SlideShare a Scribd company logo
1 of 4
Download to read offline
Sparse Observability using LP Presolve and LT
DL Factorization in IMPL
(IMPL-SparseObservability)
i n d u s t r IAL g o r i t h m s LLC. (IAL)
www.industrialgorithms.com
August 2014
Introduction
Presented in this short document is a description of our technology we call “Sparse
Observability”. Observability is the estimatability metric (Bagajewicz, 2010) to structurally
determine that an unmeasured variable or regressed parameter is either uniquely solvable
(observable) or otherwise unsolvable (unobservable) in data reconciliation and regression
(DRR) applications. Ultimately, our purpose to use efficient sparse matrix techniques is to solve
large industrial DRR flowsheets quickly and accurately.
Most other implementations of observability calculation use dense linear algebra such as
reduced row echelon form (RREF), Gauss-Jordan decomposition (Crowe et. al. 1983; Madron
1992), QR factorization which can now be considered as semi-sparse (Swartz, 1989; Sanchez
and Romagnoli, 1996), Schur complements, Cholesky factorization (Kelly, 1998a) and singular
value decomposition (SVD) (Kelly, 1999). A sparse LU decomposition with complete-pivoting
from Albuquerque and Biegler (1996) for dynamic data reconciliation observability computation
was used but it is uncertain if complete-pivoting causes extreme “fill-ins” of the lower and upper
triangular matrices essentially making them near-dense. There is another sparse observability
method using an LP sub-solver found in Kelly and Zyngier (2008) but this requires solving as
many LP sub-problems as there are unmeasured variables which can be considered as
somewhat inefficient.
IMPL’s sparse observability technique uses the variable classification and nomenclature found
in Kelly (1998b) given that if we partition or separate the unmeasured variables into independent
(B12) and dependent (B34) sub-matrices then all dependent unmeasured variables by definition
are unobservable. If any independent unmeasured variable is a (linear) function of any
dependent variable then this independent variable is of course also unobservable because it is
dependent on another non-observable variable.
This can be easily seen using the following (linearized) DRR equation (Kelly, 1998b):
A * x + [B12 B34] * [y12T
y34T
]T
+ C * z = 0 or
A * x + B12 * y12 + B34 * y34 + C * z = 0
where x represents the measured variables in column vector form, y are the unmeasured
variables and z are the fixed variables or constants with corresponding sparse Jacobian
matrices A, B12, B34 and C respectively. By multiplying both sides by B12T and rearranging
we arrive at essentially the Normal equation found in typical ordinary least squares (OLS)
regression which is unique to the classification method of Kelly (1998b):
B12T
* B12 * y12 = -B12T
* (A * x + B34 * y34 + C * z) or
y12 = -(B12T
* B12)-1
* B12T
* (B34 * y34 + A * x + C * z)
Therefore, if (B12T
* B12)-1
* B12T
* B34 has any non-zero elements (relative to machine
precision) in any row then the corresponding independent unmeasured y12 variable is a linear
function of a non-observable unmeasured y34 variable.
Hence, the technology described below is an effective and reliable way to identify unobservable
unmeasured variables in large-scale and industrial-sized flowsheet DRR problems. It should be
mentioned that once the observability calculations have been performed, then the same
matrices can be used to identify redundant and non-redundant measured variables described in
Kelly (1998b). Redundant variables are observable if there measurement reading is missing,
deleted or removed from the data content of the DRR problem. Non-redundant variables are
unobservable if their measurement is no longer available given the remaining model and data of
the DRR problem. Obviously, it is important to have as much hardware (measurement
readings) and software (model relationships) redundancy in the system as possible given the
capital, operational and maintenance costs where this sensor network design optimization
problem is addressed in Kelly and Zyngier (2008).
Technology
IMPL’s sparse observability has two steps or parts: a presolve or preprocessing step to identify
essentially trivial observable (strongly independent) unmeasured variables and a LT
DL
factorization for sparse symmetric indefinite matrices on the remaining equations and
unmeasured variables to determine the dependent unmeasured variables.
Our DRR presolve is taken directly from primal LP presolve techniques (Andersen and
Andersen, 1995) and operates on the B sparse matrix (i.e., [B12 B34]) and to our knowledge,
we are the first to adopt primal LP presolving techniques to the application of identifying
observable unmeasured variables in DRR problems.
First, it involves removing singleton rows with a single column indicating that the column or
unmeasured variable is strongly observable because it is not a function of any other
unmeasured variable. And second, it removes doubleton rows of the form b * yi – b * yj = 0 and
one of the yi or yj unmeasured variables are removed because they are mirror images of one
another; thus reducing the number of rows and columns of B. These two operations are
repeated until there is no more improvement in reducing the rows and columns of B or there are
no rows or columns left in B.
These two straightforward primal LP presolve operations are very effective in practice and
usually shrink the B matrix considerably. As an example, Figure 1 shows the unmeasured
variable’s sparse matrix B taken from Crowe et. al. (1983)’s example #3.
Figure 1. Crowe et. al. (1983)’s Example #3.
All columns or unmeasured variables are observable and we determined these to be observable
using our LP presolving techniques. For instance, the first pass or iteration of our presolve finds
eight (8) singleton rows that can be removed i.e., “Node1 N2”, “Node1 Ar”, “Node2 N2”, “Node2
Ar”, “Node3 N2”, “Node3 H2”, “Node3 NH3”, “Splitter Constraints H2”. This means, that of the
total thirteen (13) unmeasured variables, 8 are determined to be strongly observable easily.
When these 8 rows and columns are removed a second pass further removes rows “Node2
NH3”, “Node2 H2”, “Node3 Ar” and “Node4 H2”. The third and final pass removes row “Node1
H2” and now completely identifies all columns as observable.
When not all unmeasured variables are declared to be observable using our presolve method,
we then need to proceed to identifying from the reduced or presolved set, which are
independent and dependent columns using either of the methods mentioned previously.
Instead, our approach in IMPL is to use the LT
DL (Cholesky-like) factorization for symmetric
indefinite sparse matrices from Intel’s Math Kernel Library (MKL) PARDISO (Schenk and
Gartner, 2006). LT
DL is similar to Cholesky factorization except that the former performs
symmetric and other types of pivoting required to precisely determine if a symmetric matrix is
singular (Hansen, 1987) where the later does not. More specifically, indefinite matrices have
both positive and negative eigenvalues and some can be zero (semi-definite) and these types of
indefinite matrices are found in non-convex and nonlinear interior-point optimization problems
such as those solved by IPOPT.
IMPL also supports locating independent and dependent unmeasured variables using a sparse
version of the modified Cholesky factorization from Kelly (1998a) and, even though it does not
employ any pivoting techniques, it is surprisingly accurate even for B matrices with large
condition numbers i.e., maximum singular value divided by its minimum singular value. And,
IMPL further supports a sparse LU decomposition technique with partial-pivoting to check for
near-zero diagonal elements of the upper triangular matrix (U). If Ui,i ~ 0 then this (usually)
indicates that column [B]i is dependent but is less accurate than the LT
DL and our modified
Cholesky.
In summary, when the LT
DL factorization in PARDISO is performed using parallel-processing
applied to the presolved BT
B matrix, this provides a very fast, accurate and robust technique to
identify observable and unobservable unmeasured variables in large-scale systems as well as
its complement of redundant and non-redundant measured variables. For nonlinear problems,
the A, B and C sparse matrices are linearized from the converged x and y solution values and
provide only “point or local estimates” of the observability and redundancy indicators.
References
Crowe, C.M., Garcia Campos, Y.A., Hrymak, A., “Reconciliation of process flowrates by matrix
projection, part 1: linear case”, American Institute of Chemical Engineering Journal, 29, 881-
888, (1983).
Hansen, P.C., “Detection of near-singularity in Cholesky and LDL factorizations”, Journal of
Computational and Applied Mathematics, 19, 293-299, (1987).
Swartz, C.L.E., “Data reconciliation for generalized flowsheet applications”, 197th
National Meeting, American Chemical Society (ACS), Dallas, Texas, (1989).
Madron, F., “Process plant performance: measurement and data processing for optimization
and retrofits”, Ellis Horwood, New York, NY, (1992).
Andersen, E.D., Andersen, K.D., “Presolving in linear programming”, Mathematical
Programming, 71, 221-245, (1995).
Albuquerque, J.S., Biegler, L.T., “Data reconciliation and gross-error detection for dynamic
systems”, American Institute of Chemical Engineering Journal, 42, 10, (1996).
Sanchez, M., Romagnoli, J., “Use of orthogonal transformations in data classification-
reconciliation”, Computers & Chemical Engineering, 20, 483-493, (1996).
Kelly, J.D., "On finding the matrix projection in the data reconciliation solution", Computers &
Chemical Engineering, 1553, (1998a).
Kelly, J.D., "A regularization approach to the reconciliation of constrained data sets", Computers
& Chemical Engineering, 1771, (1998b).
Kelly, J.D., "Reconciliation of process data using other projection matrices", Computers &
Chemical Engineering, 785, (1999).
Schenk, O., Gartner, K., “On fast factorization pivoting methods for sparse symmetric indefinite
systems”, Electronic Transactions on Numerical Analysis, 23, 158-179, (2006).
Kelly, J.D., Zyngier, D., "A new and improved MILP formulation to optimize observability,
redundancy and precision for sensor network problems", American Institute of Chemical
Engineering Journal, 54, 1282, (2008).
Bagajewicz, M., “Smart process plants: software and hardware for accurate data and profitable
operations”, McGraw-Hill (ISBN:978-0-07-160471), (2010).

