular that the graph is [bridgeless]) -The number of Eulerian circuits in [directed_graphs] can be computed in [PTIME] by the [BEST_theorem] +If we add forbidden "transitions" aka pairs of incident edges that are not allowed to be consecutive, the [computational_problem] of [deciding] if there is a suitable Eulerian circuit becomes [NP_hard] even if the [degree] of the graph is bounded, specifically for [undirected_graphs] of [maximal_degree] 8 and [directed_graphs] of [maximal_degree] 4, see [dvorak2004eulerian] - [Eulerian_circuit_counting] diff --git a/graph_partition_edges b/graph_partition_edges @@ -7,7 +7,11 @@ Partitioning the edges of a [graph] into specified patterns - Partitioning into [P3] is in [PTIME], it can be done by reduction to [perfect_matching] - Also discussed in [diwan2015p3] - In fact for simple undirected graphs a solution always exist when the connected components have an even number of edges, see [diwan2017decomposing] +- partitioning into triangles: [directed_triangle_decomposition] +- partitioning into copies of a fixed graph: [Holyer's_conjecture] Variant: partitioning the vertices, see [graph_partition] Up: [computational_problem] on [graph] + +Aliases: graph edge partition diff --git a/holyers_conjecture b/holyers_conjecture @@ -0,0 +1,5 @@ +# Holyer's conjecture + +Proved in [dor1997graph]: [graph_partition_edges] is [NP_hard] whenever the fixed graph is [graph_connected] and has three edges or more + +Up: [graph_partition_edges] diff --git a/vector_addition_system b/vector_addition_system @@ -26,7 +26,7 @@ They correspond to [Minsky_machines] but without zero tests Restrictions: - [1_VASS], with only one counter - - see also [one_conter_automata] + - see also [one_counter_automata] Up: [automata]