pith. sign in

arxiv: 2006.05378 · v2 · submitted 2020-06-09 · 🧮 math.OC

A Graph-Based Modeling Abstraction for Optimization: Concepts and Implementation in Plasmo.jl

Jordan Jalving , Sungho Shin , Victor M. Zavala This is my paper

Pith reviewed 2026-05-24 14:19 UTC · model grok-4.3

classification 🧮 math.OC
keywords optimization modelinggraph abstractionhypergraphdecomposition algorithmsmodular modelingPlasmo.jllarge-scale optimizationJulia package
0
0 comments X

The pith

Any optimization problem can be modeled as a hierarchical hypergraph of subproblems connected by edges.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper introduces an abstraction called OptiGraph that represents optimization problems as hierarchical hypergraphs. Nodes stand for subproblems and edges capture how those subproblems connect. The goal is to let users build large models by combining smaller pieces in a natural way. The same graph structure then supports standard graph tools for partitioning or visualization and supplies structure information directly to decomposition solvers. An implementation appears in the Plasmo.jl package with examples and case studies.

Core claim

Under this abstraction, any optimization problem is treated as a hierarchical hypergraph in which nodes represent optimization subproblems and edges represent connectivity between such subproblems. The abstraction enables the modular construction of highly complex models in an intuitive manner, facilitates the use of graph analysis tools to perform partitioning, aggregation, and visualization tasks, and facilitates communication of structures to decomposition algorithms.

What carries the argument

The OptiGraph, a hierarchical hypergraph in which nodes are subproblems and edges are their connections, which organizes the model for modular assembly and direct use by graph algorithms and decomposers.

If this is right

  • Complex models can be assembled from reusable subproblem modules rather than written as single monolithic blocks.
  • Standard graph algorithms become available for tasks such as automatic partitioning and visualization of the optimization structure.
  • Decomposition methods receive explicit connectivity information from the hypergraph without additional manual encoding.
  • The same model representation supports both construction and algorithmic exploitation in one framework.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The hypergraph view could be combined with existing open graph libraries to import advanced analysis routines without new implementation.
  • Automatic translation between the OptiGraph and other modeling languages might allow existing models to gain the modular and decomposition benefits.
  • The structure might support new debugging tools that highlight connectivity issues directly on the graph rather than in equation lists.

Load-bearing premise

Representing an optimization problem as this hypergraph must keep every piece of necessary mathematical information intact and deliver net practical gains in construction and solution that exceed any added modeling cost.

What would settle it

A side-by-side test on a large-scale model showing that the hypergraph version takes longer to build or produces slower or less accurate solutions than a conventional formulation without the graph layer would show the abstraction does not deliver the claimed advantages.

Figures

Figures reproduced from arXiv: 2006.05378 by Jordan Jalving, Sungho Shin, Victor M. Zavala.