More Related Content

What's hot

Mc0079 computer based optimization methods--phpapp02
Mc0079 computer based optimization methods--phpapp02Mc0079 computer based optimization methods--phpapp02
Mc0079 computer based optimization methods--phpapp02Rabby Bhatt
 
poster_kdd_2013_LRLR
poster_kdd_2013_LRLRposter_kdd_2013_LRLR
poster_kdd_2013_LRLRXiao Cai
 
Master of Computer Application (MCA) – Semester 4 MC0079
Master of Computer Application (MCA) – Semester 4  MC0079Master of Computer Application (MCA) – Semester 4  MC0079
Master of Computer Application (MCA) – Semester 4 MC0079Aravind NC
 
Additional Relational Algebra Operations
Additional Relational Algebra OperationsAdditional Relational Algebra Operations
Additional Relational Algebra OperationsA. S. M. Shafi
 
Relational Algebra & Calculus
Relational Algebra & CalculusRelational Algebra & Calculus
Relational Algebra & CalculusAbdullah Khosa
 
Dbms ii mca-ch5-ch6-relational algebra-2013
Dbms ii mca-ch5-ch6-relational algebra-2013Dbms ii mca-ch5-ch6-relational algebra-2013
Dbms ii mca-ch5-ch6-relational algebra-2013Prosanta Ghosh
 
5 the relational algebra and calculus
5 the relational algebra and calculus5 the relational algebra and calculus
5 the relational algebra and calculusKumar
 
Relational algebra calculus
Relational algebra  calculusRelational algebra  calculus
Relational algebra calculusVaibhav Kathuria
 
Lecture 06 relational algebra and calculus
Lecture 06 relational algebra and calculusLecture 06 relational algebra and calculus
Lecture 06 relational algebra and calculusemailharmeet
 
Design of second order linear time invariant systems for deadbeat response
Design of second order linear time invariant systems for deadbeat responseDesign of second order linear time invariant systems for deadbeat response
Design of second order linear time invariant systems for deadbeat responseIAEME Publication
 
A General Method for Estimating a Linear Structural Equation System
 A General Method for Estimating a Linear Structural Equation System A General Method for Estimating a Linear Structural Equation System
A General Method for Estimating a Linear Structural Equation SystemCheer Chain Enterprise Co., Ltd.
 
Relational algebra
Relational algebraRelational algebra
Relational algebrashynajain
 
Modeling the Broker Behavior Using a BDI Agent
Modeling the Broker Behavior Using a BDI AgentModeling the Broker Behavior Using a BDI Agent
Modeling the Broker Behavior Using a BDI Agentinfopapers
 
Relational algebra ppt
Relational algebra pptRelational algebra ppt
Relational algebra pptGirdharRatne
 

What's hot (18)

Mc0079 computer based optimization methods--phpapp02
Mc0079 computer based optimization methods--phpapp02Mc0079 computer based optimization methods--phpapp02
Mc0079 computer based optimization methods--phpapp02
 
poster_kdd_2013_LRLR
poster_kdd_2013_LRLRposter_kdd_2013_LRLR
poster_kdd_2013_LRLR
 
Master of Computer Application (MCA) – Semester 4 MC0079
Master of Computer Application (MCA) – Semester 4  MC0079Master of Computer Application (MCA) – Semester 4  MC0079
Master of Computer Application (MCA) – Semester 4 MC0079
 
Additional Relational Algebra Operations
Additional Relational Algebra OperationsAdditional Relational Algebra Operations
Additional Relational Algebra Operations
 
Relational Algebra & Calculus
Relational Algebra & CalculusRelational Algebra & Calculus
Relational Algebra & Calculus
 
Dbms ii mca-ch5-ch6-relational algebra-2013
Dbms ii mca-ch5-ch6-relational algebra-2013Dbms ii mca-ch5-ch6-relational algebra-2013
Dbms ii mca-ch5-ch6-relational algebra-2013
 
5 the relational algebra and calculus
5 the relational algebra and calculus5 the relational algebra and calculus
5 the relational algebra and calculus
 
Relational algebra calculus
Relational algebra  calculusRelational algebra  calculus
Relational algebra calculus
 
Paper473
Paper473Paper473
Paper473
 
Lecture 06 relational algebra and calculus
Lecture 06 relational algebra and calculusLecture 06 relational algebra and calculus
Lecture 06 relational algebra and calculus
 
Principa
PrincipaPrincipa
Principa
 
Principa
 Principa Principa
Principa
 
Chapter6
Chapter6Chapter6
Chapter6
 
Design of second order linear time invariant systems for deadbeat response
Design of second order linear time invariant systems for deadbeat responseDesign of second order linear time invariant systems for deadbeat response
Design of second order linear time invariant systems for deadbeat response
 
A General Method for Estimating a Linear Structural Equation System
 A General Method for Estimating a Linear Structural Equation System A General Method for Estimating a Linear Structural Equation System
A General Method for Estimating a Linear Structural Equation System
 
Relational algebra
Relational algebraRelational algebra
Relational algebra
 
Modeling the Broker Behavior Using a BDI Agent
Modeling the Broker Behavior Using a BDI AgentModeling the Broker Behavior Using a BDI Agent
Modeling the Broker Behavior Using a BDI Agent
 
Relational algebra ppt
Relational algebra pptRelational algebra ppt
Relational algebra ppt
 

Viewers also liked

Lu decomposition
Lu decompositionLu decomposition
Lu decompositiongilandio
 
Exp 3 (1)3. To Formulate YBUS Matrix By Singular Transformation.
Exp 3 (1)3.	To Formulate YBUS Matrix By Singular Transformation.Exp 3 (1)3.	To Formulate YBUS Matrix By Singular Transformation.
Exp 3 (1)3. To Formulate YBUS Matrix By Singular Transformation.Shweta Yadav
 
資料科學如何幫我們更瞭解捐款人?
資料科學如何幫我們更瞭解捐款人?資料科學如何幫我們更瞭解捐款人?
資料科學如何幫我們更瞭解捐款人?Sheng-Wei (Kuan-Ta) Chen
 
