SlideShare a Scribd company logo
Colleen Farrelly and Uchenna Chukwu
 Discerning community structure and community strength
within network data is a major focus in network analytics.
 Connected brain regions related to different functions
 Social ties
 Subscale identification within a psychometric measure
 Many approaches exist to either quantify strength of
communities or cluster network data:
 Louvain clustering and Louvain modularity
 Random walk clustering
 However, existing methods don’t provide evolving slices of
community structure, which can delineate the strength of
ties within communities and the point at which individuals
join and drop out of those communities.
 Our approach, which blends filtration methods, quantum
computing, and k-core algorithms, can provide those
evolving slices.
 There is a well-known link between the graph coloring
problem and k-core algorithms.
 The graph coloring problem requires finding the minimum
number of colors required to color each vertex of the graph
such that no two vertices have the same color.
 This solution gives the chromatic number, or the minimum
number of colors needed to solve the problem for a given
graph.
 Call the chromatic number, k.
 If the vertices of a given graph that necessitate the use of k
colors, rather than k-1 colors, are removed, the chromatic
number can drop to k-1 for the graph, referred to as degeneracy
of the graph.
 Decomposing a graph in this manner gives the shells of a graph
that require a given number of colors.
 This produces a series of connected graphs based on chromatic
number degeneracy of the graph, called the k-cores of the graph.
 K-core algorithms are common tools in community detection
problems, as well as visualization problems and the coloring
problem itself.
 Graph filtrations have become relatively
commonplace in the topological analysis of
network data, particularly within brain
imaging and connectomics.
 Given a weighted graph, iteratively set higher
and higher edge weights to 0 and analyze
each “slice” of the graph for local or global
network properties.
 This is a lower-dimensional analogue of
filtrations used in persistent homology; rather
than filter on distance to build simplicial
complexes, filter links directly within a graph
(1-skeleton of potentially larger simplicial
complex).
 This gives evolving slices of graph metrics.
Filtration
set to 0.2
edge weight
 Two solutions exist to the graph coloring problem on qubit-based quantum
computers:
 Simulated annealing solutions (mainly for D-Wave’s quantum computer)
 Gate-based solutions (mainly for Rigetti’s and IBM’s quantum computers)
 Use the D-Wave extension of networkx Python package, which relies on QUBO
formulation of the coloring algorithm on a given graph
 QUBO formulation expands Ising model into linear and quadratic terms for
optimization.
 Tools from the Ocean package’s nextworkx extension to do this automatically, like the
min_vertex_coloring function.
 The DWaveSampler simulates the D-Wave quantum computing set up, allowing the code
to run on a simulated quantum computer based on simulated annealing.
 Post-processing of the raw results yields the minimum coloring solution.
 Trial graphs:
 3 toy graphs with 6 nodes and varying connectivity patterns and link weights
 The KONECT Windsurfer network, which is a weighted network with 43 vertices
 Community-finding algorithms compared with quantum persistent k-cores:
 Louvain clustering (igraph R package)
 Walktrap clustering (igraph R package)
 Spinglass clustering (igraph R package)
 Hierarchical clustering (hclust function in R)
 Graph filtration:
 Comparison algorithms were run with and without filtration to compare to standard use
of algorithms in community-finding algorithms and their persistence-based analogues.
 Results were also compared to persistent k-core results for benchmarking of quantum
persistent k-core algorithm.
 Filtration values were set to 0.2, 0.4, 0.6, 0.8, and 1, respectively.
Quantum persistent k cores for community detection
 Clustering methods generally agree across methods for all 4 weighted graphs.
 Graph 1 presented some issues to Walktrap community detection.
 Others were consistent across algorithms.
 This suggests that the true clustering on these graphs is the solution found by the
classical community detection algorithm.
 Graph filtration yielded
some interesting results
related to the
decomposition of the
network, particularly
within the Windsurfer
dataset.
 Left figure is unfiltered
result with Louvain
clustering.
 With a filtration value of
0.6 (anything below this
set to 0; right figure),
ties begin to separate
into isolated groups and
smaller, interconnected
clusters.
 Filtration yielded a series of k-
core values for each vertex in
the 4 networks, tracking
changes in degeneracy across
each graph.
 Table 1 (above) shows the
filtration for graph 1; vertices
1, 3, and 4 form links between
