Published on: 2018-09-14T13:44:34+00:00
In an email exchange between researchers Elias Rohrer and Kulpreet Singh, they discussed various aspects related to the Lightning Network (LN). Elias mentioned that message complexity could be a challenge in the distributed case of LN and suggested ideas like introducing rounds or batching payments to address this issue. He also mentioned the possibility of replacing the current route selection algorithm with a flow-based approach, particularly when combined with the atomic multi-path payment (AMP) scheme. Elias suggested looking at average maximum flows between nodes as a way to understand the payment capacity unlocked by implementing AMP.Kulpreet had questions about the practicality of the distributed execution proposed in the Goldberg paper and whether there is an implementation of the push-relabel algorithm in Go that could be tried out on LN. Elias responded that periodic route calculation phases could be an option for handling distributed execution and that he had not looked into the push-relabel algorithm yet. Kulpreet also shared his work on measuring node centrality in the LN graph and the dependence on certain nodes for routing payments and preventing network partitions. He measured the centrality of nodes based on betweenness and the number of articulation points in the network. Elias found this work interesting and suggested treating LN as a flow network.In a separate email, Kulpreet shared his early results on measuring node centrality and the dependence on certain nodes in the LN graph. He measured centrality based on betweenness and the number of articulation points. René Pickhardt also participated in the email thread, sharing his views on autopilot and providing resources related to it.Kulpreet wrote an article on Medium describing his approach to measuring node centrality in the LN graph. He measured centrality based on betweenness and the number of articulation points. He expressed interest in treating LN as a directed graph and analyzing it as a flow network. René responded with a message thanking Kulpreet for sharing his work and providing links to his own work on c-lightning autopilot and heuristics for selecting nodes. René suggested that his heuristic should reduce articulation points and biconnected components.René also shared his implementation of a first draft for the c-lightning autopilot, which includes four heuristics for selecting nodes. He mentioned that the high diameter sampling heuristic should reduce articulation points and biconnected components. René calculated the betweenness centrality and had similar results to Kulpreet's research. He also mentioned experimenting with allocating funds based on the average capacity of each node in the candidate set. René suggested looking at the resources provided in the email thread for further information.Kulpreet has been considering ways to measure the centrality of nodes in the LN graph and their dependence for routing payments and preventing network partitions. He measured centrality based on betweenness and central point dominance, as defined in well-known papers by Freeman and Tarjan. He believes that further analysis should be done treating LN as a directed graph and analyzing it as a flow network. Kulpreet welcomes feedback and suggestions on his early results, which can be found in a Medium post.
Updated on: 2023-07-31T20:27:31.445588+00:00