applications of an adjoint...
TRANSCRIPT
![Page 1: Applications of an Adjoint Capabilityflowhead.sems.qmul.ac.uk/events/conference2012/downloads/prese… · Current best (NASA-Cart3d) 32 With permission, from: M. Nemec, M. Aftosmis,](https://reader033.vdocuments.mx/reader033/viewer/2022060606/605b6e51f9e4604740056a2e/html5/thumbnails/1.jpg)
Applications of an Adjoint Capability
1
Richard Dwight, Jouke de Baar, Dirk-Jan Boon,
Hester Bijl, Fulvio Scarano
Aerodynamics Group TU Delft
![Page 2: Applications of an Adjoint Capabilityflowhead.sems.qmul.ac.uk/events/conference2012/downloads/prese… · Current best (NASA-Cart3d) 32 With permission, from: M. Nemec, M. Aftosmis,](https://reader033.vdocuments.mx/reader033/viewer/2022060606/605b6e51f9e4604740056a2e/html5/thumbnails/2.jpg)
Adjoint applications
1. Adjoint solution fields
2. Applications
• Error estimation and goal-oriented adaptation
• Response surfaces in high-dimensional spaces
2
• Response surfaces in high-dimensional spaces
• Kriging
• Global optimization
• Uncertainty quantification
• 4D-var data assimilation
• Frequency-domain solvers
![Page 3: Applications of an Adjoint Capabilityflowhead.sems.qmul.ac.uk/events/conference2012/downloads/prese… · Current best (NASA-Cart3d) 32 With permission, from: M. Nemec, M. Aftosmis,](https://reader033.vdocuments.mx/reader033/viewer/2022060606/605b6e51f9e4604740056a2e/html5/thumbnails/3.jpg)
Adjoint solution fields
• Consider the PDE-constrained minimization:
• subject to
3
with W the flow variables, X the mesh and D the design variables.
• Goal: determine the derivatives of I with respect to D.• We define the Lagrangian which is identical to I and its derivatives
with respect to the design variables D:
![Page 4: Applications of an Adjoint Capabilityflowhead.sems.qmul.ac.uk/events/conference2012/downloads/prese… · Current best (NASA-Cart3d) 32 With permission, from: M. Nemec, M. Aftosmis,](https://reader033.vdocuments.mx/reader033/viewer/2022060606/605b6e51f9e4604740056a2e/html5/thumbnails/4.jpg)
Supersonic NACA0012, Mach=1.5Information transport in hyperbolic problems
4
![Page 5: Applications of an Adjoint Capabilityflowhead.sems.qmul.ac.uk/events/conference2012/downloads/prese… · Current best (NASA-Cart3d) 32 With permission, from: M. Nemec, M. Aftosmis,](https://reader033.vdocuments.mx/reader033/viewer/2022060606/605b6e51f9e4604740056a2e/html5/thumbnails/5.jpg)
Supersonic NACA0012 PrimalDesign parameter = Mach number
5
![Page 6: Applications of an Adjoint Capabilityflowhead.sems.qmul.ac.uk/events/conference2012/downloads/prese… · Current best (NASA-Cart3d) 32 With permission, from: M. Nemec, M. Aftosmis,](https://reader033.vdocuments.mx/reader033/viewer/2022060606/605b6e51f9e4604740056a2e/html5/thumbnails/6.jpg)
Supersonic NACA0012 AdjointCost function = Stagnation pressure
6
![Page 7: Applications of an Adjoint Capabilityflowhead.sems.qmul.ac.uk/events/conference2012/downloads/prese… · Current best (NASA-Cart3d) 32 With permission, from: M. Nemec, M. Aftosmis,](https://reader033.vdocuments.mx/reader033/viewer/2022060606/605b6e51f9e4604740056a2e/html5/thumbnails/7.jpg)
Supersonic NACA0012 AdjointCost function = Point pressure
7
![Page 8: Applications of an Adjoint Capabilityflowhead.sems.qmul.ac.uk/events/conference2012/downloads/prese… · Current best (NASA-Cart3d) 32 With permission, from: M. Nemec, M. Aftosmis,](https://reader033.vdocuments.mx/reader033/viewer/2022060606/605b6e51f9e4604740056a2e/html5/thumbnails/8.jpg)