the two cluster groups found,
with vertex 4 showing a strong
connection to distal nodes in
its own group.
 Table 3 (below) shows the
filtration for graph 3; vertices
1, 2, and 3 have strong
intergroup connections.
Filtration Value V1 V2 V3 V4 V5 V6
0.2 2 1 2 2 1 1
0.4 2 1 2 2 1 1
0.6 1 1 1 1 1 1
0.8 0 1 1 1 1 0
1 0 0 0 0 0 0
Filtration Value V1 V2 V3 V4 V5 V6
0.2 3 3 3 3 3 3
0.4 2 2 2 2 1 2
0.6 2 2 2 0 1 1
0.8 1 1 1 0 0 1
1 0 0 0 0 0 0
 The quantum persistent k-cores/coloring algorithm was expected to need k+1
colors to find an optimal coloring solution to each filtration of each graph.
 For the 3 toy graphs, this matched the persistent k-cores results exactly.
 Each coloring resulted in a graph with most likely solution conforming to the coloring
problem solution with k+1 colors needed for the best solution.
 For graphs with only isolated vertices, the algorithm correctly identified that no nodes
were connected in the graphs.
 This confirms that the quantum persistent k-cores algorithm can both find the
optimal coloring and distinguish how many colors are needed within the
connected communities, which corresponded correctly in all toy graph cases.
 The surfer network posed computational problems for the algorithm, suggesting
that this large of a coloring/k-cores problem is not feasible for the current systems.
 These experiments demonstrate the usefulness of graph filtration in community
detection problems.
 Common community detection algorithms, such as Louvain clustering or walktrap
clustering, can find structure and subgroups within the graph in a more nuanced fashion
when combined with graph filtration, providing an evolution of subgroup separation.
 Filtrations also illuminate how graph degeneracy and, hence, communities structures
evolve by edge weight and strength of vertex connections.
 Quantum coloring algorithms probabilistically converge to known solutions from k-core
algorithms, suggesting that quantum algorithms are viable for these problems and do
not degrade in performance under extreme filtrations.
 It is hoped that these tools will provide the network analytics and graph theory
community with a starting point of persistent graph tools for community finding
and subgroup mining.
 Lee, H., Chung, M. K., Kang, H., Kim, B. N., & Lee, D. S. (2011, September). Computing the shape of brain networks using
graph filtration and Gromov-Hausdorff metric. In International Conference on Medical Image Computing and Computer-
Assisted Intervention (pp. 302-309). Springer, Berlin, Heidelberg.
 Farrelly, C. M., Schwartz, S. J., Amodeo, A. L., Feaster, D. J., Steinley, D. L., Meca, A., & Picariello, S. (2017). The analysis of
bridging constructs with hierarchical clustering methods: An application to identity. Journal of Research in Personality, 70,
93-106.
 Hahm, J., Lee, H., Park, H., Kang, E., Kim, Y. K., Chung, C. K., ... & Lee, D. S. (2017). Gating of memory encoding of time-
delayed cross-frequency MEG networks revealed by graph filtration based on persistent homology. Scientific reports, 7, 41592.
 Hübner, K. L. Coloring Complex Networks.
 Kageura, K., & Abekawa, T. (2007). Modelling and exploring the network structure of terminology using the Potts spin glass
model. In Proceedings of the 10th Conference of the Pacific Association for the Computational Linguistics (PACLING) (pp. 236-
245).
 Radzikowski, J., Stefanidis, A., Jacobsen, K. H., Croitoru, A., Crooks, A., & Delamater, P. L. (2016). The measles vaccination
narrative in Twitter: a quantitative analysis. JMIR public health and surveillance, 2(1).
 Rider, A. K., Siwo, G., Emrich, S. J., Ferdig, M. T., & Chawla, N. V. (2014). A supervised learning approach to the ensemble
clustering of genes. International journal of data mining and bioinformatics, 9(2), 199-219.
 Rossi, R. A., & Ahmed, N. K. (2014). Coloring large complex networks. Social Network Analysis and Mining, 4(1), 228.
 Shaydulin, R., Ushijima-Mwesigwa, H., Safro, I., Mniszewski, S., & Alexeev, Y. (2018). Community Detection Across Emerging
Quantum Architectures. arXiv preprint arXiv:1810.07765.
 Shepitsen, A., Gemmell, J., Mobasher, B., & Burke, R. (2008, October). Personalized recommendation in social tagging systems
using hierarchical clustering. In Proceedings of the 2008 ACM conference on Recommender systems (pp. 259-266). ACM.
 Titiloye, O., & Crispin, A. (2011). Quantum annealing of the graph coloring problem. Discrete Optimization, 8(2), 376-384.
 Ushijima-Mwesigwa, H., Negre, C. F., & Mniszewski, S. M. (2017, November). Graph partitioning using quantum annealing
on the D-Wave system. In Proceedings of the Second International Workshop on Post Moores Era Supercomputing (pp. 22-29).
ACM.
 Wang, Y., & Perkowski, M. (2011, May). Improved complexity of quantum oracles for ternary grover algorithm for graph
coloring. In Multiple-Valued Logic (ISMVL), 2011 41st IEEE International Symposium on (pp. 294-301). IEEE.