MATLAB programs Power System Simulation lab (Electrical Engineer)
MATLAB programs Power System Simulation  lab (Electrical Engineer)MATLAB programs Power System Simulation  lab (Electrical Engineer)
MATLAB programs Power System Simulation lab (Electrical Engineer)Mathankumar S
 
Power System Modelling And Simulation Lab
Power System Modelling And Simulation LabPower System Modelling And Simulation Lab
Power System Modelling And Simulation LabSachin Airan
 
Презентація команди Васильківської ЗОШ І-ІІІ ступенів №1
Презентація команди Васильківської ЗОШ І-ІІІ ступенів №1Презентація команди Васильківської ЗОШ І-ІІІ ступенів №1
Презентація команди Васильківської ЗОШ І-ІІІ ступенів №108600 Vasilkov
 
Hybrid Dynamic Simulation (HDS) Industrial Modeling Framework (HDS-IMF)
Hybrid Dynamic Simulation (HDS)  Industrial Modeling Framework (HDS-IMF)Hybrid Dynamic Simulation (HDS)  Industrial Modeling Framework (HDS-IMF)
Hybrid Dynamic Simulation (HDS) Industrial Modeling Framework (HDS-IMF)Alkis Vazacopoulos
 
Advanced Production Accounting of a Flotation Plant
Advanced Production Accounting of a Flotation PlantAdvanced Production Accounting of a Flotation Plant
Advanced Production Accounting of a Flotation PlantAlkis Vazacopoulos
 
Intro ed x k-ix norrköping
Intro ed x k-ix norrköpingIntro ed x k-ix norrköping
Intro ed x k-ix norrköpingCormac McGrath
 
Development of Communicative Competence Analysis
Development of Communicative Competence AnalysisDevelopment of Communicative Competence Analysis
Development of Communicative Competence AnalysisJoan Cahuich Perez
 
Презентация Васильківської ЗОШ І-ІІІ ст№9 "Вулиця Бурки"
Презентация Васильківської ЗОШ І-ІІІ ст№9 "Вулиця Бурки"Презентация Васильківської ЗОШ І-ІІІ ст№9 "Вулиця Бурки"
Презентация Васильківської ЗОШ І-ІІІ ст№9 "Вулиця Бурки"08600 Vasilkov
 
Endocrine system chapter 11
Endocrine system chapter 11Endocrine system chapter 11
Endocrine system chapter 11Kaylal88
 
Stock Decomposition Heuristic for Scheduling: A Priority Dispatch Rule Approach
Stock Decomposition Heuristic for Scheduling: A Priority Dispatch Rule ApproachStock Decomposition Heuristic for Scheduling: A Priority Dispatch Rule Approach
Stock Decomposition Heuristic for Scheduling: A Priority Dispatch Rule ApproachAlkis Vazacopoulos
 

Viewers also liked (20)

Meeting13
Meeting13Meeting13
Meeting13
 
Lu decomposition
Lu decompositionLu decomposition
Lu decomposition
 
Es272 ch4b
Es272 ch4bEs272 ch4b
Es272 ch4b
 
Exp 3 (1)3. To Formulate YBUS Matrix By Singular Transformation.
Exp 3 (1)3.	To Formulate YBUS Matrix By Singular Transformation.Exp 3 (1)3.	To Formulate YBUS Matrix By Singular Transformation.
Exp 3 (1)3. To Formulate YBUS Matrix By Singular Transformation.
 
資料科學如何幫我們更瞭解捐款人?
資料科學如何幫我們更瞭解捐款人?資料科學如何幫我們更瞭解捐款人?
資料科學如何幫我們更瞭解捐款人?
 
MATLAB programs Power System Simulation lab (Electrical Engineer)
MATLAB programs Power System Simulation  lab (Electrical Engineer)MATLAB programs Power System Simulation  lab (Electrical Engineer)
MATLAB programs Power System Simulation lab (Electrical Engineer)
 
Power System Modelling And Simulation Lab
Power System Modelling And Simulation LabPower System Modelling And Simulation Lab
Power System Modelling And Simulation Lab
 
Презентація команди Васильківської ЗОШ І-ІІІ ступенів №1
Презентація команди Васильківської ЗОШ І-ІІІ ступенів №1Презентація команди Васильківської ЗОШ І-ІІІ ступенів №1
Презентація команди Васильківської ЗОШ І-ІІІ ступенів №1
 
Chakala belan
Chakala belanChakala belan
Chakala belan
 
Hybrid Dynamic Simulation (HDS) Industrial Modeling Framework (HDS-IMF)
Hybrid Dynamic Simulation (HDS)  Industrial Modeling Framework (HDS-IMF)Hybrid Dynamic Simulation (HDS)  Industrial Modeling Framework (HDS-IMF)
Hybrid Dynamic Simulation (HDS) Industrial Modeling Framework (HDS-IMF)
 
Advanced Production Accounting of a Flotation Plant
Advanced Production Accounting of a Flotation PlantAdvanced Production Accounting of a Flotation Plant
Advanced Production Accounting of a Flotation Plant
 
Intro ed x k-ix norrköping
Intro ed x k-ix norrköpingIntro ed x k-ix norrköping
Intro ed x k-ix norrköping
 
6 vamos al mundo
6 vamos al mundo6 vamos al mundo
6 vamos al mundo
 
Development of Communicative Competence Analysis
Development of Communicative Competence AnalysisDevelopment of Communicative Competence Analysis
Development of Communicative Competence Analysis
 
Презентация Васильківської ЗОШ І-ІІІ ст№9 "Вулиця Бурки"
Презентация Васильківської ЗОШ І-ІІІ ст№9 "Вулиця Бурки"Презентация Васильківської ЗОШ І-ІІІ ст№9 "Вулиця Бурки"
Презентация Васильківської ЗОШ І-ІІІ ст№9 "Вулиця Бурки"
 
Jet fuelsupplychaindesign
Jet fuelsupplychaindesignJet fuelsupplychaindesign
Jet fuelsupplychaindesign
 
CV__DEC 2016 160816
CV__DEC 2016 160816CV__DEC 2016 160816
CV__DEC 2016 160816
 
Pendahuluan
PendahuluanPendahuluan
Pendahuluan
 
Endocrine system chapter 11
Endocrine system chapter 11Endocrine system chapter 11
Endocrine system chapter 11
 
Stock Decomposition Heuristic for Scheduling: A Priority Dispatch Rule Approach
Stock Decomposition Heuristic for Scheduling: A Priority Dispatch Rule ApproachStock Decomposition Heuristic for Scheduling: A Priority Dispatch Rule Approach
Stock Decomposition Heuristic for Scheduling: A Priority Dispatch Rule Approach
 

Similar to Sparse Observability using LP Presolve and LTDL Factorization in IMPL (IMPL-SparseObservability)

Missing-Value Handling in Dynamic Model Estimation using IMPL
Missing-Value Handling in Dynamic Model Estimation using IMPL Missing-Value Handling in Dynamic Model Estimation using IMPL
Missing-Value Handling in Dynamic Model Estimation using IMPL Alkis Vazacopoulos
 
A Regularization Approach to the Reconciliation of Constrained Data Sets
A Regularization Approach to the Reconciliation of Constrained Data SetsA Regularization Approach to the Reconciliation of Constrained Data Sets
A Regularization Approach to the Reconciliation of Constrained Data SetsAlkis Vazacopoulos
 
Application of smooth transition autoregressive (star) models for exchange rate
Application of smooth transition autoregressive (star) models for exchange rateApplication of smooth transition autoregressive (star) models for exchange rate
Application of smooth transition autoregressive (star) models for exchange rateAlexander Decker
 
IBM SPSS Statistics Algorithms.pdf
IBM SPSS Statistics Algorithms.pdfIBM SPSS Statistics Algorithms.pdf
IBM SPSS Statistics Algorithms.pdfNorafizah Samawi
 
