A secondary goal of the ESAMLab is to develop software tools that allow implementation of new methodologies in practice. The following codes were developed as part of this effort. All the programs are distributed in the hope that they will be useful to the research community, but they are offered WITHOUT ANY WARRANTY. (List of programs is in chronological order, from most recent)
FEM Laplace-Beltrami Spectrum estimator for mesh and voxel data.- Matlab code that computes the spectrum of the Laplace-Beltrami operator, used for SPC of mesh and voxel scanned parts.
This set of Matlab functions accompanies the paper “A registration-free approach for Statistical Process Control of 3D
scanned objects via FEM”, Precision Engineering, 74, pp. 247-263 (2022),by X. Zhao and EDC. It computes the eigendecomposition of the LB estimator, obtained via Finite Element Methods, solving for the first “k” eigenvalues and eigenvectors. For the paper, see:
https://doi.org/10.101 /j.precisioneng.2021.10.018.
Download the code here.
OptimaRegion- an R package for the computation of confidence regions on the location of optima of Thin Plate Spline and polynomial models.
This is an R package that allows users to compute confidence regions on optima of response surface models (polynomial and Thin Plate Splines) using bootstrapping, and hence, it requires no normality assumptions. It accompanies the paper “ Confidence regions for the location of Response Surface Optima: the R package OptimaRegion”, Communications in Statistics, Simulation and Computation, https://doi.org/10.1080/03610918.2020.1823412, (2020). Please visit the CRAN web site for information about this package:
OptimizeWeigher- a program for the near optimal setup of a multihead weighing machine
This R code implements the methods in the paper “Optimal Targets Setup of a Multihead Weighing Machine” by E. del Castillo, A. Beretta, and Q. Semeraro, EJOR (2019).
SPS- Sparse Precision Selection algorithm for fitting Gaussian Random Field models to large datasets
This MATLAB code accompanies the paper ““On the Theoretical Guarantees for Parameter Estimation of Gaussian Random Field Models: a Sparse Precision Matrix Approach”, by Sam Davanloo Tajbakhsh, S. Aybat and E. del Castillo, uploaded to arXiv:1405.5576, 2016. It allows users to fit either a single isotropic GRF (recommended for about n<1600), or, if n>1600 it can use a segmentation approach in which case an anisotropic model can result. The code runs on Matlab 2014 and 2015b, and has been tested in Macintosh OS X Mavericks, Yosemite and El Capitan, and in MS Windows versions 7 and 8. Consult the readme file for installation and use instructions.
Readme file SPS code (zipped file)
GGP- Geodesic Gaussian Process (GGP) model fitting
This set of MATLAB programs accompanies the paper “Geodesic Gaussian Processes for the Reconstruction of a Free-Form Surface” by Enrique del Castillo, Bianca M. Colosimo and Sam D. Tajbakhsh, Technometrics, 2015. They fit a Geodesic Gaussian Process as explained in that paper to N points (x; y; z) (point cloud data), typically collected using a scanner or non-contact device, and provides the best estimate of the underlying ‘true” surface in the presence of measurement noise in each of the 3 coordinates for all of the data sets in the paper. There are two sets of programs, one recommended for small problems up to N=1600 points, and one for larger datasets. Since the main program for big and small N require common functions, it is suggested to simply place all files in the same folder. The programs were developed with MATLAB 2011b.
Readme file–SMALL N — Readme file–BIG N — GGP Code and Dataset
BEGO- A fully Bayesian Efficient Global Optimization Algorithm
This set of zipped MATLAB .m functions implement the EGO method described in the paper “A Fully Bayesian Approach to Efficient Global Optimization Algorithm” by S. Tajbakhsh, E. del Castillo, and J.L. Rosenberger.
— README_BEGO — BayesianEGOCode
SSA (Statistical Shape Analysis tools)
This is a set of MATLAB functions that implement algorithms for the statistical analysis of shapes. Included are algorithms for 2-D Procrustes registration, solution to the matching problem between 2 shapes, analysis of landmark variance in tangent space, and analysis of 2 factor experiments when the responses are geometrical shapes. See the included README file. Some functions require the Statistics and Optimization toolboxes.
STRPD (Spatial-Temporal Robust Parameter Design): Bayesian Optimization of a Functional Response System using Spatiotemporal Gaussian Processes
A zipped set of MATLAB functions that implements the bayesian model fitting, covariance modeling (including a check for separability) and profile response bayesian optimization methods in the paper by H. Alshraideh and E. del Castillo, “Gaussian Process Modeling and Optimization of Profile Response Experiments”, Quality & Reliability Engineering Int., 30(4), 2014. This type of model provides more flexible shapes of the profiles than the random effects model and is recommended when the randoms effects model does not fit well. A GUI running within Matlab is provided. Requires Matlab’s Global Optimization toolbox.