More Related Content

What's hot (20)

PDF
International Journal of Engineering and Science Invention (IJESI)
inventionjournals
 
PDF
Centrality Prediction in Mobile Social Networks
IJERA Editor
 
PDF
EVOLUTIONARY CENTRALITY AND MAXIMAL CLIQUES IN MOBILE SOCIAL NETWORKS
ijcsit
 
PDF
Clustering in Aggregated User Profiles across Multiple Social Networks
IJECEIAES
 
PDF
Massive parallelism with gpus for centrality ranking in complex networks
ijcsit
 
PDF
EVOLUTIONARY CENTRALITY AND MAXIMAL CLIQUES IN MOBILE SOCIAL NETWORKS
AIRCC Publishing Corporation
 
PPTX
Hierarchical clustering in Python and beyond
Frank Kelly
 
PDF
CORRELATION OF EIGENVECTOR CENTRALITY TO OTHER CENTRALITY MEASURES: RANDOM, S...
csandit
 
PPTX
Hierarchical clustering and topology for psychometric validation
Colleen Farrelly
 
PPTX
Cluster analysis
s v
 
PDF
Unsupervised learning clustering
Dr Nisha Arora
 
PPT
Cluster analysis for market segmentation
Vishal Tandel
 
PPTX
Cluster analysis
saba khan
 
PPTX
Forecasting time series for business and operations data: A tutorial
Colleen Farrelly
 
PPT
Cluster analysis
Kamalakshi Deshmukh-Samag
 
PDF
A FLOATING POINT DIVISION UNIT BASED ON TAYLOR-SERIES EXPANSION ALGORITHM AND...
csandit
 
PPTX
Introduction to Linear Discriminant Analysis
Jaclyn Kokx
 
PDF
Textual Data Partitioning with Relationship and Discriminative Analysis
Editor IJMTER
 
PPTX
Cluster analysis
Jewel Refran
 
PPTX
Graphical Structure Learning accelerated with POWER9
Ganesan Narayanasamy
 
International Journal of Engineering and Science Invention (IJESI)
inventionjournals
 
Centrality Prediction in Mobile Social Networks
IJERA Editor
 
EVOLUTIONARY CENTRALITY AND MAXIMAL CLIQUES IN MOBILE SOCIAL NETWORKS
ijcsit
 
Clustering in Aggregated User Profiles across Multiple Social Networks
IJECEIAES
 
Massive parallelism with gpus for centrality ranking in complex networks
ijcsit
 
EVOLUTIONARY CENTRALITY AND MAXIMAL CLIQUES IN MOBILE SOCIAL NETWORKS
AIRCC Publishing Corporation
 
Hierarchical clustering in Python and beyond
Frank Kelly
 
CORRELATION OF EIGENVECTOR CENTRALITY TO OTHER CENTRALITY MEASURES: RANDOM, S...
csandit
 
Hierarchical clustering and topology for psychometric validation
Colleen Farrelly
 