Figure 1
Figure 1. Figure 1: Natural gas transmission network and possible partitioning strategies. The network lay [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Space and space-time unfolding of natural gas network (uncovering components). [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Network topology of coupled natural gas (in blue) and electric (in red) networks (left), space [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: OptiGraph with three OptiNodes connected by four OptiEdges . 2.2 Syntax and Usage We now introduce the Plasmo.jl implementation of the OptiGraph and show how to use its syntax to create, analyze, and solve an optimization model. Plasmo.jl implements the graph model described by (2.1) and illustrated in [PITH_FULL_IMAGE:figures/full_fig_p009_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Output visuals for Code Snippet 1. Graph topology obtained with plot function (left) and graph matrix representation obtained with spy function (right) [PITH_FULL_IMAGE:figures/full_fig_p011_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: An OptiGraph that contains three subgraphs. The subgraphs are coupled through the global edge e0 in OptiGraph G. 2.3.1 Example 2: Hierarchical Graph with Global Edge An OptiGraph object manages its own nodes and edge in a self-contained manner (without re￾quiring references to other higher-level graphs). Consequently, we can define subgraphs indepen￾dently (in a modular manner) and these can be coupled tog… view at source ↗
Figure 7
Figure 7. Figure 7: An OptiGraph that contains three subgraphs. The subgraphs are coupled to the global node n0 in OptiGraph G. defined on a high-level graph. This feature is key because each node can have its own syntax (syn￾tax does not need to be consistent or non-redundant over the entire model). This is a fundamental difference with algebraic modeling languages (syntax has to be consistent and non-redundant over the enti… view at source ↗
Figure 8
Figure 8. Figure 8: Output visuals for Code Snippet 2 showing hierarchical structure of the OptiGraph with three subgraphs [PITH_FULL_IMAGE:figures/full_fig_p015_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Output visuals for Code Snippet 3 showing hierarchical structure of the OptiGraph with three subgraphs. an OptiGraph. We inspect the nodes, edges, linking constraints, and subgraphs using getoptinodes, getoptiedges, getlinkconstraints, and getsubgraphs functions, and we can collect all of the corresponding graph attributes using recursive versions of these functions (all nodes, all edges, all linkconstrain… view at source ↗
Figure 10
Figure 10. Figure 10: Typical graph representations used in partitioning tools. A hypergraph can be projected to [PITH_FULL_IMAGE:figures/full_fig_p018_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Core partitioning functionality in Plasmo.jl. (left) A Partition with nine OptiNodes, (middle) corresponding OptiGraph containing three subgraphs, and (right) subgraphs aggre￾gated into OptiNodes. (a) incident edges (b) neighborhood (c) expand [PITH_FULL_IMAGE:figures/full_fig_p019_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: Core graph topology functions in Plasmo.jl. (left) querying incident edges to a sub￾graph, (middle) querying a subgraph neighborhood, and (right) expanding a subgraph. picts core topology functions we commonly use in the framework. We can query the incident OptiEdges (Figure 12a) to a set of OptiNodes (or a subgraph) to inspect coupling (i.e. inspect incident linking constraints). We can also obtain the n… view at source ↗
Figure 13
Figure 13. Figure 13: Output visuals for Code Snippet 4 showing graph structure of dynamic optimization problem [PITH_FULL_IMAGE:figures/full_fig_p021_13.png] view at source ↗
Figure 14
Figure 14. Figure 14: Output visuals for Code Snippet 5 showing partitions and reordering of dynamic optimization problem. well-balanced and note that the matrix is now rearranged into a banded structure that is typical of dynamic optimization problems (partitioning automatically induces reordering). Plasmo.jl can use other graph representations to perform partitioning. For instance, we can create a traditional graph represent… view at source ↗
Figure 15
Figure 15. Figure 15: Output visuals for Code Snippet 7 showing aggregated graph of dynamic optimization problem [PITH_FULL_IMAGE:figures/full_fig_p024_15.png] view at source ↗
Figure 16
Figure 16. Figure 16: Output visuals for Code Snippet 8 showing overlapping subgraphs [PITH_FULL_IMAGE:figures/full_fig_p025_16.png] view at source ↗
Figure 17
Figure 17. Figure 17: Block structure (left) and nested block structure (right) of an [PITH_FULL_IMAGE:figures/full_fig_p026_17.png] view at source ↗
Figure 18
Figure 18. Figure 18: Depiction of Schwarz Algorithm. The original graph containing two subgraphs ( [PITH_FULL_IMAGE:figures/full_fig_p031_18.png] view at source ↗
Figure 19
Figure 19. Figure 19: Graph layouts of the natural gas network problem. The graph is colored by the physical [PITH_FULL_IMAGE:figures/full_fig_p036_19.png] view at source ↗
Figure 20
Figure 20. Figure 20: Partitioning results with 48 partitions. Imbalance versus the number of linking constraints [PITH_FULL_IMAGE:figures/full_fig_p038_20.png] view at source ↗
Figure 21
Figure 21. Figure 21: Depiction of DC OPF problem with four partitions. The original calculated partitions with [PITH_FULL_IMAGE:figures/full_fig_p041_21.png] view at source ↗
Figure 22
Figure 22. Figure 22: Comparison of Schwarz algorithm for different values of overlap [PITH_FULL_IMAGE:figures/full_fig_p041_22.png] view at source ↗
read the original abstract

We present a general graph-based modeling abstraction for optimization that we call an OptiGraph. Under this abstraction, any optimization problem is treated as a hierarchical hypergraph in which nodes represent optimization subproblems and edges represent connectivity between such subproblems. The abstraction enables the modular construction of highly complex models in an intuitive manner, facilitates the use of graph analysis tools (to perform partitioning, aggregation, and visualization tasks), and facilitates communication of structures to decomposition algorithms. We provide an open-source implementation of the abstraction in the Julia-based package Plasmo.jl. We provide tutorial examples and large application case studies to illustrate the capabilities.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 4 minor

Summary. The paper introduces the OptiGraph abstraction, under which any optimization problem is represented as a hierarchical hypergraph with nodes as subproblems and hyperedges as connectivity relations. This enables modular model construction, application of graph analysis tools for partitioning and visualization, and direct interfacing with decomposition algorithms. The work provides an open-source Julia implementation in Plasmo.jl together with tutorial examples and large-scale case studies demonstrating the mapping in practice.