Bayesian Optimization of a Functional Response System using a Random Effects Models
A set of zipped MATLAB functions that implement the methods described in the paper by E. Del Castillo, B. M. Colosimo and H. Alshraideh, “Bayesian Modeling and Robust Optimization of Functional Responses affected by Noise Factors”, JQT, 44, 2, 2012. See the included README file. Some functions require the Statistics and Optimization toolboxes.
— BayesianOptimizationPrograms
MuBoundS (Multivariate bounded adjustment under fixed adjustment costs)
MuBoundS is a software that computes optimal bounded adjustment parameters for a multivariate bounded adjustment setting, either in the presence of a fixed cost of adjustment or, in the absence of cost information, with the help of Mean Squared Deviation (MSD) and Average Adjustment Interval (AAI) values. It accompanies the paper “An Approach to Multivariate Bounded Process Adjustment” by N. Govind et al. (ESAMLab technical paper, 2016).
–Readme file (txt) — MATLAB code
DUALCONTROL (multiple response sequential optimizer)
Dual Control algorithm which accompanies the paper “A Matrix-T Approach to the Sequential Design of Optimization Experiments” by E.del Castillo and E. Santiago (IIE Transactions, 2010). This MATLAB program will reproduce all examples in the paper; different processes are simulated and optimized on-line (note computing times may be long for repeated simulation/optimizations). For a stand-alone version to use in practice, modifications will be needed: instead of simulating “y” in the main loop, enter the observed responses in vector y and iterate manually.). Copy the following .m program into a folder in your Matlab path and type “help DualControl”. The program requires the optimization and statistics toolboxes.
GADOE (Genetic Algorithm for Design of RPD Experiments)
This set of MATLAB .m programs find a DOE for Robust Parameter Design using a genetic optimization algorithm and accompanies the paper “A New Design Criterion for Robust Parameter Design” by E. del Castillo et al. (JQT, 2007). Extract all files into a folder on the MATLAB path. Type “help optimizeDOE.m” to see calling instructions (that is the initial program which calls all other functions). — GADOE
ADC (Adaptive Deadband Control)
The objective of this MATLAB program is to assist users to adjust a production process, in which the mean of the quality characteristic can drift randomly, in the presence of a fixed adjustment cost. The discussion on this method can be found in paper “Adaptive Deadband Control of a Drifting Process with Unknown Parameters” by Z. Lian and E. del Castillo (Statistics & Probability Letters, 2007).
— User’s manual — AdaptiveDeadband (3 Zipped files)
SAUFAC (Setup Adjustment With Fixed Adjustment Cost)
SAUFAC is a computer program based on the paper on “Setup Adjustment Under Unknown Process Parameters and Fixed Adjustment Cost” by Z. Lian and E. del Castillo (J. of Statistical Planning and Inference, 2006). It computes the control tables for two cases of setup adjustment problems under fixed adjustment cost, unknown-variance case (chart.R) and known-variance case (chart_n.R). A brief user’s manual and the complete package are provided.
— Link to chart.R — Link to chart.R — User’s Manual — SAUFAC package Zip files
SMControl (Sequential Monte Carlo Setup adjustment for Multiple Lots)
SMControl is a MATLAB program for the solution of the multiple lot setup adjustment problem when the process parameters are unknown. It accompanies the paper “Setup Adjustment of Multiple Lots Using a Sequential Monte Carlo Method” (Z. Lian et al, Technometrics, 2006).
— Read me/user’s manuals — Matlab .m programs (zipped)
CONREG (Confidence Region computation of optimal solutions in constrained RSM problems)
CONREG computes and displays the confidence regions for constrained optima of regression models. It accompanies the paper “A General Approach to Confidence Regions for Optimal Factor Levels of Response Surfaces” by J. Peterson et al. (Biometrics, 2002), and “Computation of Confidence Regions for Optimal Factor Levels in Constrained Response Surface Problems“, by S. Cahya et al (J. of Computational and Graphical Statistics, 2004). The program runs under MATLAB version 5.3. For installation and step-by-step on how to run the program, consult the user’s guide.
— User’s manual in pdf — CONREG
BH (Box-Hunter confidence region computation for unconstrained stationary points in RSM)
The BH program is a software tool for computation and display of confidence regions on the location of the stationary point of a quadratic response surface. It accompanies the paper “A Tool for Computing Confidence Regions on the Stationary Point of a Response Surface” by E. del Castillo and S. Cahya (The American Statistician, 2001). The program runs under MAPLE versions 5, 6, and now it also runs in v. 12. For installation and step-by-step on how to run the program, consult the user’s guide.
— User’s guide (MS Word) — Maple V — Maple VI — Maple 12 (NEW, Feb. 2011)