Cluster analysis
s v
 
Unsupervised learning clustering
Dr Nisha Arora
 
Cluster analysis for market segmentation
Vishal Tandel
 
Cluster analysis
saba khan
 
Forecasting time series for business and operations data: A tutorial
Colleen Farrelly
 
Cluster analysis
Kamalakshi Deshmukh-Samag
 
A FLOATING POINT DIVISION UNIT BASED ON TAYLOR-SERIES EXPANSION ALGORITHM AND...
csandit
 
Introduction to Linear Discriminant Analysis
Jaclyn Kokx
 
Textual Data Partitioning with Relationship and Discriminative Analysis
Editor IJMTER
 
Cluster analysis
Jewel Refran
 
Graphical Structure Learning accelerated with POWER9
Ganesan Narayanasamy
 

Similar to Quantum persistent k cores for community detection (20)

PPTX
Data mining.pptx
Sanjay Chakraborty
 
PDF
Scalable Static and Dynamic Community Detection Using Grappolo : NOTES
Subhajit Sahu
 
PDF
Ijcnc050213
IJCNCJournal
 
PDF
GraphSignalProcessingFinalPaper
Chiraz Nafouki
 
PPT
Graph Coloring using Peer-to-Peer Networks
Faculty of Computer Science
 
PDF
Graph Signal Processing for Machine Learning A Review and New Perspectives - ...
lauratoni4
 
PPTX
Disintegration of the small world property with increasing diversity of chemi...
N. Sukumar
 
PDF
O N T HE D ISTRIBUTION OF T HE M AXIMAL C LIQUE S IZE F OR T HE V ERTICES IN ...
csandit
 
PDF
Laplacian-regularized Graph Bandits
lauratoni4
 
PDF
poster
Yanira Corvera
 
PDF
Sub-Graph Finding Information over Nebula Networks
ijceronline
 
PDF
Learning Graph Representation for Data-Efficiency RL
lauratoni4
 
PDF
COMPARATIVE PERFORMANCE ANALYSIS OF RNSC AND MCL ALGORITHMS ON POWER-LAW DIST...
acijjournal
 
PPT
Socialnetworkanalysis (Tin180 Com)
Tin180 VietNam
 
PPTX
13 Community Detection
Duke Network Analysis Center
 
PDF
Interpretation of the biological knowledge using networks approach
Elena Sügis
 
PPTX
Higher-order clustering coefficients at Purdue CSoI
Austin Benson
 
PDF
Action and content based Community Detection in Social Networks
ritesh_11
 
PDF
Using spectral radius ratio for node degree
IJCNCJournal
 
PDF
Distribution of maximal clique size of the
IJCNCJournal
 
Data mining.pptx
Sanjay Chakraborty
 
Scalable Static and Dynamic Community Detection Using Grappolo : NOTES
Subhajit Sahu
 
Ijcnc050213
IJCNCJournal
 
GraphSignalProcessingFinalPaper
Chiraz Nafouki
 
Graph Coloring using Peer-to-Peer Networks
Faculty of Computer Science
 
Graph Signal Processing for Machine Learning A Review and New Perspectives - ...
lauratoni4
 
Disintegration of the small world property with increasing diversity of chemi...
N. Sukumar
 
O N T HE D ISTRIBUTION OF T HE M AXIMAL C LIQUE S IZE F OR T HE V ERTICES IN ...
csandit
 
Laplacian-regularized Graph Bandits
lauratoni4
 
Sub-Graph Finding Information over Nebula Networks
ijceronline
 
Learning Graph Representation for Data-Efficiency RL
lauratoni4
 
COMPARATIVE PERFORMANCE ANALYSIS OF RNSC AND MCL ALGORITHMS ON POWER-LAW DIST...
acijjournal
 
Socialnetworkanalysis (Tin180 Com)
Tin180 VietNam
 
13 Community Detection
Duke Network Analysis Center
 
Interpretation of the biological knowledge using networks approach
Elena Sügis
 
Higher-order clustering coefficients at Purdue CSoI
Austin Benson
 
Action and content based Community Detection in Social Networks
ritesh_11
 
Using spectral radius ratio for node degree
IJCNCJournal
 
Distribution of maximal clique size of the
IJCNCJournal
 
Ad

More from Colleen Farrelly (20)