Significance. If the abstraction preserves mathematical structure without prohibitive overhead, it would offer a practical advance in modeling complex, structured optimization problems by unifying graph-theoretic tools with algebraic modeling. The open-source package, tutorial examples, and case studies constitute concrete, reproducible contributions that lower the barrier to adopting hierarchical and graph-based decomposition strategies.

major comments (2)
  1. [§4] §4 (case studies): the demonstrations show that models can be constructed and solved via the OptiGraph interface, but no quantitative comparison of modeling effort, memory footprint, or solve time versus a direct JuMP or Pyomo formulation is reported; this leaves the claim that advantages exceed overhead unverified for the largest instances.
  2. [§2.1] §2.1 (OptiGraph definition): the statement that the hypergraph representation 'preserves all necessary mathematical information' is asserted without an explicit equivalence theorem or invariant that maps the original objective, constraints, and variable domains onto the node/edge data structures; a short formal statement would make the central claim load-bearing rather than descriptive.
minor comments (4)
  1. [§2] Notation for hyperedges versus ordinary edges is introduced without a consistent typographic distinction (e.g., bold versus script font) that persists through the implementation section.
  2. [Figure 3] Figure 3 (partitioning example) lacks axis labels and a legend indicating which colors correspond to which subgraphs, reducing readability for readers unfamiliar with the package.
  3. [§1] The related-work paragraph cites only a handful of decomposition frameworks; adding references to graph-based modeling libraries (e.g., Plasmo predecessors or NetworkX-based optimization interfaces) would better situate the contribution.
  4. [§3] Several code snippets in the tutorial section use inline comments that duplicate the surrounding prose; condensing them would improve conciseness.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive comments and positive recommendation. We address the two major comments below and will incorporate revisions accordingly.

read point-by-point responses

  1. Referee: [§4] §4 (case studies): the demonstrations show that models can be constructed and solved via the OptiGraph interface, but no quantitative comparison of modeling effort, memory footprint, or solve time versus a direct JuMP or Pyomo formulation is reported; this leaves the claim that advantages exceed overhead unverified for the largest instances.

    Authors: We agree that explicit quantitative comparisons would make the practical advantages clearer. In the revised manuscript we will add a short subsection in §4 reporting modeling effort (measured in lines of code) and memory footprint for the smallest case study against an equivalent direct JuMP formulation. For the largest instances we will add a brief discussion noting that the OptiGraph layer introduces only a thin wrapper with negligible overhead on the underlying JuMP models; a full timing comparison on the largest instances is outside the scope of the current work but can be pursued in follow-up studies. revision: yes

  2. Referee: [§2.1] §2.1 (OptiGraph definition): the statement that the hypergraph representation 'preserves all necessary mathematical information' is asserted without an explicit equivalence theorem or invariant that maps the original objective, constraints, and variable domains onto the node/edge data structures; a short formal statement would make the central claim load-bearing rather than descriptive.

    Authors: We accept this observation. In the revised §2.1 we will insert a concise formal statement (one paragraph) that defines the mapping: each node stores its local objective, constraints, and variable domains; each hyperedge stores the linking constraints and shared variables; and the overall problem is recovered by taking the union of all node objectives and constraints together with the linking constraints on the hyperedges. This makes the preservation claim explicit without altering the surrounding exposition. revision: yes

Circularity Check

0 steps flagged

No significant circularity; modeling abstraction is self-contained

full rationale

The paper introduces an OptiGraph abstraction for representing optimization problems as hierarchical hypergraphs, supplies a Julia implementation in Plasmo.jl, and demonstrates it via tutorials and case studies. No derivation chain exists that reduces a claimed result to fitted inputs, self-definitions, or self-citation load-bearing premises. The central claim is a modeling proposal whose validity rests on practical utility shown through code and examples rather than any mathematical reduction to prior inputs. This matches the default expectation of no circularity for papers presenting new abstractions and implementations.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 1 invented entities

The central claim rests on the utility of the newly defined OptiGraph entity together with standard background assumptions from graph theory and optimization; no free parameters are fitted and no ad-hoc axioms beyond domain standards are invoked.

axioms (1)
  • standard math Standard graph-theoretic concepts (hypergraphs, hierarchical nesting, connectivity via edges) apply directly to representing optimization subproblem structure.
    The abstraction is built on established definitions of graphs and hypergraphs without additional unproved lemmas.
invented entities (1)
  • OptiGraph no independent evidence
    purpose: Core modeling abstraction that represents any optimization problem as a hierarchical hypergraph of subproblems
    Newly introduced entity whose practical value is asserted by the paper; no independent falsifiable evidence outside the implementation is supplied.

pith-pipeline@v0.9.0 · 5632 in / 1389 out tokens · 49729 ms · 2026-05-24T14:19:12.077456+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Graph-Based Modeling and Decomposition of Energy Infrastructures

    math.OC 2020-10 accept novelty 7.0

    Graph-based modeling with restricted additive Schwarz decomposition in interior-point methods accelerates transient gas network optimization and multi-period AC optimal power flow by over 300%.