Supersonic NACA0012 AdjointCost function = Trailing edge pressure
8
![Page 9: Applications of an Adjoint Capabilityflowhead.sems.qmul.ac.uk/events/conference2012/downloads/prese… · Current best (NASA-Cart3d) 32 With permission, from: M. Nemec, M. Aftosmis,](https://reader033.vdocuments.mx/reader033/viewer/2022060606/605b6e51f9e4604740056a2e/html5/thumbnails/9.jpg)
Supersonic NACA0012 AdjointCost function = Drag
9
![Page 10: Applications of an Adjoint Capabilityflowhead.sems.qmul.ac.uk/events/conference2012/downloads/prese… · Current best (NASA-Cart3d) 32 With permission, from: M. Nemec, M. Aftosmis,](https://reader033.vdocuments.mx/reader033/viewer/2022060606/605b6e51f9e4604740056a2e/html5/thumbnails/10.jpg)
3d High-lift Gradient EvaluationAdjoint Field
10
![Page 11: Applications of an Adjoint Capabilityflowhead.sems.qmul.ac.uk/events/conference2012/downloads/prese… · Current best (NASA-Cart3d) 32 With permission, from: M. Nemec, M. Aftosmis,](https://reader033.vdocuments.mx/reader033/viewer/2022060606/605b6e51f9e4604740056a2e/html5/thumbnails/11.jpg)
3d High-lift Gradient Evaluation Adjoint Field
11
![Page 12: Applications of an Adjoint Capabilityflowhead.sems.qmul.ac.uk/events/conference2012/downloads/prese… · Current best (NASA-Cart3d) 32 With permission, from: M. Nemec, M. Aftosmis,](https://reader033.vdocuments.mx/reader033/viewer/2022060606/605b6e51f9e4604740056a2e/html5/thumbnails/12.jpg)
3d High-lift Gradient EvaluationAdjoint Field
12
![Page 13: Applications of an Adjoint Capabilityflowhead.sems.qmul.ac.uk/events/conference2012/downloads/prese… · Current best (NASA-Cart3d) 32 With permission, from: M. Nemec, M. Aftosmis,](https://reader033.vdocuments.mx/reader033/viewer/2022060606/605b6e51f9e4604740056a2e/html5/thumbnails/13.jpg)
Overview of applications – as of today
• Shape optimization (Pironneau 1973, Jameson 1988, Eleshaky 1997)
• Efficient global optimization (EGO) (Ely 2004, Ghattas 2009)
• Error estimation• A posteriori (Giles 1997, Becker and Rannacher 1998)
• Error transport equation (Roache 1998, Cesnik 2002)
• Goal-oriented mesh adaptation (Darmofal 2004, Barth 2005, Dwight 2007)
• Frequency-domain methods• Linear frequency-domain solver (Campobasso 2007)
• Schur’s-complement eigenvalue solver (Badcock 2010)
13
• Schur’s-complement eigenvalue solver (Badcock 2010)
• Uncertainty quantification• First-order moment methods
• Parameter reduction (Loeven 2009)
• Surrogate modelling (Mavriplis 2010, Dwight 2010)
• Inverse problems + data assimilation• Deterministic (4D-var, Least-squares) (Dwight 2011)
• Stochastic (McMC, Hessian) (Fox 2009, Ghattas 2010)
• Petrov-Galerkin Reduced-order modelling (ROMs) (Farhat 2010)
![Page 14: Applications of an Adjoint Capabilityflowhead.sems.qmul.ac.uk/events/conference2012/downloads/prese… · Current best (NASA-Cart3d) 32 With permission, from: M. Nemec, M. Aftosmis,](https://reader033.vdocuments.mx/reader033/viewer/2022060606/605b6e51f9e4604740056a2e/html5/thumbnails/14.jpg)
Error estimation + Goal-oriented adaptation
Problem: Local gradient adaptation doesn‘t work for hyperbolic
problems (e.g. Euler equations):
14