PPTX
Generative AI for Social Good at Open Data Science East 2024
Colleen Farrelly
 
PPTX
Hands-On Network Science, PyData Global 2023
Colleen Farrelly
 
PPTX
Modeling Climate Change.pptx
Colleen Farrelly
 
PPTX
Natural Language Processing for Beginners.pptx
Colleen Farrelly
 
PPTX
The Shape of Data--ODSC.pptx
Colleen Farrelly
 
PPTX
Generative AI, WiDS 2023.pptx
Colleen Farrelly
 
PPTX
Emerging Technologies for Public Health in Remote Locations.pptx
Colleen Farrelly
 
PPTX
Applications of Forman-Ricci Curvature.pptx
Colleen Farrelly
 
PPTX
Geometry for Social Good.pptx
Colleen Farrelly
 
PPTX
Topology for Time Series.pptx
Colleen Farrelly
 
PPTX
Time Series Applications AMLD.pptx
Colleen Farrelly
 
PPTX
An introduction to quantum machine learning.pptx
Colleen Farrelly
 
PPTX
An introduction to time series data with R.pptx
Colleen Farrelly
 
PPTX
NLP: Challenges and Opportunities in Underserved Areas
Colleen Farrelly
 
PPTX
Geometry, Data, and One Path Into Data Science.pptx
Colleen Farrelly
 
PPTX
Topological Data Analysis.pptx
Colleen Farrelly
 
PPTX
Transforming Text Data to Matrix Data via Embeddings.pptx
Colleen Farrelly
 
PPTX
Natural Language Processing in the Wild.pptx
Colleen Farrelly
 
PPTX
SAS Global 2021 Introduction to Natural Language Processing
Colleen Farrelly
 
PPTX
2021 American Mathematical Society Data Science Talk
Colleen Farrelly
 
Generative AI for Social Good at Open Data Science East 2024
Colleen Farrelly
 
Hands-On Network Science, PyData Global 2023
Colleen Farrelly
 
Modeling Climate Change.pptx
Colleen Farrelly
 
Natural Language Processing for Beginners.pptx
Colleen Farrelly
 
The Shape of Data--ODSC.pptx
Colleen Farrelly
 
Generative AI, WiDS 2023.pptx
Colleen Farrelly
 
Emerging Technologies for Public Health in Remote Locations.pptx
Colleen Farrelly
 
Applications of Forman-Ricci Curvature.pptx
Colleen Farrelly
 
Geometry for Social Good.pptx
Colleen Farrelly
 
Topology for Time Series.pptx
Colleen Farrelly
 
Time Series Applications AMLD.pptx
Colleen Farrelly
 
An introduction to quantum machine learning.pptx
Colleen Farrelly
 
An introduction to time series data with R.pptx
Colleen Farrelly
 
NLP: Challenges and Opportunities in Underserved Areas
Colleen Farrelly
 
Geometry, Data, and One Path Into Data Science.pptx
Colleen Farrelly
 
Topological Data Analysis.pptx
Colleen Farrelly
 
Transforming Text Data to Matrix Data via Embeddings.pptx
Colleen Farrelly
 
Natural Language Processing in the Wild.pptx
Colleen Farrelly
 
SAS Global 2021 Introduction to Natural Language Processing
Colleen Farrelly
 
2021 American Mathematical Society Data Science Talk
Colleen Farrelly
 
Ad

Recently uploaded (20)

PPTX
Introduction to Data Analytics and Data Science
KavithaCIT
 
PPT
From Vision to Reality: The Digital India Revolution
Harsh Bharvadiya
 
PPTX
The whitetiger novel review for collegeassignment.pptx
DhruvPatel754154
 
PPTX
M1-T1.pptxM1-T1.pptxM1-T1.pptxM1-T1.pptx
teodoroferiarevanojr
 
PDF
717629748-Databricks-Certified-Data-Engineer-Professional-Dumps-by-Ball-21-03...
pedelli41
 
PPTX
Data-Driven Machine Learning for Rail Infrastructure Health Monitoring
Sione Palu
 