  2. A Julia Framework for Graph-Structured Nonlinear Optimization

    math.OC 2022-04 unverdicted novelty 4.0

    A Julia framework combines Plasmo.jl and MadNLP.jl to model and solve graph-structured nonlinear optimization problems, demonstrated on a large stochastic gas network instance with over 1.7 million variables.

Reference graph

Works this paper leans on

85 extracted references · 85 canonical work pages · cited by 2 Pith papers · 1 internal anchor

  1. [1]

    PETSc/TS: A Modern Scalable ODE/DAE Solver Library

    A BHYANKAR , S., B ROWN , J., C ONSTANTINESCU , E. M., G HOSH , D., S MITH , B. F., AND ZHANG , H. Petsc/ts: A modern scalable ode/dae solver library. arXiv preprint arXiv:1806.01437 (2018)

    work page internal anchor Pith review Pith/arXiv arXiv 2018

  2. [2]

    Towards a Generic Algorithm for Identifying High-Quality Decompo- sitions of Optimization Problems

    A LLMAN , A., T ANG , W., AND DAOUTIDIS , P. Towards a Generic Algorithm for Identifying High-Quality Decompo- sitions of Optimization Problems. In Computer Aided Chemical Engineering. 2018

    work page 2018

  3. [3]

    Decode: a community-based algorithm for generating high-quality decompositions of optimization problems

    A LLMAN , A., T ANG , W., AND DAOUTIDIS , P. Decode: a community-based algorithm for generating high-quality decompositions of optimization problems. Optimization and Engineering (06 2019)

    work page 2019

  4. [4]

    D., C OFFRIN , C., D EMARCO , C., D IAO , R., F ER- RIS , M., F LISCOUNAKIS , S., G REENE , S., H UANG , R., J OSZ , C., K ORAB , R., L ESIEUTRE , B., M AEGHT , J., M OLZAHN , D

    B ABAEINEJADSAROOKOLAEE , S., B IRCHFIELD , A., C HRISTIE , R. D., C OFFRIN , C., D EMARCO , C., D IAO , R., F ER- RIS , M., F LISCOUNAKIS , S., G REENE , S., H UANG , R., J OSZ , C., K ORAB , R., L ESIEUTRE , B., M AEGHT , J., M OLZAHN , D. K., O VERBYE , T. J., P ANCIATICI , P., P ARK , B., S NODGRASS , J., AND ZIMMERMAN , R. The power grid library for ...

    work page 2019

  5. [5]

    Gephi: An open source software for exploring and manipulating networks, 2009

    B ASTIAN , M., H EYMANN , S., AND JACOMY , M. Gephi: An open source software for exploring and manipulating networks, 2009

    work page 2009

  6. [6]

    E., M ALAGUTI , E., AND TRAVERSI , E

    B ERGNER , M., C APRARA , A., C ESELLI , A., F URINI , F., L ¨UBBECKE , M. E., M ALAGUTI , E., AND TRAVERSI , E. Auto- matic dantzig–wolfe reformulation of mixed integer programs. Mathematical Programming 149, 1 (feb 2015), 391–424

    work page 2015

  7. [7]

    Efficient stochastic programming in julia, 2019

    B IEL , M., AND JOHANSSON , M. Efficient stochastic programming in julia, 2019

    work page 2019

  8. [8]

    Distributed optimization and statistical learning via the alternating direction method of multipliers

    B OYD , S., P ARIKH , N., C HU, E., P ELEATO , B., AND ECKSTEIN , J. Distributed optimization and statistical learning via the alternating direction method of multipliers. Found. Trends Mach. Learn. 3, 1 (Jan. 2011), 1122

    work page 2011

  9. [9]

    B RUNAUD , B., AND GROSSMANN , I. E. Perspectives in multilevel decision-making in the process industry. Frontiers of Engineering Management 4, 3 (2017), 1–34

    work page 2017

  10. [10]

    F., C HEN , S., AND ZAVALA, V

    C AO, Y., F UENTES -C ORTES , L. F., C HEN , S., AND ZAVALA, V. M. Scalable modeling and solution of stochastic multi- objective optimization problems. Computers & Chemical Engineering 99 (2017), 185–197

    work page 2017

  11. [11]

    C AO, Y., AND ZAVALA, V. M. A scalable global optimization algorithm for stochastic nonlinear programs

    work page

  12. [12]

    PaToH (Partitioning Tool for Hypergraphs)

    C ¸ATALY ¨UREK , ¨U., AND AYKANAT , C. PaToH (Partitioning Tool for Hypergraphs) . Springer US, Boston, MA, 2011, pp. 1479–1487

    work page 2011

  13. [13]