Penalized Regressions with Different Tuning Parameter Choosing Criteria and t...
Penalized Regressions with Different Tuning Parameter Choosing Criteria and t...Penalized Regressions with Different Tuning Parameter Choosing Criteria and t...
Penalized Regressions with Different Tuning Parameter Choosing Criteria and t...CSCJournals
 
A Robust Method Based On LOVO Functions For Solving Least Squares Problems
A Robust Method Based On LOVO Functions For Solving Least Squares ProblemsA Robust Method Based On LOVO Functions For Solving Least Squares Problems
A Robust Method Based On LOVO Functions For Solving Least Squares ProblemsDawn Cook
 
from_data_to_differential_equations.ppt
from_data_to_differential_equations.pptfrom_data_to_differential_equations.ppt
from_data_to_differential_equations.pptashutoshvb1
 
Nonnegative Garrote as a Variable Selection Method in Panel Data
Nonnegative Garrote as a Variable Selection Method in Panel DataNonnegative Garrote as a Variable Selection Method in Panel Data
Nonnegative Garrote as a Variable Selection Method in Panel DataIJCSIS Research Publications
 
Scalable Constrained Spectral Clustering
Scalable Constrained Spectral ClusteringScalable Constrained Spectral Clustering
Scalable Constrained Spectral Clustering1crore projects
 
Structural Dynamic Reanalysis of Beam Elements Using Regression Method
Structural Dynamic Reanalysis of Beam Elements Using Regression MethodStructural Dynamic Reanalysis of Beam Elements Using Regression Method
Structural Dynamic Reanalysis of Beam Elements Using Regression MethodIOSR Journals
 
A delay decomposition approach to robust stability analysis of uncertain syst...
A delay decomposition approach to robust stability analysis of uncertain syst...A delay decomposition approach to robust stability analysis of uncertain syst...
A delay decomposition approach to robust stability analysis of uncertain syst...ISA Interchange
 
Reduction of-linear-time-invariant-systems-using-routh-approximation-and-pso
Reduction of-linear-time-invariant-systems-using-routh-approximation-and-psoReduction of-linear-time-invariant-systems-using-routh-approximation-and-pso
Reduction of-linear-time-invariant-systems-using-routh-approximation-and-psoCemal Ardil
 
Machine Learning.pdf
Machine Learning.pdfMachine Learning.pdf
Machine Learning.pdfBeyaNasr1
 
Bg2420212027
Bg2420212027Bg2420212027
Bg2420212027IJMER
 
EVALUATING SYMMETRIC INFORMATION GAP BETWEEN DYNAMICAL SYSTEMS USING PARTICLE...
EVALUATING SYMMETRIC INFORMATION GAP BETWEEN DYNAMICAL SYSTEMS USING PARTICLE...EVALUATING SYMMETRIC INFORMATION GAP BETWEEN DYNAMICAL SYSTEMS USING PARTICLE...
EVALUATING SYMMETRIC INFORMATION GAP BETWEEN DYNAMICAL SYSTEMS USING PARTICLE...Zac Darcy
 
Piecewise Controller Design for Affine Fuzzy Systems
Piecewise Controller Design for Affine Fuzzy SystemsPiecewise Controller Design for Affine Fuzzy Systems
Piecewise Controller Design for Affine Fuzzy SystemsISA Interchange
 
Piecewise controller design for affine fuzzy systems
Piecewise controller design for affine fuzzy systemsPiecewise controller design for affine fuzzy systems
Piecewise controller design for affine fuzzy systemsISA Interchange
 

Similar to Sparse Observability using LP Presolve and LTDL Factorization in IMPL (IMPL-SparseObservability) (20)

Missing-Value Handling in Dynamic Model Estimation using IMPL
Missing-Value Handling in Dynamic Model Estimation using IMPL Missing-Value Handling in Dynamic Model Estimation using IMPL
Missing-Value Handling in Dynamic Model Estimation using IMPL
 
A Regularization Approach to the Reconciliation of Constrained Data Sets
A Regularization Approach to the Reconciliation of Constrained Data SetsA Regularization Approach to the Reconciliation of Constrained Data Sets
A Regularization Approach to the Reconciliation of Constrained Data Sets
 
Application of smooth transition autoregressive (star) models for exchange rate
Application of smooth transition autoregressive (star) models for exchange rateApplication of smooth transition autoregressive (star) models for exchange rate
Application of smooth transition autoregressive (star) models for exchange rate
 
IBM SPSS Statistics Algorithms.pdf
IBM SPSS Statistics Algorithms.pdfIBM SPSS Statistics Algorithms.pdf
IBM SPSS Statistics Algorithms.pdf
 
Penalized Regressions with Different Tuning Parameter Choosing Criteria and t...
Penalized Regressions with Different Tuning Parameter Choosing Criteria and t...Penalized Regressions with Different Tuning Parameter Choosing Criteria and t...
Penalized Regressions with Different Tuning Parameter Choosing Criteria and t...
 
A Robust Method Based On LOVO Functions For Solving Least Squares Problems
A Robust Method Based On LOVO Functions For Solving Least Squares ProblemsA Robust Method Based On LOVO Functions For Solving Least Squares Problems
A Robust Method Based On LOVO Functions For Solving Least Squares Problems
 
from_data_to_differential_equations.ppt
from_data_to_differential_equations.pptfrom_data_to_differential_equations.ppt
from_data_to_differential_equations.ppt
 
Nonnegative Garrote as a Variable Selection Method in Panel Data
Nonnegative Garrote as a Variable Selection Method in Panel DataNonnegative Garrote as a Variable Selection Method in Panel Data
Nonnegative Garrote as a Variable Selection Method in Panel Data
 
Colombo14a
Colombo14aColombo14a
Colombo14a
 
Ee35741744
Ee35741744Ee35741744
Ee35741744
 
Scalable Constrained Spectral Clustering
Scalable Constrained Spectral ClusteringScalable Constrained Spectral Clustering
Scalable Constrained Spectral Clustering
 
Structural Dynamic Reanalysis of Beam Elements Using Regression Method
Structural Dynamic Reanalysis of Beam Elements Using Regression MethodStructural Dynamic Reanalysis of Beam Elements Using Regression Method
Structural Dynamic Reanalysis of Beam Elements Using Regression Method
 
The Short-term Swap Rate Models in China
The Short-term Swap Rate Models in ChinaThe Short-term Swap Rate Models in China
The Short-term Swap Rate Models in China
 
A delay decomposition approach to robust stability analysis of uncertain syst...
A delay decomposition approach to robust stability analysis of uncertain syst...A delay decomposition approach to robust stability analysis of uncertain syst...
A delay decomposition approach to robust stability analysis of uncertain syst...
 
Reduction of-linear-time-invariant-systems-using-routh-approximation-and-pso
Reduction of-linear-time-invariant-systems-using-routh-approximation-and-psoReduction of-linear-time-invariant-systems-using-routh-approximation-and-pso
Reduction of-linear-time-invariant-systems-using-routh-approximation-and-pso
 
Machine Learning.pdf
Machine Learning.pdfMachine Learning.pdf
Machine Learning.pdf
 
Bg2420212027
Bg2420212027Bg2420212027
Bg2420212027
 
EVALUATING SYMMETRIC INFORMATION GAP BETWEEN DYNAMICAL SYSTEMS USING PARTICLE...
EVALUATING SYMMETRIC INFORMATION GAP BETWEEN DYNAMICAL SYSTEMS USING PARTICLE...EVALUATING SYMMETRIC INFORMATION GAP BETWEEN DYNAMICAL SYSTEMS USING PARTICLE...
EVALUATING SYMMETRIC INFORMATION GAP BETWEEN DYNAMICAL SYSTEMS USING PARTICLE...
 