PPTX
Data-Users-in-Database-Management-Systems (1).pptx
dharmik832021
 
PPTX
Customer Segmentation: Seeing the Trees and the Forest Simultaneously
Sione Palu
 
PPTX
IP_Journal_Articles_2025IP_Journal_Articles_2025
mishell212144
 
PPTX
Insurance-Analytics-Branch-Dashboard (1).pptx
trivenisapate02
 
PDF
blockchain123456789012345678901234567890
tanvikhunt1003
 
PPTX
Multiscale Segmentation of Survey Respondents: Seeing the Trees and the Fores...
Sione Palu
 
PPTX
UVA-Ortho-PPT-Final-1.pptx Data analytics relevant to the top
chinnusindhu1
 
PPTX
7 Easy Ways to Improve Clarity in Your BI Reports
sophiegracewriter
 
PDF
202501214233242351219 QASS Session 2.pdf
lauramejiamillan
 
PPTX
Pipeline Automatic Leak Detection for Water Distribution Systems
Sione Palu
 
PPTX
Probability systematic sampling methods.pptx
PrakashRajput19
 
PDF
SUMMER INTERNSHIP REPORT[1] (AutoRecovered) (6) (1).pdf
pandeydiksha814
 
PDF
apidays Munich 2025 - The Double Life of the API Product Manager, Emmanuel Pa...
apidays
 
PDF
202501214233242351219 QASS Session 2.pdf
lauramejiamillan
 
Introduction to Data Analytics and Data Science
KavithaCIT
 
From Vision to Reality: The Digital India Revolution
Harsh Bharvadiya
 
The whitetiger novel review for collegeassignment.pptx
DhruvPatel754154
 
M1-T1.pptxM1-T1.pptxM1-T1.pptxM1-T1.pptx
teodoroferiarevanojr
 
717629748-Databricks-Certified-Data-Engineer-Professional-Dumps-by-Ball-21-03...
pedelli41
 
Data-Driven Machine Learning for Rail Infrastructure Health Monitoring
Sione Palu
 
Data-Users-in-Database-Management-Systems (1).pptx
dharmik832021
 
Customer Segmentation: Seeing the Trees and the Forest Simultaneously
Sione Palu
 
IP_Journal_Articles_2025IP_Journal_Articles_2025
mishell212144
 
Insurance-Analytics-Branch-Dashboard (1).pptx
trivenisapate02
 
blockchain123456789012345678901234567890
tanvikhunt1003
 
Multiscale Segmentation of Survey Respondents: Seeing the Trees and the Fores...
Sione Palu
 
UVA-Ortho-PPT-Final-1.pptx Data analytics relevant to the top
chinnusindhu1
 
7 Easy Ways to Improve Clarity in Your BI Reports
sophiegracewriter
 
202501214233242351219 QASS Session 2.pdf
lauramejiamillan
 
Pipeline Automatic Leak Detection for Water Distribution Systems
Sione Palu
 
Probability systematic sampling methods.pptx
PrakashRajput19
 
SUMMER INTERNSHIP REPORT[1] (AutoRecovered) (6) (1).pdf
pandeydiksha814
 
apidays Munich 2025 - The Double Life of the API Product Manager, Emmanuel Pa...
apidays
 