    Y., AND ZAVALA, V

    C HIANG , N. Y., AND ZAVALA, V. M. Large-scale optimal control of interconnected natural gas and electrical trans- mission systems. Applied Energy 168 (2016), 226–235

    work page 2016

  14. [14]

    Powermodels.jl: An open-source framework for explor- ing power flow formulations

    C OFFRIN , C., B ENT, R., S UNDAR , K., N G, Y., AND LUBIN , M. Powermodels.jl: An open-source framework for explor- ing power flow formulations. In 2018 Power Systems Computation Conference (PSCC) (June 2018), pp. 1–8

    work page 2018

  15. [15]

    A structure-conveying modelling lan- guage for mathematical and stochastic programming

    C OLOMBO , M., G ROTHEY , A., H OGG , J., W OODSEND , K., AND GONDZIO , J. A structure-conveying modelling lan- guage for mathematical and stochastic programming. Mathematical Programming Computation (2009)

    work page 2009

  16. [16]

    J., C ASTILLO , E., M ´INGUEZ , R., AND GARC´IA-B ERTRAND , R

    C ONEJO , A. J., C ASTILLO , E., M ´INGUEZ , R., AND GARC´IA-B ERTRAND , R. Decomposition techniques in mathematical programming: Engineering and science applications. 2006

    work page 2006

  17. [17]

    D., B OMAN , E

    D EVINE , K. D., B OMAN , E. G., H EAPHY , R. T., B ISSELING , R. H., AND CATALYUREK , U. V. Parallel hypergraph partitioning for scientific computing. In 20th International Parallel and Distributed Processing Symposium, IPDPS 2006 (2006)

    work page 2006

  18. [18]

    W., AND BIEGLER , L

    D OWLING , A. W., AND BIEGLER , L. T. A framework for efficient large scale equation-oriented flowsheet optimization. Computers & Chemical Engineering 72 (2015), 3–20

    work page 2015

  19. [19]

    Jump: A modeling language for mathematical optimization

    D UNNING , I., H UCHETTE , J., AND LUBIN , M. Jump: A modeling language for mathematical optimization. SIAM Review 59, 2 (2017), 295–320

    work page 2017

  20. [20]

    Disropt : a python framework

    F ARINA , F., C AMISA , A., T ESTA , A., N OTARNICOLA , I., AND NOTARSTEFANO , G. Disropt : a python framework. 1–14

    work page

  21. [21]

    C., AND HORN , J

    F ERRIS , M. C., AND HORN , J. D. Partitioning mathematical programs for parallel solution. Mathematical Programming 80, 1 (jan 1998), 35–61

    work page 1998

  22. [22]

    F ISHER , M. L. APPLICATIONS ORIENTED GUIDE TO LAGRANGIAN RELAXATION. Interfaces (1985)

    work page 1985

  23. [23]

    M., AND KERNIGHAN , B

    F OURER , R., G AY, D. M., AND KERNIGHAN , B. AMPL: A Mathematical Programming Language. Springer-Verlag, Berlin, Heidelberg, 1989, p. 150151

    work page 1989

  24. [24]

    F ROMMER , A., AND SZYLD , D. B. An algebraic convergence theory for restricted additive schwarz methods using weighted max norms. SIAM Journal on Numerical Analysis 39, 2 (2002), 463–479

    work page 2002

  25. [25]

    Parallel Interior Point Solver for Structured Quadratic Programs: Application to Financial Planning Problems

    G ONDZIO , J., AND GROTHEY , A. Parallel Interior Point Solver for Structured Quadratic Programs: Application to Financial Planning Problems. J. Annals of Operations Research 152, 1 (2006), 319–339

    work page 2006

  26. [26]

    Parallel interior-point solver for structured linear programs

    G ONDZIO , J., AND SARKISSIAN , R. Parallel interior-point solver for structured linear programs. Mathematical Program- ming (2003)

    work page 2003

  27. [27]

    Global Optimization

    G ROSSMAN , I. Global Optimization. Springer, 2013

    work page 2013

  28. [28]

    G ROSSMANN , I. E. Advances in mathematical programming models for enterprise-wide optimization. Computers and Chemical Engineering 47 (2012), 2–18

    work page 2012

  29. [29]

    PSMG-A Parallel Structured Model Generator for Mathematical Programming

    G ROTHEY , A., AND QIANG , F. PSMG-A Parallel Structured Model Generator for Mathematical Programming. Work- ingpaper, Optimization Online, 2014. Graph-Based Modeling for Optimization 49

    work page 2014

  30. [30]