Piecewise Controller Design for Affine Fuzzy Systems
Piecewise Controller Design for Affine Fuzzy SystemsPiecewise Controller Design for Affine Fuzzy Systems
Piecewise Controller Design for Affine Fuzzy Systems
 
Piecewise controller design for affine fuzzy systems
Piecewise controller design for affine fuzzy systemsPiecewise controller design for affine fuzzy systems
Piecewise controller design for affine fuzzy systems
 

More from Alkis Vazacopoulos

Automatic Fine-tuning Xpress-MP to Solve MIP
Automatic Fine-tuning Xpress-MP to Solve MIPAutomatic Fine-tuning Xpress-MP to Solve MIP
Automatic Fine-tuning Xpress-MP to Solve MIPAlkis Vazacopoulos
 
Amazing results with ODH|CPLEX
Amazing results with ODH|CPLEXAmazing results with ODH|CPLEX
Amazing results with ODH|CPLEXAlkis Vazacopoulos
 
Bia project poster fantasy football
Bia project poster  fantasy football Bia project poster  fantasy football
Bia project poster fantasy football Alkis Vazacopoulos
 
NFL Game schedule optimization
NFL Game schedule optimization NFL Game schedule optimization
NFL Game schedule optimization Alkis Vazacopoulos
 
2017 Business Intelligence & Analytics Corporate Event Stevens Institute of T...
2017 Business Intelligence & Analytics Corporate Event Stevens Institute of T...2017 Business Intelligence & Analytics Corporate Event Stevens Institute of T...
2017 Business Intelligence & Analytics Corporate Event Stevens Institute of T...Alkis Vazacopoulos
 
Very largeoptimizationparallel
Very largeoptimizationparallelVery largeoptimizationparallel
Very largeoptimizationparallelAlkis Vazacopoulos
 
Optimization Direct: Introduction and recent case studies
Optimization Direct: Introduction and recent case studiesOptimization Direct: Introduction and recent case studies
Optimization Direct: Introduction and recent case studiesAlkis Vazacopoulos
 
Informs 2016 Solving Planning and Scheduling Problems with CPLEX
Informs 2016 Solving Planning and Scheduling Problems with CPLEX Informs 2016 Solving Planning and Scheduling Problems with CPLEX
Informs 2016 Solving Planning and Scheduling Problems with CPLEX Alkis Vazacopoulos
 
Finite Impulse Response Estimation of Gas Furnace Data in IMPL Industrial Mod...
Finite Impulse Response Estimation of Gas Furnace Data in IMPL Industrial Mod...Finite Impulse Response Estimation of Gas Furnace Data in IMPL Industrial Mod...
Finite Impulse Response Estimation of Gas Furnace Data in IMPL Industrial Mod...Alkis Vazacopoulos
 
Industrial Modeling Service (IMS-IMPL)
Industrial Modeling Service (IMS-IMPL)Industrial Modeling Service (IMS-IMPL)
Industrial Modeling Service (IMS-IMPL)Alkis Vazacopoulos
 
Dither Signal Design Problem (DSDP) for Closed-Loop Estimation Industrial Mod...
Dither Signal Design Problem (DSDP) for Closed-Loop Estimation Industrial Mod...Dither Signal Design Problem (DSDP) for Closed-Loop Estimation Industrial Mod...
Dither Signal Design Problem (DSDP) for Closed-Loop Estimation Industrial Mod...Alkis Vazacopoulos
 
Distillation Curve Optimization Using Monotonic Interpolation
Distillation Curve Optimization Using Monotonic InterpolationDistillation Curve Optimization Using Monotonic Interpolation
Distillation Curve Optimization Using Monotonic InterpolationAlkis Vazacopoulos
 
Multi-Utility Scheduling Optimization (MUSO) Industrial Modeling Framework (M...
Multi-Utility Scheduling Optimization (MUSO) Industrial Modeling Framework (M...Multi-Utility Scheduling Optimization (MUSO) Industrial Modeling Framework (M...
Multi-Utility Scheduling Optimization (MUSO) Industrial Modeling Framework (M...Alkis Vazacopoulos
 
Advanced Parameter Estimation (APE) for Motor Gasoline Blending (MGB) Indust...
Advanced Parameter Estimation (APE) for Motor Gasoline Blending (MGB)  Indust...Advanced Parameter Estimation (APE) for Motor Gasoline Blending (MGB)  Indust...
Advanced Parameter Estimation (APE) for Motor Gasoline Blending (MGB) Indust...Alkis Vazacopoulos
 
Partial Differential Equations (PDE’s) Industrial Modeling Framework (PDE-IMF)
Partial Differential Equations (PDE’s)  Industrial Modeling Framework (PDE-IMF)Partial Differential Equations (PDE’s)  Industrial Modeling Framework (PDE-IMF)
Partial Differential Equations (PDE’s) Industrial Modeling Framework (PDE-IMF)Alkis Vazacopoulos
 

More from Alkis Vazacopoulos (20)

Automatic Fine-tuning Xpress-MP to Solve MIP
Automatic Fine-tuning Xpress-MP to Solve MIPAutomatic Fine-tuning Xpress-MP to Solve MIP
Automatic Fine-tuning Xpress-MP to Solve MIP
 
Data mining 2004
Data mining 2004Data mining 2004
Data mining 2004
 
Amazing results with ODH|CPLEX
Amazing results with ODH|CPLEXAmazing results with ODH|CPLEX
Amazing results with ODH|CPLEX
 
Bia project poster fantasy football
Bia project poster  fantasy football Bia project poster  fantasy football
Bia project poster fantasy football
 
NFL Game schedule optimization
NFL Game schedule optimization NFL Game schedule optimization
NFL Game schedule optimization
 
2017 Business Intelligence & Analytics Corporate Event Stevens Institute of T...
2017 Business Intelligence & Analytics Corporate Event Stevens Institute of T...2017 Business Intelligence & Analytics Corporate Event Stevens Institute of T...
2017 Business Intelligence & Analytics Corporate Event Stevens Institute of T...
 
Posters 2017
Posters 2017Posters 2017
Posters 2017
 
Very largeoptimizationparallel
Very largeoptimizationparallelVery largeoptimizationparallel
Very largeoptimizationparallel
 
Retail Pricing Optimization
Retail Pricing Optimization Retail Pricing Optimization
Retail Pricing Optimization
 
Optimization Direct: Introduction and recent case studies
Optimization Direct: Introduction and recent case studiesOptimization Direct: Introduction and recent case studies
Optimization Direct: Introduction and recent case studies
 
Informs 2016 Solving Planning and Scheduling Problems with CPLEX
Informs 2016 Solving Planning and Scheduling Problems with CPLEX Informs 2016 Solving Planning and Scheduling Problems with CPLEX
Informs 2016 Solving Planning and Scheduling Problems with CPLEX
 
ODHeuristics
ODHeuristicsODHeuristics
ODHeuristics
 
Finite Impulse Response Estimation of Gas Furnace Data in IMPL Industrial Mod...
Finite Impulse Response Estimation of Gas Furnace Data in IMPL Industrial Mod...Finite Impulse Response Estimation of Gas Furnace Data in IMPL Industrial Mod...
Finite Impulse Response Estimation of Gas Furnace Data in IMPL Industrial Mod...
 
Industrial Modeling Service (IMS-IMPL)
Industrial Modeling Service (IMS-IMPL)Industrial Modeling Service (IMS-IMPL)
Industrial Modeling Service (IMS-IMPL)
 
Dither Signal Design Problem (DSDP) for Closed-Loop Estimation Industrial Mod...
Dither Signal Design Problem (DSDP) for Closed-Loop Estimation Industrial Mod...Dither Signal Design Problem (DSDP) for Closed-Loop Estimation Industrial Mod...
Dither Signal Design Problem (DSDP) for Closed-Loop Estimation Industrial Mod...
 