![Page 15: Applications of an Adjoint Capabilityflowhead.sems.qmul.ac.uk/events/conference2012/downloads/prese… · Current best (NASA-Cart3d) 32 With permission, from: M. Nemec, M. Aftosmis,](https://reader033.vdocuments.mx/reader033/viewer/2022060606/605b6e51f9e4604740056a2e/html5/thumbnails/15.jpg)
Goal-oriented adaptation: Idea
Flow Adjoint
15
k(2) . dCD/dk(2)k(4)dCD/dk(4)
![Page 16: Applications of an Adjoint Capabilityflowhead.sems.qmul.ac.uk/events/conference2012/downloads/prese… · Current best (NASA-Cart3d) 32 With permission, from: M. Nemec, M. Aftosmis,](https://reader033.vdocuments.mx/reader033/viewer/2022060606/605b6e51f9e4604740056a2e/html5/thumbnails/16.jpg)
A Posteriori Error EstimationError representation for linear problems
• Consider the linear problem:
• and let the functional of interest be of the form:
16
• where the inner product may be defined by e.g.:
![Page 17: Applications of an Adjoint Capabilityflowhead.sems.qmul.ac.uk/events/conference2012/downloads/prese… · Current best (NASA-Cart3d) 32 With permission, from: M. Nemec, M. Aftosmis,](https://reader033.vdocuments.mx/reader033/viewer/2022060606/605b6e51f9e4604740056a2e/html5/thumbnails/17.jpg)
A Posteriori Error EstimationError representation for linear problems
• Also define the adjoint operator L* of L as:
• giving the adjoint problem:
17
• Now for a linear problem the error satisfies the same eqn as the
solution:
• where:
![Page 18: Applications of an Adjoint Capabilityflowhead.sems.qmul.ac.uk/events/conference2012/downloads/prese… · Current best (NASA-Cart3d) 32 With permission, from: M. Nemec, M. Aftosmis,](https://reader033.vdocuments.mx/reader033/viewer/2022060606/605b6e51f9e4604740056a2e/html5/thumbnails/18.jpg)
A Posteriori Error Estimation Error representation for linear problems
• But the error in J can also be repesented in terms of the solution
of the adjoint problem:
18
• So two formulations for the error are:
• Compare with the dual equivalence relation:
![Page 19: Applications of an Adjoint Capabilityflowhead.sems.qmul.ac.uk/events/conference2012/downloads/prese… · Current best (NASA-Cart3d) 32 With permission, from: M. Nemec, M. Aftosmis,](https://reader033.vdocuments.mx/reader033/viewer/2022060606/605b6e51f9e4604740056a2e/html5/thumbnails/19.jpg)
Convergence of error: Drag
19
![Page 20: Applications of an Adjoint Capabilityflowhead.sems.qmul.ac.uk/events/conference2012/downloads/prese… · Current best (NASA-Cart3d) 32 With permission, from: M. Nemec, M. Aftosmis,](https://reader033.vdocuments.mx/reader033/viewer/2022060606/605b6e51f9e4604740056a2e/html5/thumbnails/20.jpg)
Location of error: Drag
20
![Page 21: Applications of an Adjoint Capabilityflowhead.sems.qmul.ac.uk/events/conference2012/downloads/prese… · Current best (NASA-Cart3d) 32 With permission, from: M. Nemec, M. Aftosmis,](https://reader033.vdocuments.mx/reader033/viewer/2022060606/605b6e51f9e4604740056a2e/html5/thumbnails/21.jpg)
Convergence of error: Lift
21
![Page 22: Applications of an Adjoint Capabilityflowhead.sems.qmul.ac.uk/events/conference2012/downloads/prese… · Current best (NASA-Cart3d) 32 With permission, from: M. Nemec, M. Aftosmis,](https://reader033.vdocuments.mx/reader033/viewer/2022060606/605b6e51f9e4604740056a2e/html5/thumbnails/22.jpg)
Location of error: Lift
22