    Highly scalable parallel algorithms for sparse matrix factorization

    G UPTA , A., K ARYPIS , G., AND KUMAR , V. Highly scalable parallel algorithms for sparse matrix factorization. IEEE Transactions on Parallel and Distributed Systems 8, 5 (1997), 502–520

    work page 1997

  31. [31]

    SnapVX: A Network-Based Convex Optimization Solver

    H ALLAC , D., W ONG , C., D IAMOND , S., S OSIC , R., B OYD , S., AND LESKOVEC , J. SnapVX: A Network-Based Convex Optimization Solver. Journal of Machine Learning Research 0 (2017), 1–5

    work page 2017

  32. [32]

    E., L AIRD , C

    H ART, W. E., L AIRD , C. D., W ATSON , J.-P., W OODRUFF , D. L., H ACKEBEIL , G. A., N ICHOLSON , B. L., AND SIIROLA , J. D. Pyomo–optimization modeling in python, second ed., vol. 67. Springer Science & Business Media, 2017

    work page 2017

  33. [33]

    H EO, S., R ANGARAJAN , S., D AOUTIDIS , P., AND JOGWAR , S. S. Graph reduction of complex energy-integrated net- works: Process systems applications. AIChE Journal (2014)

    work page 2014

  34. [34]

    H IJAZI , H. L. Gravity : A mathematical modeling language for optimization and machine learning

    work page

  35. [35]

    H ¨UBNER , J., S CHMIDT , M., AND STEINBACH , M. C. Optimization techniques for tree-structured nonlinear problems. Computational Management Science (2020)

    work page 2020

  36. [36]

    Parallel algebraic modeling for stochastic optimization

    H UCHETTE , J., L UBIN , M., AND PETRA , C. Parallel algebraic modeling for stochastic optimization. In Proceedings of HPTCDL 2014: 1st Workshop for High Performance Technical Computing in Dynamic Languages - Held in Conjunction with SC 2014: The International Conference for High Performance Computing, Networking, Storage and Analysis (2014)

    work page 2014

  37. [37]

    J ALVING , J., A BHYANKAR , S., K IM, K., H ERELD , M., AND ZAVALA, V. M. A graph-based computational framework for simulation and optimisation of coupled infrastructure networks. IET Generation, Transmission & Distribution(2017), 1–14

    work page 2017

  38. [38]

    J ALVING , J., C AO, Y., AND ZAVALA, V. M. Graph-based modeling and simulation of complex systems. Computers & Chemical Engineering 125 (2019), 134–154

    work page 2019

  39. [39]

    J ALVING , J., AND ZAVALA, V. M. An optimization-based state estimation framework for large-scale natural gas net- works. Industrial & Engineering Chemistry Research 57, 17 (2018), 5966–5979

    work page 2018

  40. [40]

    X., H UANG , J., AND ZHANG , R

    J IANG , W., Q I, J., Y U, J. X., H UANG , J., AND ZHANG , R. HyperX: A Scalable Hypergraph Framework. IEEE Transac- tions on Knowledge and Data Engineering (2018)

    work page 2018

  41. [41]

    S., R ANGARAJAN , S., AND DAOUTIDIS , P

    J OGWAR , S. S., R ANGARAJAN , S., AND DAOUTIDIS , P. Reduction of complex energy-integrated process networks using graph theory. Computers and Chemical Engineering (2015)

    work page 2015

  42. [42]

    P., AND LAIRD , C

    K ANG , J., C AO, Y., WORD , D. P., AND LAIRD , C. D. An interior-point method for efficient solution of block-structured NLP problems using an implicit Schur-complement decomposition. Computers and Chemical Engineering (2014)

    work page 2014

  43. [43]

    D., AND ZAVALA, V

    K ANG , J., C HIANG , N., L AIRD , C. D., AND ZAVALA, V. M. Nonlinear programming strategies on high-performance computers. In Decision and Control (CDC), 2015 IEEE 54th Annual Conference on (2015), IEEE, pp. 4612–4620

    work page 2015

  44. [44]

    A Fast and High Quality Multilevel Scheme for Partitioning Irregular Graphs

    K ARYPIS , G., AND KUMAR , V. A Fast and High Quality Multilevel Scheme for Partitioning Irregular Graphs. SIAM Journal on Scientific Computing 20, 1 (1998), 359–392

    work page 1998

  45. [45]

    Multilevel k-way hypergraph partitioning

    K ARYPIS , G., AND KUMAR , V. Multilevel k-way hypergraph partitioning. In Proceedings of the 36th Annual ACM/IEEE Design Automation Conference (New York, NY, USA, 1999), DAC ’99, ACM, pp. 343–348

    work page 1999

  46. [46]

    G., AND ZAVALA, V

    K IM, K., P ETRA , C. G., AND ZAVALA, V. M. An asynchronous bundle-trust-region method for dual decomposition of stochastic mixed-integer programming. SIAM Journal on Optimization 29, 1 (2019), 318–342

    work page 2019

  47. [47]