Xmr im
Xmr imXmr im
Xmr im
 
Distillation Curve Optimization Using Monotonic Interpolation
Distillation Curve Optimization Using Monotonic InterpolationDistillation Curve Optimization Using Monotonic Interpolation
Distillation Curve Optimization Using Monotonic Interpolation
 
Multi-Utility Scheduling Optimization (MUSO) Industrial Modeling Framework (M...
Multi-Utility Scheduling Optimization (MUSO) Industrial Modeling Framework (M...Multi-Utility Scheduling Optimization (MUSO) Industrial Modeling Framework (M...
Multi-Utility Scheduling Optimization (MUSO) Industrial Modeling Framework (M...
 
Advanced Parameter Estimation (APE) for Motor Gasoline Blending (MGB) Indust...
Advanced Parameter Estimation (APE) for Motor Gasoline Blending (MGB)  Indust...Advanced Parameter Estimation (APE) for Motor Gasoline Blending (MGB)  Indust...
Advanced Parameter Estimation (APE) for Motor Gasoline Blending (MGB) Indust...
 
Partial Differential Equations (PDE’s) Industrial Modeling Framework (PDE-IMF)
Partial Differential Equations (PDE’s)  Industrial Modeling Framework (PDE-IMF)Partial Differential Equations (PDE’s)  Industrial Modeling Framework (PDE-IMF)
Partial Differential Equations (PDE’s) Industrial Modeling Framework (PDE-IMF)
 

Recently uploaded

꧁❤ Greater Noida Call Girls Delhi ❤꧂ 9711199171 ☎️ Hard And Sexy Vip Call
꧁❤ Greater Noida Call Girls Delhi ❤꧂ 9711199171 ☎️ Hard And Sexy Vip Call꧁❤ Greater Noida Call Girls Delhi ❤꧂ 9711199171 ☎️ Hard And Sexy Vip Call
꧁❤ Greater Noida Call Girls Delhi ❤꧂ 9711199171 ☎️ Hard And Sexy Vip Callshivangimorya083
 
DBA Basics: Getting Started with Performance Tuning.pdf
DBA Basics: Getting Started with Performance Tuning.pdfDBA Basics: Getting Started with Performance Tuning.pdf
DBA Basics: Getting Started with Performance Tuning.pdfJohn Sterrett
 
(PARI) Call Girls Wanowrie ( 7001035870 ) HI-Fi Pune Escorts Service
(PARI) Call Girls Wanowrie ( 7001035870 ) HI-Fi Pune Escorts Service(PARI) Call Girls Wanowrie ( 7001035870 ) HI-Fi Pune Escorts Service
(PARI) Call Girls Wanowrie ( 7001035870 ) HI-Fi Pune Escorts Serviceranjana rawat
 
Saket, (-DELHI )+91-9654467111-(=)CHEAP Call Girls in Escorts Service Saket C...
Saket, (-DELHI )+91-9654467111-(=)CHEAP Call Girls in Escorts Service Saket C...Saket, (-DELHI )+91-9654467111-(=)CHEAP Call Girls in Escorts Service Saket C...
Saket, (-DELHI )+91-9654467111-(=)CHEAP Call Girls in Escorts Service Saket C...Sapana Sha
 
PKS-TGC-1084-630 - Stage 1 Proposal.pptx
PKS-TGC-1084-630 - Stage 1 Proposal.pptxPKS-TGC-1084-630 - Stage 1 Proposal.pptx
PKS-TGC-1084-630 - Stage 1 Proposal.pptxPramod Kumar Srivastava
 
办理学位证纽约大学毕业证(NYU毕业证书)原版一比一
办理学位证纽约大学毕业证(NYU毕业证书)原版一比一办理学位证纽约大学毕业证(NYU毕业证书)原版一比一
办理学位证纽约大学毕业证(NYU毕业证书)原版一比一fhwihughh
 
1:1定制(UQ毕业证)昆士兰大学毕业证成绩单修改留信学历认证原版一模一样
1:1定制(UQ毕业证)昆士兰大学毕业证成绩单修改留信学历认证原版一模一样1:1定制(UQ毕业证)昆士兰大学毕业证成绩单修改留信学历认证原版一模一样
1:1定制(UQ毕业证)昆士兰大学毕业证成绩单修改留信学历认证原版一模一样vhwb25kk
 
Brighton SEO | April 2024 | Data Storytelling
Brighton SEO | April 2024 | Data StorytellingBrighton SEO | April 2024 | Data Storytelling
Brighton SEO | April 2024 | Data StorytellingNeil Barnes
 
Predictive Analysis - Using Insight-informed Data to Determine Factors Drivin...
Predictive Analysis - Using Insight-informed Data to Determine Factors Drivin...Predictive Analysis - Using Insight-informed Data to Determine Factors Drivin...
Predictive Analysis - Using Insight-informed Data to Determine Factors Drivin...ThinkInnovation
 
RA-11058_IRR-COMPRESS Do 198 series of 1998
RA-11058_IRR-COMPRESS Do 198 series of 1998RA-11058_IRR-COMPRESS Do 198 series of 1998
RA-11058_IRR-COMPRESS Do 198 series of 1998YohFuh
 
Industrialised data - the key to AI success.pdf
Industrialised data - the key to AI success.pdfIndustrialised data - the key to AI success.pdf
Industrialised data - the key to AI success.pdfLars Albertsson
 
04242024_CCC TUG_Joins and Relationships
04242024_CCC TUG_Joins and Relationships04242024_CCC TUG_Joins and Relationships
04242024_CCC TUG_Joins and Relationshipsccctableauusergroup
 
GA4 Without Cookies [Measure Camp AMS]
GA4 Without Cookies [Measure Camp AMS]GA4 Without Cookies [Measure Camp AMS]
GA4 Without Cookies [Measure Camp AMS]📊 Markus Baersch
 
EMERCE - 2024 - AMSTERDAM - CROSS-PLATFORM TRACKING WITH GOOGLE ANALYTICS.pptx
EMERCE - 2024 - AMSTERDAM - CROSS-PLATFORM  TRACKING WITH GOOGLE ANALYTICS.pptxEMERCE - 2024 - AMSTERDAM - CROSS-PLATFORM  TRACKING WITH GOOGLE ANALYTICS.pptx
EMERCE - 2024 - AMSTERDAM - CROSS-PLATFORM TRACKING WITH GOOGLE ANALYTICS.pptxthyngster
 
Indian Call Girls in Abu Dhabi O5286O24O8 Call Girls in Abu Dhabi By Independ...
Indian Call Girls in Abu Dhabi O5286O24O8 Call Girls in Abu Dhabi By Independ...Indian Call Girls in Abu Dhabi O5286O24O8 Call Girls in Abu Dhabi By Independ...
Indian Call Girls in Abu Dhabi O5286O24O8 Call Girls in Abu Dhabi By Independ...dajasot375
 
From idea to production in a day – Leveraging Azure ML and Streamlit to build...
From idea to production in a day – Leveraging Azure ML and Streamlit to build...From idea to production in a day – Leveraging Azure ML and Streamlit to build...
From idea to production in a day – Leveraging Azure ML and Streamlit to build...Florian Roscheck
 
办理(Vancouver毕业证书)加拿大温哥华岛大学毕业证成绩单原版一比一
办理(Vancouver毕业证书)加拿大温哥华岛大学毕业证成绩单原版一比一办理(Vancouver毕业证书)加拿大温哥华岛大学毕业证成绩单原版一比一
办理(Vancouver毕业证书)加拿大温哥华岛大学毕业证成绩单原版一比一F La
 
Dubai Call Girls Wifey O52&786472 Call Girls Dubai
Dubai Call Girls Wifey O52&786472 Call Girls DubaiDubai Call Girls Wifey O52&786472 Call Girls Dubai
Dubai Call Girls Wifey O52&786472 Call Girls Dubaihf8803863
 

Recently uploaded (20)

꧁❤ Greater Noida Call Girls Delhi ❤꧂ 9711199171 ☎️ Hard And Sexy Vip Call
꧁❤ Greater Noida Call Girls Delhi ❤꧂ 9711199171 ☎️ Hard And Sexy Vip Call꧁❤ Greater Noida Call Girls Delhi ❤꧂ 9711199171 ☎️ Hard And Sexy Vip Call
꧁❤ Greater Noida Call Girls Delhi ❤꧂ 9711199171 ☎️ Hard And Sexy Vip Call
 
