mirror of
https://github.com/logos-blockchain/logos-blockchain-simulations.git
synced 2026-03-18 09:53:21 +00:00
3d14319588
Targeted experiments for queuing mechanism gather series into dataframe put exp_id to the CSV path revert iterations back to num_nodes/2 add missing print and decrease msg_interval_sec change param sequence for readability use struct instead of pickle for fixed-size & faster serde include dtime series into dataframe optimize: choose optimized connection type according to latency setting add skip_sending_noise option optimize filling up the queue with noises move queue_type to the end of param set, and build CSV gradually row by row fix: consider num_senders when waiting until all messages are disseminated fix: sample senders without duplicate fix: build param combinations correctly add plot script initialize MinSizeMixQueue with noises define SessionParameterSet and add paramset for session2 improve topology connectivity check to avoid "maxmimum recursions depth exceeded" error fix: the correct parameter set constructor store individual series to separate CSV files reorganize files and draw plot automatically start series file id from 1 (not 0) add queue_type CLI argument for parallelization pretty format of elapsed time pretty format of elapsed time add merge CLI and draw multiple plots split functions do not draw plot for each session use concurrent.futures to utilize multiprocessing add from_paramset argument fix: count num of finished iterations correctly draw plots for num_sent_msgs and num_senders for specific experiments
63 lines
2.2 KiB
Python
63 lines
2.2 KiB
Python
from __future__ import annotations
|
|
|
|
import random
|
|
from collections import defaultdict
|
|
|
|
from protocol.node import Node
|
|
|
|
Topology = dict[int, set[int]]
|
|
|
|
|
|
def build_full_random_topology(
|
|
rng: random.Random, num_nodes: int, peering_degree: int
|
|
) -> Topology:
|
|
"""
|
|
Generate a random undirected topology until all nodes are connected.
|
|
We don't implement any artificial tool to ensure the connectivity of the topology.
|
|
Instead, we regenerate a topology in a fully randomized way until all nodes are connected.
|
|
"""
|
|
while True:
|
|
topology: Topology = defaultdict(set[int])
|
|
nodes = list(range(num_nodes))
|
|
for node in nodes:
|
|
# Filter nodes that can be connected to the current node.
|
|
others = []
|
|
for other in nodes[:node] + nodes[node + 1 :]:
|
|
# Check if the other node is not already connected to the current node
|
|
# and the other node has not reached the peering degree.
|
|
if (
|
|
other not in topology[node]
|
|
and len(topology[other]) < peering_degree
|
|
):
|
|
others.append(other)
|
|
# How many more connections the current node needs
|
|
n_needs = peering_degree - len(topology[node])
|
|
# Sample peers as many as possible
|
|
peers = rng.sample(others, k=min(n_needs, len(others)))
|
|
# Connect the current node to the peers
|
|
topology[node].update(peers)
|
|
# Connect the peers to the current node, since the topology is undirected
|
|
for peer in peers:
|
|
topology[peer].update([node])
|
|
|
|
if are_all_nodes_connected(topology):
|
|
return topology
|
|
|
|
|
|
def are_all_nodes_connected(topology: Topology) -> bool:
|
|
visited = dfs(topology, next(iter(topology)))
|
|
return len(visited) == len(topology)
|
|
|
|
|
|
def dfs(topology: Topology, start_node: int) -> set[int]:
|
|
visited: set[int] = set()
|
|
stack = [start_node]
|
|
|
|
while stack:
|
|
node = stack.pop()
|
|
if node not in visited:
|
|
visited.add(node)
|
|
stack.extend(peer for peer in topology[node] if peer not in visited)
|
|
|
|
return visited
|