    K IM, K., AND ZAVALA, V. M. Algorithmic innovations and software for the dual decomposition method applied to stochastic mixed-integer programs. Mathematical Programming Computation (2018)

    work page 2018

  48. [48]

    Structure-Exploiting Interior Point Methods

    K OUROUNIS , D., AND SCHENK , O. Structure-Exploiting Interior Point Methods

    work page

  49. [49]

    Decomposition algorithms for stochastic programming on a computational grid

    L INDEROTH , J., AND WRIGHT , S. Decomposition algorithms for stochastic programming on a computational grid. Computational Optimization and Applications (2003)

    work page 2003

  50. [50]

    G., AND ANITESCU , M

    L UBIN , M., P ETRA , C. G., AND ANITESCU , M. The parallel solution of dense saddle-point linear systems arising in stochastic programming. Optimization Methods and Software (2012)

    work page 2012

  51. [51]

    GNU Linear Programming Kit Version 4.32.http://www.gnu.org/software/glpk/glpk.html, 2000–2012

    M AKHORIN , A. GNU Linear Programming Kit Version 4.32.http://www.gnu.org/software/glpk/glpk.html, 2000–2012

    work page 2000

  52. [52]

    M ARAVELIAS , C. T. General framework and modeling approach classification for chemical production scheduling. AIChE Journal (2012)

    work page 2012

  53. [53]

    E., E LMQVIST , H., AND OTTER , M

    M ATTSSON , S. E., E LMQVIST , H., AND OTTER , M. Physical system modeling with modelica. Control Engineering Practice 6, 4 (1998), 501–510

    work page 1998

  54. [54]

    Hype: Massive hypergraph partitioning with neighborhood expansion

    M AYER, C., M AYER, R., B HOWMIK , S., E PPLE , L., AND ROTHERMEL , K. Hype: Massive hypergraph partitioning with neighborhood expansion. 2018 IEEE International Conference on Big Data (Big Data) (2018), 458–467

    work page 2018

  55. [55]

    Graph representation and decomposition of ODE/hyperbolic PDE systems

    M OHARIR , M., K ANG , L., D AOUTIDIS , P., AND ALMANSOORI , A. Graph representation and decomposition of ODE/hyperbolic PDE systems. Computers and Chemical Engineering (2017)

    work page 2017

  56. [56]

    N EWMAN , M. E. Modularity and community structure in networks. Proceedings of the National Academy of Sciences of the United States of America (2006)

    work page 2006

  57. [57]

    Simulation of transient gas flows in networks

    O SIADACZ , A. Simulation of transient gas flows in networks. International Journal for Numerical Methods in Fluids 4 , 1 (1984), 13–24

    work page 1984

  58. [58]

    A., AND MARKOV , I

    P APA, D. A., AND MARKOV , I. L. Hypergraph partitioning and clustering. In Handbook of Approximation Algorithms and Metaheuristics. 2007. 50 Jordan Jalving, Sungho Shin, Victor M. Zavala

    work page 2007

  59. [59]

    Distillating knowledge about scotch

    P ELLEGRINI , F. Distillating knowledge about scotch. In Combinatorial Scientific Computing (Dagstuhl, Germany, 2009), U. Naumann, O. Schenk, H. D. Simon, and S. Toledo, Eds., no. 09061 in Dagstuhl Seminar Proceedings, Schloss Dagstuhl - Leibniz-Zentrum fuer Informatik, Germany

    work page 2009

  60. [60]

    V., W RIGHT , S

    R AO, C. V., W RIGHT , S. J., AND RAWLINGS , J. B. Application of interior-point methods to model predictive control. Journal of optimization theory and applications 99, 3 (1998), 723–757

    work page 1998

  61. [61]

    B., AND MAYNE, D

    R AWLINGS , J. B., AND MAYNE, D. Model predictive control: Theory and design, 2 ed. 2018

    work page 2018

  62. [62]

    M., K OCH , T., AND M ¨OST, D

    R EHFELDT , D., H OBBIE , H., S CH ¨ONHEIT , D., G LEIXNER , A. M., K OCH , T., AND M ¨OST, D. A massively parallel interior-point solver for linear energy system models with block structure

    work page

  63. [63]

    S., L AIRD , C

    R ODRIGUEZ , J. S., L AIRD , C. D., AND ZAVALA, V. M. Scalable preconditioning of block-structured linear algebra systems using ADMM. Computers and Chemical Engineering 133 (2020), 106478

    work page 2020

  64. [64]

    V., AND GROSSMANN , I

    S AHINIDIS , N. V., AND GROSSMANN , I. E. Convergence properties of generalized benders decomposition. Computers and Chemical Engineering (1991)

    work page 1991

  65. [65]