![Page 23: Applications of an Adjoint Capabilityflowhead.sems.qmul.ac.uk/events/conference2012/downloads/prese… · Current best (NASA-Cart3d) 32 With permission, from: M. Nemec, M. Aftosmis,](https://reader033.vdocuments.mx/reader033/viewer/2022060606/605b6e51f9e4604740056a2e/html5/thumbnails/23.jpg)
Location of error: Lift
23
![Page 24: Applications of an Adjoint Capabilityflowhead.sems.qmul.ac.uk/events/conference2012/downloads/prese… · Current best (NASA-Cart3d) 32 With permission, from: M. Nemec, M. Aftosmis,](https://reader033.vdocuments.mx/reader033/viewer/2022060606/605b6e51f9e4604740056a2e/html5/thumbnails/24.jpg)
Location of error: Lift
24
![Page 25: Applications of an Adjoint Capabilityflowhead.sems.qmul.ac.uk/events/conference2012/downloads/prese… · Current best (NASA-Cart3d) 32 With permission, from: M. Nemec, M. Aftosmis,](https://reader033.vdocuments.mx/reader033/viewer/2022060606/605b6e51f9e4604740056a2e/html5/thumbnails/25.jpg)
Location of error: Lift
25
![Page 26: Applications of an Adjoint Capabilityflowhead.sems.qmul.ac.uk/events/conference2012/downloads/prese… · Current best (NASA-Cart3d) 32 With permission, from: M. Nemec, M. Aftosmis,](https://reader033.vdocuments.mx/reader033/viewer/2022060606/605b6e51f9e4604740056a2e/html5/thumbnails/26.jpg)
Convergence of all cases: Lift
26
![Page 27: Applications of an Adjoint Capabilityflowhead.sems.qmul.ac.uk/events/conference2012/downloads/prese… · Current best (NASA-Cart3d) 32 With permission, from: M. Nemec, M. Aftosmis,](https://reader033.vdocuments.mx/reader033/viewer/2022060606/605b6e51f9e4604740056a2e/html5/thumbnails/27.jpg)
Convergence of error: ONERA M6
27
![Page 28: Applications of an Adjoint Capabilityflowhead.sems.qmul.ac.uk/events/conference2012/downloads/prese… · Current best (NASA-Cart3d) 32 With permission, from: M. Nemec, M. Aftosmis,](https://reader033.vdocuments.mx/reader033/viewer/2022060606/605b6e51f9e4604740056a2e/html5/thumbnails/28.jpg)
Convergence of error: ONERA M6
28
![Page 29: Applications of an Adjoint Capabilityflowhead.sems.qmul.ac.uk/events/conference2012/downloads/prese… · Current best (NASA-Cart3d) 32 With permission, from: M. Nemec, M. Aftosmis,](https://reader033.vdocuments.mx/reader033/viewer/2022060606/605b6e51f9e4604740056a2e/html5/thumbnails/29.jpg)
Error Transport Equation
Grid 5
29
![Page 30: Applications of an Adjoint Capabilityflowhead.sems.qmul.ac.uk/events/conference2012/downloads/prese… · Current best (NASA-Cart3d) 32 With permission, from: M. Nemec, M. Aftosmis,](https://reader033.vdocuments.mx/reader033/viewer/2022060606/605b6e51f9e4604740056a2e/html5/thumbnails/30.jpg)
Error Transport Equation: RAE2822
30
![Page 31: Applications of an Adjoint Capabilityflowhead.sems.qmul.ac.uk/events/conference2012/downloads/prese… · Current best (NASA-Cart3d) 32 With permission, from: M. Nemec, M. Aftosmis,](https://reader033.vdocuments.mx/reader033/viewer/2022060606/605b6e51f9e4604740056a2e/html5/thumbnails/31.jpg)
Error estimate: Lift and Drag
31
![Page 32: Applications of an Adjoint Capabilityflowhead.sems.qmul.ac.uk/events/conference2012/downloads/prese… · Current best (NASA-Cart3d) 32 With permission, from: M. Nemec, M. Aftosmis,](https://reader033.vdocuments.mx/reader033/viewer/2022060606/605b6e51f9e4604740056a2e/html5/thumbnails/32.jpg)
Current best (NASA-Cart3d)
32
With permission, from: M. Nemec, M. Aftosmis, “Adjoint-Based Adaptive
Mesh Refinement for Complex Geometries”, AIAA-2008-0725, 2008.
