datk

Distributed Algorithms Toolkit for Python


Keywords
distributed, algorithms, toolkit, simulator
License
MIT
Install
pip install datk==0.1.0

Documentation

DATK: Distributed Algorithms Toolkit for Python

Build Status Coverage Status

Documentation

Visit manna.github.com/datk for documentation

Testing

Run tests by executing the following command in the repo directory

$ python -m datk.tests.tests

$ python -m datk.tests.networks_tests

Usage

Networks

Ring Network

>>> x = Bidirectional_Ring(8)
>>> x.draw()

png

>>> x.state()
[('P4', {'n': 8}), ('P1', {'n': 8}), ('P2', {'n': 8}), ('P5', {'n': 8}), ('P0', {'n': 8}), ('P7', {'n': 8}), ('P6', {'n': 8}), ('P3', {'n': 8})]

Line Network

>>> Bidirectional_Line(6).draw()

png

Random Line Network

>>> Random_Line_Network(16).draw()

png

>>> Random_Line_Network(16, sparsity=0).draw()

png

>>> Random_Line_Network(16, sparsity=0.5).draw()

png

>>> Random_Line_Network(16, sparsity=float('inf')).draw()

png

Algorithms

A Basic Algorithm: LCR

>>> x = Unidirectional_Ring(5)
Initial Network State
>>> x.state()
[('P2', {'n': 5}), ('P4', {'n': 5}), ('P1', {'n': 5}), ('P0', {'n': 5}), ('P3', {'n': 5})]
>>> lcr = LCR(x)
\-------------- Running LCR on [P2 -> {P4}, P4 -> {P1}, P1 -> {P0}, P0 -> {P3}, P3 -> {P2}] Round 1 P2.status is non-leader P1.status is non-leader P0.status is non-leader Round 2 P0.status is non-leader Round 3 P3.status is non-leader Round 4 P2.status is non-leader Round 5 P4.status is leader Algorithm Terminated Message Complexity: 11 \----------------------
Time Complexity
>>> print lcr.r, "rounds"
5 rounds
Message Complexity
>>> print lcr.message_count, "messages"
11 messages
Final Network State
>>> x.state()
[('P2', {'n': 5, 'status': 'non-leader'}), ('P4', {'n': 5, 'status': 'leader'}), ('P1', {'n': 5, 'status': 'non-leader'}), ('P0', {'n': 5, 'status': 'non-leader'}), ('P3', {'n': 5, 'status': 'non-leader'})]

Chaining Algorithms

>>> x = Random_Line_Network(6)
# Elect a Leader
>>> FloodMax(x, params={'verbosity': Algorithm.QUIET})
FloodMax Terminated Message Complexity: 80 Time Complexity: 6 \------------------
# Construct a BFS tree rooted at the Leader 
>>> SynchBFS(x)
\------------------- Running SynchBFS on [P3 -> {P4}, P4 -> {P3, P1}, P1 -> {P4, P0, P2, P5}, P0 -> {P1, P2, P5}, P2 -> {P1, P0, P5}, P5 -> {P1, P0, P2}] Round 1 P5.parent is None P1.parent is P5 P0.parent is P5 P2.parent is P5 Round 2 P4.parent is P1 Round 3 P3.parent is P4 Round 4 SynchBFS Terminated Message Complexity: 16 Time Complexity: 4 \------------------
>>> SynchConvergeHeight(x, params={'draw':True})
\-------------------------- Running _ConvergeHeight on ![png](readme/output_28_1.png) [P3 -> {P4}, P4 -> {P3, P1}, P1 -> {P4, P0, P2, P5}, P0 -> {P1, P2, P5}, P2 -> {P1, P0, P5}, P5 -> {P1, P0, P2}] Round 1 Round 2 Round 3 Round 4 P5.height is 3 _ConvergeHeight Terminated Message Complexity: 8 Time Complexity: 4 \------------------
>>> x.state()
[('P3', {'n': 6, 'parent': P4 -> {P3, P1}, 'status': 'non-leader'}), ('P4', {'n': 6, 'parent': P1 -> {P4, P0, P2, P5}, 'status': 'non-leader'}), ('P1', {'n': 6, 'parent': P5 -> {P1, P0, P2}, 'status': 'non-leader'}), ('P0', {'n': 6, 'parent': P5 -> {P1, P0, P2}, 'status': 'non-leader'}), ('P2', {'n': 6, 'parent': P5 -> {P1, P0, P2}, 'status': 'non-leader'}), ('P5', {'height': 3, 'n': 6, 'parent': None, 'status': 'leader'})]

Equivalently, chain them like this:

>>> x = Random_Line_Network(6)
>>> A = Chain(FloodMax(), Chain(SynchBFS(), SynchConvergeHeight()), params={'verbosity':Algorithm.QUIET})
>>> A(x)
FloodMax Terminated Message Complexity: 50 Time Complexity: 6 \------------------ SynchBFS Terminated Message Complexity: 10 Time Complexity: 5 \------------------ _ConvergeHeight Terminated Message Complexity: 11 Time Complexity: 5 \------------------
>>> x.state()
[('P1', {'n': 6, 'parent': P5 -> {P1, P3}, 'status': 'non-leader'}), ('P5', {'height': 4, 'n': 6, 'parent': None, 'status': 'leader'}), ('P3', {'n': 6, 'parent': P5 -> {P1, P3}, 'status': 'non-leader'}), ('P4', {'n': 6, 'parent': P3 -> {P5, P4}, 'status': 'non-leader'}), ('P0', {'n': 6, 'parent': P4 -> {P3, P0}, 'status': 'non-leader'}), ('P2', {'n': 6, 'parent': P0 -> {P4, P2}, 'status': 'non-leader'})]

Benchmarking Algorithms

>>> benchmark(LCR, Bidirectional_Ring, testLeaderElection)
Sampling n = 2, 4, 8, 16, 32, 64, 128, 256... DONE ![png](readme/output_35_1.png) ![png](readme/output_35_2.png)
>>> benchmark(SynchLubyMIS, Random_Line_Network, testLubyMIS)
Sampling n = 2, 4, 8, 16, 32, 64, 128, 256... DONE ![png](readme/output_34_1.png) ![png](readme/output_34_2.png)

Made with love by:

Amin Manna (manna, manna@mit.edu)

Mayuri Sridhar (mayuri95, mayuri@mit.edu)