202501214233242351219 QASS Session 2.pdf
lauramejiamillan
 

Quantum persistent k cores for community detection

  • 1. Colleen Farrelly and Uchenna Chukwu
  • 2.  Discerning community structure and community strength within network data is a major focus in network analytics.  Connected brain regions related to different functions  Social ties  Subscale identification within a psychometric measure  Many approaches exist to either quantify strength of communities or cluster network data:  Louvain clustering and Louvain modularity  Random walk clustering  However, existing methods don’t provide evolving slices of community structure, which can delineate the strength of ties within communities and the point at which individuals join and drop out of those communities.  Our approach, which blends filtration methods, quantum computing, and k-core algorithms, can provide those evolving slices.
  • 3.  There is a well-known link between the graph coloring problem and k-core algorithms.  The graph coloring problem requires finding the minimum number of colors required to color each vertex of the graph such that no two vertices have the same color.  This solution gives the chromatic number, or the minimum number of colors needed to solve the problem for a given graph.  Call the chromatic number, k.  If the vertices of a given graph that necessitate the use of k colors, rather than k-1 colors, are removed, the chromatic number can drop to k-1 for the graph, referred to as degeneracy of the graph.  Decomposing a graph in this manner gives the shells of a graph that require a given number of colors.  This produces a series of connected graphs based on chromatic number degeneracy of the graph, called the k-cores of the graph.  K-core algorithms are common tools in community detection problems, as well as visualization problems and the coloring problem itself.
  • 4.  Graph filtrations have become relatively commonplace in the topological analysis of network data, particularly within brain imaging and connectomics.  Given a weighted graph, iteratively set higher and higher edge weights to 0 and analyze each “slice” of the graph for local or global network properties.  This is a lower-dimensional analogue of filtrations used in persistent homology; rather than filter on distance to build simplicial complexes, filter links directly within a graph (1-skeleton of potentially larger simplicial complex).  This gives evolving slices of graph metrics. Filtration set to 0.2 edge weight
  • 5.  Two solutions exist to the graph coloring problem on qubit-based quantum computers:  Simulated annealing solutions (mainly for D-Wave’s quantum computer)  Gate-based solutions (mainly for Rigetti’s and IBM’s quantum computers)  Use the D-Wave extension of networkx Python package, which relies on QUBO formulation of the coloring algorithm on a given graph  QUBO formulation expands Ising model into linear and quadratic terms for optimization.  Tools from the Ocean package’s nextworkx extension to do this automatically, like the min_vertex_coloring function.  The DWaveSampler simulates the D-Wave quantum computing set up, allowing the code to run on a simulated quantum computer based on simulated annealing.  Post-processing of the raw results yields the minimum coloring solution.
  • 6.  Trial graphs:  3 toy graphs with 6 nodes and varying connectivity patterns and link weights  The KONECT Windsurfer network, which is a weighted network with 43 vertices  Community-finding algorithms compared with quantum persistent k-cores:  Louvain clustering (igraph R package)  Walktrap clustering (igraph R package)  Spinglass clustering (igraph R package)  Hierarchical clustering (hclust function in R)  Graph filtration:  Comparison algorithms were run with and without filtration to compare to standard use of algorithms in community-finding algorithms and their persistence-based analogues.  Results were also compared to persistent k-core results for benchmarking of quantum persistent k-core algorithm.  Filtration values were set to 0.2, 0.4, 0.6, 0.8, and 1, respectively.
  • 8.  Clustering methods generally agree across methods for all 4 weighted graphs.  Graph 1 presented some issues to Walktrap community detection.  Others were consistent across algorithms.  This suggests that the true clustering on these graphs is the solution found by the classical community detection algorithm.
  • 9.  Graph filtration yielded some interesting results related to the decomposition of the network, particularly within the Windsurfer dataset.  Left figure is unfiltered result with Louvain clustering.  With a filtration value of 0.6 (anything below this set to 0; right figure), ties begin to separate into isolated groups and smaller, interconnected clusters.