![Page 33: Applications of an Adjoint Capabilityflowhead.sems.qmul.ac.uk/events/conference2012/downloads/prese… · Current best (NASA-Cart3d) 32 With permission, from: M. Nemec, M. Aftosmis,](https://reader033.vdocuments.mx/reader033/viewer/2022060606/605b6e51f9e4604740056a2e/html5/thumbnails/33.jpg)
Response surfaces with imprecise gradientsUncertainty quantification
33
![Page 34: Applications of an Adjoint Capabilityflowhead.sems.qmul.ac.uk/events/conference2012/downloads/prese… · Current best (NASA-Cart3d) 32 With permission, from: M. Nemec, M. Aftosmis,](https://reader033.vdocuments.mx/reader033/viewer/2022060606/605b6e51f9e4604740056a2e/html5/thumbnails/34.jpg)
34
![Page 35: Applications of an Adjoint Capabilityflowhead.sems.qmul.ac.uk/events/conference2012/downloads/prese… · Current best (NASA-Cart3d) 32 With permission, from: M. Nemec, M. Aftosmis,](https://reader033.vdocuments.mx/reader033/viewer/2022060606/605b6e51f9e4604740056a2e/html5/thumbnails/35.jpg)
Shocks
35
![Page 36: Applications of an Adjoint Capabilityflowhead.sems.qmul.ac.uk/events/conference2012/downloads/prese… · Current best (NASA-Cart3d) 32 With permission, from: M. Nemec, M. Aftosmis,](https://reader033.vdocuments.mx/reader033/viewer/2022060606/605b6e51f9e4604740056a2e/html5/thumbnails/36.jpg)
Shocks
36
![Page 37: Applications of an Adjoint Capabilityflowhead.sems.qmul.ac.uk/events/conference2012/downloads/prese… · Current best (NASA-Cart3d) 32 With permission, from: M. Nemec, M. Aftosmis,](https://reader033.vdocuments.mx/reader033/viewer/2022060606/605b6e51f9e4604740056a2e/html5/thumbnails/37.jpg)
Aside: Conditional probabilityBayes’ theorem
Combining a:
37
Combining a:
1. Prior belief on parameters.
2. Likelihood of observations.
Bayes’ theorem gives a posterior (updated belief) on parameters.