DBA Basics: Getting Started with Performance Tuning.pdf
DBA Basics: Getting Started with Performance Tuning.pdfDBA Basics: Getting Started with Performance Tuning.pdf
DBA Basics: Getting Started with Performance Tuning.pdf
 
(PARI) Call Girls Wanowrie ( 7001035870 ) HI-Fi Pune Escorts Service
(PARI) Call Girls Wanowrie ( 7001035870 ) HI-Fi Pune Escorts Service(PARI) Call Girls Wanowrie ( 7001035870 ) HI-Fi Pune Escorts Service
(PARI) Call Girls Wanowrie ( 7001035870 ) HI-Fi Pune Escorts Service
 
Saket, (-DELHI )+91-9654467111-(=)CHEAP Call Girls in Escorts Service Saket C...
Saket, (-DELHI )+91-9654467111-(=)CHEAP Call Girls in Escorts Service Saket C...Saket, (-DELHI )+91-9654467111-(=)CHEAP Call Girls in Escorts Service Saket C...
Saket, (-DELHI )+91-9654467111-(=)CHEAP Call Girls in Escorts Service Saket C...
 
PKS-TGC-1084-630 - Stage 1 Proposal.pptx
PKS-TGC-1084-630 - Stage 1 Proposal.pptxPKS-TGC-1084-630 - Stage 1 Proposal.pptx
PKS-TGC-1084-630 - Stage 1 Proposal.pptx
 
办理学位证纽约大学毕业证(NYU毕业证书)原版一比一
办理学位证纽约大学毕业证(NYU毕业证书)原版一比一办理学位证纽约大学毕业证(NYU毕业证书)原版一比一
办理学位证纽约大学毕业证(NYU毕业证书)原版一比一
 
1:1定制(UQ毕业证)昆士兰大学毕业证成绩单修改留信学历认证原版一模一样
1:1定制(UQ毕业证)昆士兰大学毕业证成绩单修改留信学历认证原版一模一样1:1定制(UQ毕业证)昆士兰大学毕业证成绩单修改留信学历认证原版一模一样
1:1定制(UQ毕业证)昆士兰大学毕业证成绩单修改留信学历认证原版一模一样
 
Brighton SEO | April 2024 | Data Storytelling
Brighton SEO | April 2024 | Data StorytellingBrighton SEO | April 2024 | Data Storytelling
Brighton SEO | April 2024 | Data Storytelling
 
VIP Call Girls Service Charbagh { Lucknow Call Girls Service 9548273370 } Boo...
VIP Call Girls Service Charbagh { Lucknow Call Girls Service 9548273370 } Boo...VIP Call Girls Service Charbagh { Lucknow Call Girls Service 9548273370 } Boo...
VIP Call Girls Service Charbagh { Lucknow Call Girls Service 9548273370 } Boo...
 
Predictive Analysis - Using Insight-informed Data to Determine Factors Drivin...
Predictive Analysis - Using Insight-informed Data to Determine Factors Drivin...Predictive Analysis - Using Insight-informed Data to Determine Factors Drivin...
Predictive Analysis - Using Insight-informed Data to Determine Factors Drivin...
 
RA-11058_IRR-COMPRESS Do 198 series of 1998
RA-11058_IRR-COMPRESS Do 198 series of 1998RA-11058_IRR-COMPRESS Do 198 series of 1998
RA-11058_IRR-COMPRESS Do 198 series of 1998
 
Decoding Loan Approval: Predictive Modeling in Action
Decoding Loan Approval: Predictive Modeling in ActionDecoding Loan Approval: Predictive Modeling in Action
Decoding Loan Approval: Predictive Modeling in Action
 
Industrialised data - the key to AI success.pdf
Industrialised data - the key to AI success.pdfIndustrialised data - the key to AI success.pdf
Industrialised data - the key to AI success.pdf
 
04242024_CCC TUG_Joins and Relationships
04242024_CCC TUG_Joins and Relationships04242024_CCC TUG_Joins and Relationships
04242024_CCC TUG_Joins and Relationships
 
GA4 Without Cookies [Measure Camp AMS]
GA4 Without Cookies [Measure Camp AMS]GA4 Without Cookies [Measure Camp AMS]
GA4 Without Cookies [Measure Camp AMS]
 
EMERCE - 2024 - AMSTERDAM - CROSS-PLATFORM TRACKING WITH GOOGLE ANALYTICS.pptx
EMERCE - 2024 - AMSTERDAM - CROSS-PLATFORM  TRACKING WITH GOOGLE ANALYTICS.pptxEMERCE - 2024 - AMSTERDAM - CROSS-PLATFORM  TRACKING WITH GOOGLE ANALYTICS.pptx
EMERCE - 2024 - AMSTERDAM - CROSS-PLATFORM TRACKING WITH GOOGLE ANALYTICS.pptx
 
Indian Call Girls in Abu Dhabi O5286O24O8 Call Girls in Abu Dhabi By Independ...
Indian Call Girls in Abu Dhabi O5286O24O8 Call Girls in Abu Dhabi By Independ...Indian Call Girls in Abu Dhabi O5286O24O8 Call Girls in Abu Dhabi By Independ...
Indian Call Girls in Abu Dhabi O5286O24O8 Call Girls in Abu Dhabi By Independ...
 
From idea to production in a day – Leveraging Azure ML and Streamlit to build...
From idea to production in a day – Leveraging Azure ML and Streamlit to build...From idea to production in a day – Leveraging Azure ML and Streamlit to build...
From idea to production in a day – Leveraging Azure ML and Streamlit to build...
 
办理(Vancouver毕业证书)加拿大温哥华岛大学毕业证成绩单原版一比一
办理(Vancouver毕业证书)加拿大温哥华岛大学毕业证成绩单原版一比一办理(Vancouver毕业证书)加拿大温哥华岛大学毕业证成绩单原版一比一
办理(Vancouver毕业证书)加拿大温哥华岛大学毕业证成绩单原版一比一
 
Dubai Call Girls Wifey O52&786472 Call Girls Dubai
Dubai Call Girls Wifey O52&786472 Call Girls DubaiDubai Call Girls Wifey O52&786472 Call Girls Dubai
Dubai Call Girls Wifey O52&786472 Call Girls Dubai
 

Sparse Observability using LP Presolve and LTDL Factorization in IMPL (IMPL-SparseObservability)