  • 10.  Filtration yielded a series of k- core values for each vertex in the 4 networks, tracking changes in degeneracy across each graph.  Table 1 (above) shows the filtration for graph 1; vertices 1, 3, and 4 form links between the two cluster groups found, with vertex 4 showing a strong connection to distal nodes in its own group.  Table 3 (below) shows the filtration for graph 3; vertices 1, 2, and 3 have strong intergroup connections. Filtration Value V1 V2 V3 V4 V5 V6 0.2 2 1 2 2 1 1 0.4 2 1 2 2 1 1 0.6 1 1 1 1 1 1 0.8 0 1 1 1 1 0 1 0 0 0 0 0 0 Filtration Value V1 V2 V3 V4 V5 V6 0.2 3 3 3 3 3 3 0.4 2 2 2 2 1 2 0.6 2 2 2 0 1 1 0.8 1 1 1 0 0 1 1 0 0 0 0 0 0
  • 11.  The quantum persistent k-cores/coloring algorithm was expected to need k+1 colors to find an optimal coloring solution to each filtration of each graph.  For the 3 toy graphs, this matched the persistent k-cores results exactly.  Each coloring resulted in a graph with most likely solution conforming to the coloring problem solution with k+1 colors needed for the best solution.  For graphs with only isolated vertices, the algorithm correctly identified that no nodes were connected in the graphs.  This confirms that the quantum persistent k-cores algorithm can both find the optimal coloring and distinguish how many colors are needed within the connected communities, which corresponded correctly in all toy graph cases.  The surfer network posed computational problems for the algorithm, suggesting that this large of a coloring/k-cores problem is not feasible for the current systems.
  • 12.  These experiments demonstrate the usefulness of graph filtration in community detection problems.  Common community detection algorithms, such as Louvain clustering or walktrap clustering, can find structure and subgroups within the graph in a more nuanced fashion when combined with graph filtration, providing an evolution of subgroup separation.  Filtrations also illuminate how graph degeneracy and, hence, communities structures evolve by edge weight and strength of vertex connections.  Quantum coloring algorithms probabilistically converge to known solutions from k-core algorithms, suggesting that quantum algorithms are viable for these problems and do not degrade in performance under extreme filtrations.  It is hoped that these tools will provide the network analytics and graph theory community with a starting point of persistent graph tools for community finding and subgroup mining.
  • 13.  Lee, H., Chung, M. K., Kang, H., Kim, B. N., & Lee, D. S. (2011, September). Computing the shape of brain networks using graph filtration and Gromov-Hausdorff metric. In International Conference on Medical Image Computing and Computer- Assisted Intervention (pp. 302-309). Springer, Berlin, Heidelberg.  Farrelly, C. M., Schwartz, S. J., Amodeo, A. L., Feaster, D. J., Steinley, D. L., Meca, A., & Picariello, S. (2017). The analysis of bridging constructs with hierarchical clustering methods: An application to identity. Journal of Research in Personality, 70, 93-106.  Hahm, J., Lee, H., Park, H., Kang, E., Kim, Y. K., Chung, C. K., ... & Lee, D. S. (2017). Gating of memory encoding of time- delayed cross-frequency MEG networks revealed by graph filtration based on persistent homology. Scientific reports, 7, 41592.  Hübner, K. L. Coloring Complex Networks.  Kageura, K., & Abekawa, T. (2007). Modelling and exploring the network structure of terminology using the Potts spin glass model. In Proceedings of the 10th Conference of the Pacific Association for the Computational Linguistics (PACLING) (pp. 236- 245).  Radzikowski, J., Stefanidis, A., Jacobsen, K. H., Croitoru, A., Crooks, A., & Delamater, P. L. (2016). The measles vaccination narrative in Twitter: a quantitative analysis. JMIR public health and surveillance, 2(1).  Rider, A. K., Siwo, G., Emrich, S. J., Ferdig, M. T., & Chawla, N. V. (2014). A supervised learning approach to the ensemble clustering of genes. International journal of data mining and bioinformatics, 9(2), 199-219.  Rossi, R. A., & Ahmed, N. K. (2014). Coloring large complex networks. Social Network Analysis and Mining, 4(1), 228.  Shaydulin, R., Ushijima-Mwesigwa, H., Safro, I., Mniszewski, S., & Alexeev, Y. (2018). Community Detection Across Emerging Quantum Architectures. arXiv preprint arXiv:1810.07765.  Shepitsen, A., Gemmell, J., Mobasher, B., & Burke, R. (2008, October). Personalized recommendation in social tagging systems using hierarchical clustering. In Proceedings of the 2008 ACM conference on Recommender systems (pp. 259-266). ACM.  Titiloye, O., & Crispin, A. (2011). Quantum annealing of the graph coloring problem. Discrete Optimization, 8(2), 376-384.  Ushijima-Mwesigwa, H., Negre, C. F., & Mniszewski, S. M. (2017, November). Graph partitioning using quantum annealing on the D-Wave system. In Proceedings of the Second International Workshop on Post Moores Era Supercomputing (pp. 22-29). ACM.  Wang, Y., & Perkowski, M. (2011, May). Improved complexity of quantum oracles for ternary grover algorithm for graph coloring. In Multiple-Valued Logic (ISMVL), 2011 41st IEEE International Symposium on (pp. 294-301). IEEE.