    Architectures for distributed and hierarchical Model Predictive Control - A review, 2009

    S CATTOLINI , R. Architectures for distributed and hierarchical Model Predictive Control - A review, 2009

    work page 2009

  66. [66]

    k-way hypergraph partition- ing via n-level recursive bisection

    S CHLAG , S., H ENNE , V., H EUER , T., M EYERHENKE , H., S ANDERS , P., AND SCHULZ , C. k-way hypergraph partition- ing via n-level recursive bisection. In 18th Workshop on Algorithm Engineering and Experiments, (ALENEX 2016) (2016), pp. 53–67

    work page 2016

  67. [67]

    Graph Partitioning for High-Performance Scientific Simulations

    S CHLOEGEL , K., K ARYPIS , G., AND KUMAR , V. Graph Partitioning for High-Performance Scientific Simulations. Morgan Kaufmann Publishers Inc., San Francisco, CA, USA, 2003, p. 491541

    work page 2003

  68. [68]

    K., H ESS , C., S TEIGER , C., T EICHMANN , M., J ACOB , J., B ERNARDES -LIMA , F., H ANGU , R., AND HAYRAPETYAN , S

    S CHULZ , C., B AYER, S. K., H ESS , C., S TEIGER , C., T EICHMANN , M., J ACOB , J., B ERNARDES -LIMA , F., H ANGU , R., AND HAYRAPETYAN , S. Course notes: Graph partitioning and graph clustering in theory and practice, 2015

    work page 2015

  69. [69]

    F., AND OTHER CONTRIBUTORS

    S ETH BROMBERGER , J. F., AND OTHER CONTRIBUTORS . Juliagraphs/lightgraphs.jl: an optimized graphs package for the julia programming language, 2017

    work page 2017

  70. [70]

    S HIN , S., A NITESCU , M., AND ZAVALA, V. M. Overlapping schwarz decomposition for constrained quadratic pro- grams, 2020

    work page 2020

  71. [71]

    S HIN , S., F AULWASSER , T., Z ANON , M., AND ZAVALA, V. M. A parallel decomposition scheme for solving long- horizon optimal control problems. In 2019 IEEE 58th Conference on Decision and Control (CDC) (2019), pp. 5264–5271

    work page 2019

  72. [72]

    S HIN , S., AND ZAVALA, V. M. Multi-grid schemes for multi-scale coordination of energy systems. In Energy Markets and Responsive Grids. Springer, 2018, pp. 195–222

    work page 2018

  73. [73]

    S HIN , S., AND ZAVALA, V. M. Multi-grid schemes for multi-scale coordination of energy systems. Energy Markets and Responsive Grids (2018)

    work page 2018

  74. [74]

    M., AND ANITESCU , M

    S HIN , S., Z AVALA, V. M., AND ANITESCU , M. Decentralized schemes with overlap for solving graph-structured optimization problems. IEEE Transactions on Control of Network Systems(2020)

    work page 2020

  75. [75]

    S TEINBACH , M. C. Tree-sparse convex programs. Mathematical Methods of Operations Research (2003)

    work page 2003

  76. [76]

    Dc optimal power flow formulation andsolution using quadprogj, 2010

    S UN, J., AND L, T. Dc optimal power flow formulation andsolution using quadprogj, 2010

    work page 2010

  77. [77]

    B., AND DAOUTIDIS , P

    T ANG , W., A LLMAN , A., P OURKARGAR , D. B., AND DAOUTIDIS , P. Optimal decomposition for distributed optimiza- tion in nonlinear model predictive control through community detection.Computers & Chemical Engineering 111(2017), 43–54

    work page 2017

  78. [78]

    Network decomposition for distributed control through community detection in inputoutput bipartite graphs

    T ANG , W., AND DAOUTIDIS , P. Network decomposition for distributed control through community detection in inputoutput bipartite graphs. Journal of Process Control (2018)

    work page 2018

  79. [79]

    W ACHTER , A., AND BIEGLER , L. T. On the implementation of an interior-point filter line-search algorithm for large- scale nonlinear programming. 25–57

    work page

  80. [80]

    Computational Experience with Hypergraph-Based Methods for Automatic Decomposi- tion in Discrete Optimization

    W ANG , J., AND RALPHS , T. Computational Experience with Hypergraph-Based Methods for Automatic Decomposi- tion in Discrete Optimization. In Integration of AI and OR Techniques in Constraint Programming for Combinatorial Opti- mization Problems (Berlin, Heidelberg, 2013), C. Gomes and M. Sellmann, Eds., Springer Berlin Heidelberg, pp. 394–402

    work page 2013

Showing first 80 references.