  • 1. Sparse Observability using LP Presolve and LT DL Factorization in IMPL (IMPL-SparseObservability) i n d u s t r IAL g o r i t h m s LLC. (IAL) www.industrialgorithms.com August 2014 Introduction Presented in this short document is a description of our technology we call “Sparse Observability”. Observability is the estimatability metric (Bagajewicz, 2010) to structurally determine that an unmeasured variable or regressed parameter is either uniquely solvable (observable) or otherwise unsolvable (unobservable) in data reconciliation and regression (DRR) applications. Ultimately, our purpose to use efficient sparse matrix techniques is to solve large industrial DRR flowsheets quickly and accurately. Most other implementations of observability calculation use dense linear algebra such as reduced row echelon form (RREF), Gauss-Jordan decomposition (Crowe et. al. 1983; Madron 1992), QR factorization which can now be considered as semi-sparse (Swartz, 1989; Sanchez and Romagnoli, 1996), Schur complements, Cholesky factorization (Kelly, 1998a) and singular value decomposition (SVD) (Kelly, 1999). A sparse LU decomposition with complete-pivoting from Albuquerque and Biegler (1996) for dynamic data reconciliation observability computation was used but it is uncertain if complete-pivoting causes extreme “fill-ins” of the lower and upper triangular matrices essentially making them near-dense. There is another sparse observability method using an LP sub-solver found in Kelly and Zyngier (2008) but this requires solving as many LP sub-problems as there are unmeasured variables which can be considered as somewhat inefficient. IMPL’s sparse observability technique uses the variable classification and nomenclature found in Kelly (1998b) given that if we partition or separate the unmeasured variables into independent (B12) and dependent (B34) sub-matrices then all dependent unmeasured variables by definition are unobservable. If any independent unmeasured variable is a (linear) function of any dependent variable then this independent variable is of course also unobservable because it is dependent on another non-observable variable. This can be easily seen using the following (linearized) DRR equation (Kelly, 1998b): A * x + [B12 B34] * [y12T y34T ]T + C * z = 0 or A * x + B12 * y12 + B34 * y34 + C * z = 0 where x represents the measured variables in column vector form, y are the unmeasured variables and z are the fixed variables or constants with corresponding sparse Jacobian matrices A, B12, B34 and C respectively. By multiplying both sides by B12T and rearranging we arrive at essentially the Normal equation found in typical ordinary least squares (OLS) regression which is unique to the classification method of Kelly (1998b): B12T * B12 * y12 = -B12T * (A * x + B34 * y34 + C * z) or y12 = -(B12T * B12)-1 * B12T * (B34 * y34 + A * x + C * z)
  • 2. Therefore, if (B12T * B12)-1 * B12T * B34 has any non-zero elements (relative to machine precision) in any row then the corresponding independent unmeasured y12 variable is a linear function of a non-observable unmeasured y34 variable. Hence, the technology described below is an effective and reliable way to identify unobservable unmeasured variables in large-scale and industrial-sized flowsheet DRR problems. It should be mentioned that once the observability calculations have been performed, then the same matrices can be used to identify redundant and non-redundant measured variables described in Kelly (1998b). Redundant variables are observable if there measurement reading is missing, deleted or removed from the data content of the DRR problem. Non-redundant variables are unobservable if their measurement is no longer available given the remaining model and data of the DRR problem. Obviously, it is important to have as much hardware (measurement readings) and software (model relationships) redundancy in the system as possible given the capital, operational and maintenance costs where this sensor network design optimization problem is addressed in Kelly and Zyngier (2008). Technology IMPL’s sparse observability has two steps or parts: a presolve or preprocessing step to identify essentially trivial observable (strongly independent) unmeasured variables and a LT DL factorization for sparse symmetric indefinite matrices on the remaining equations and unmeasured variables to determine the dependent unmeasured variables. Our DRR presolve is taken directly from primal LP presolve techniques (Andersen and Andersen, 1995) and operates on the B sparse matrix (i.e., [B12 B34]) and to our knowledge, we are the first to adopt primal LP presolving techniques to the application of identifying observable unmeasured variables in DRR problems. First, it involves removing singleton rows with a single column indicating that the column or unmeasured variable is strongly observable because it is not a function of any other unmeasured variable. And second, it removes doubleton rows of the form b * yi – b * yj = 0 and one of the yi or yj unmeasured variables are removed because they are mirror images of one another; thus reducing the number of rows and columns of B. These two operations are repeated until there is no more improvement in reducing the rows and columns of B or there are no rows or columns left in B. These two straightforward primal LP presolve operations are very effective in practice and usually shrink the B matrix considerably. As an example, Figure 1 shows the unmeasured variable’s sparse matrix B taken from Crowe et. al. (1983)’s example #3.
  • 3. Figure 1. Crowe et. al. (1983)’s Example #3. All columns or unmeasured variables are observable and we determined these to be observable using our LP presolving techniques. For instance, the first pass or iteration of our presolve finds eight (8) singleton rows that can be removed i.e., “Node1 N2”, “Node1 Ar”, “Node2 N2”, “Node2 Ar”, “Node3 N2”, “Node3 H2”, “Node3 NH3”, “Splitter Constraints H2”. This means, that of the total thirteen (13) unmeasured variables, 8 are determined to be strongly observable easily. When these 8 rows and columns are removed a second pass further removes rows “Node2 NH3”, “Node2 H2”, “Node3 Ar” and “Node4 H2”. The third and final pass removes row “Node1 H2” and now completely identifies all columns as observable. When not all unmeasured variables are declared to be observable using our presolve method, we then need to proceed to identifying from the reduced or presolved set, which are independent and dependent columns using either of the methods mentioned previously. Instead, our approach in IMPL is to use the LT DL (Cholesky-like) factorization for symmetric indefinite sparse matrices from Intel’s Math Kernel Library (MKL) PARDISO (Schenk and Gartner, 2006). LT DL is similar to Cholesky factorization except that the former performs symmetric and other types of pivoting required to precisely determine if a symmetric matrix is singular (Hansen, 1987) where the later does not. More specifically, indefinite matrices have both positive and negative eigenvalues and some can be zero (semi-definite) and these types of indefinite matrices are found in non-convex and nonlinear interior-point optimization problems such as those solved by IPOPT. IMPL also supports locating independent and dependent unmeasured variables using a sparse version of the modified Cholesky factorization from Kelly (1998a) and, even though it does not employ any pivoting techniques, it is surprisingly accurate even for B matrices with large condition numbers i.e., maximum singular value divided by its minimum singular value. And, IMPL further supports a sparse LU decomposition technique with partial-pivoting to check for near-zero diagonal elements of the upper triangular matrix (U). If Ui,i ~ 0 then this (usually) indicates that column [B]i is dependent but is less accurate than the LT DL and our modified Cholesky.
  • 4. In summary, when the LT DL factorization in PARDISO is performed using parallel-processing applied to the presolved BT B matrix, this provides a very fast, accurate and robust technique to identify observable and unobservable unmeasured variables in large-scale systems as well as its complement of redundant and non-redundant measured variables. For nonlinear problems, the A, B and C sparse matrices are linearized from the converged x and y solution values and provide only “point or local estimates” of the observability and redundancy indicators. References Crowe, C.M., Garcia Campos, Y.A., Hrymak, A., “Reconciliation of process flowrates by matrix projection, part 1: linear case”, American Institute of Chemical Engineering Journal, 29, 881- 888, (1983). Hansen, P.C., “Detection of near-singularity in Cholesky and LDL factorizations”, Journal of Computational and Applied Mathematics, 19, 293-299, (1987). Swartz, C.L.E., “Data reconciliation for generalized flowsheet applications”, 197th National Meeting, American Chemical Society (ACS), Dallas, Texas, (1989). Madron, F., “Process plant performance: measurement and data processing for optimization and retrofits”, Ellis Horwood, New York, NY, (1992). Andersen, E.D., Andersen, K.D., “Presolving in linear programming”, Mathematical Programming, 71, 221-245, (1995). Albuquerque, J.S., Biegler, L.T., “Data reconciliation and gross-error detection for dynamic systems”, American Institute of Chemical Engineering Journal, 42, 10, (1996). Sanchez, M., Romagnoli, J., “Use of orthogonal transformations in data classification- reconciliation”, Computers & Chemical Engineering, 20, 483-493, (1996). Kelly, J.D., "On finding the matrix projection in the data reconciliation solution", Computers & Chemical Engineering, 1553, (1998a). Kelly, J.D., "A regularization approach to the reconciliation of constrained data sets", Computers & Chemical Engineering, 1771, (1998b). Kelly, J.D., "Reconciliation of process data using other projection matrices", Computers & Chemical Engineering, 785, (1999). Schenk, O., Gartner, K., “On fast factorization pivoting methods for sparse symmetric indefinite systems”, Electronic Transactions on Numerical Analysis, 23, 158-179, (2006). Kelly, J.D., Zyngier, D., "A new and improved MILP formulation to optimize observability, redundancy and precision for sensor network problems", American Institute of Chemical Engineering Journal, 54, 1282, (2008). Bagajewicz, M., “Smart process plants: software and hardware for accurate data and profitable operations”, McGraw-Hill (ISBN:978-0-07-160471), (2010).