![Page 38: Applications of an Adjoint Capabilityflowhead.sems.qmul.ac.uk/events/conference2012/downloads/prese… · Current best (NASA-Cart3d) 32 With permission, from: M. Nemec, M. Aftosmis,](https://reader033.vdocuments.mx/reader033/viewer/2022060606/605b6e51f9e4604740056a2e/html5/thumbnails/38.jpg)
Uncertainty Reduction with Bayes
Answer!Prior: best guess atinput uncertainties
38
Answer! input uncertainties
Statistical model: relates thesimulation to the data – e.g.:
![Page 39: Applications of an Adjoint Capabilityflowhead.sems.qmul.ac.uk/events/conference2012/downloads/prese… · Current best (NASA-Cart3d) 32 With permission, from: M. Nemec, M. Aftosmis,](https://reader033.vdocuments.mx/reader033/viewer/2022060606/605b6e51f9e4604740056a2e/html5/thumbnails/39.jpg)
Likelihood Prior Posterior
Uncertainty Reduction with Bayes
39
![Page 40: Applications of an Adjoint Capabilityflowhead.sems.qmul.ac.uk/events/conference2012/downloads/prese… · Current best (NASA-Cart3d) 32 With permission, from: M. Nemec, M. Aftosmis,](https://reader033.vdocuments.mx/reader033/viewer/2022060606/605b6e51f9e4604740056a2e/html5/thumbnails/40.jpg)
Kriging interpolation
Data from pedagogical windtunnel tests
40Results from group 19 (2011)
![Page 41: Applications of an Adjoint Capabilityflowhead.sems.qmul.ac.uk/events/conference2012/downloads/prese… · Current best (NASA-Cart3d) 32 With permission, from: M. Nemec, M. Aftosmis,](https://reader033.vdocuments.mx/reader033/viewer/2022060606/605b6e51f9e4604740056a2e/html5/thumbnails/41.jpg)
Kriging interpolation
Interpolant with uncertainty estimate
41Results from group 19 (2011)
![Page 42: Applications of an Adjoint Capabilityflowhead.sems.qmul.ac.uk/events/conference2012/downloads/prese… · Current best (NASA-Cart3d) 32 With permission, from: M. Nemec, M. Aftosmis,](https://reader033.vdocuments.mx/reader033/viewer/2022060606/605b6e51f9e4604740056a2e/html5/thumbnails/42.jpg)
Kriging – From a Bayesian PerspectiveMultivariate normal distributions
42
![Page 43: Applications of an Adjoint Capabilityflowhead.sems.qmul.ac.uk/events/conference2012/downloads/prese… · Current best (NASA-Cart3d) 32 With permission, from: M. Nemec, M. Aftosmis,](https://reader033.vdocuments.mx/reader033/viewer/2022060606/605b6e51f9e4604740056a2e/html5/thumbnails/43.jpg)
Kriging – From a Bayesian PerspectiveMultivariate normal distributions
43
![Page 44: Applications of an Adjoint Capabilityflowhead.sems.qmul.ac.uk/events/conference2012/downloads/prese… · Current best (NASA-Cart3d) 32 With permission, from: M. Nemec, M. Aftosmis,](https://reader033.vdocuments.mx/reader033/viewer/2022060606/605b6e51f9e4604740056a2e/html5/thumbnails/44.jpg)
Kriging – A Bayesian PerspectiveMultivariate normal distributions
44
![Page 45: Applications of an Adjoint Capabilityflowhead.sems.qmul.ac.uk/events/conference2012/downloads/prese… · Current best (NASA-Cart3d) 32 With permission, from: M. Nemec, M. Aftosmis,](https://reader033.vdocuments.mx/reader033/viewer/2022060606/605b6e51f9e4604740056a2e/html5/thumbnails/45.jpg)
Kriging – From a Bayesian PerspectiveIncorporating observations
45
![Page 46: Applications of an Adjoint Capabilityflowhead.sems.qmul.ac.uk/events/conference2012/downloads/prese… · Current best (NASA-Cart3d) 32 With permission, from: M. Nemec, M. Aftosmis,](https://reader033.vdocuments.mx/reader033/viewer/2022060606/605b6e51f9e4604740056a2e/html5/thumbnails/46.jpg)
Kriging – From a Bayesian PerspectiveIncorporating observations
46
![Page 47: Applications of an Adjoint Capabilityflowhead.sems.qmul.ac.uk/events/conference2012/downloads/prese… · Current best (NASA-Cart3d) 32 With permission, from: M. Nemec, M. Aftosmis,](https://reader033.vdocuments.mx/reader033/viewer/2022060606/605b6e51f9e4604740056a2e/html5/thumbnails/47.jpg)
Kriging – From a Bayesian PerspectiveIncorporating observations
47
![Page 48: Applications of an Adjoint Capabilityflowhead.sems.qmul.ac.uk/events/conference2012/downloads/prese… · Current best (NASA-Cart3d) 32 With permission, from: M. Nemec, M. Aftosmis,](https://reader033.vdocuments.mx/reader033/viewer/2022060606/605b6e51f9e4604740056a2e/html5/thumbnails/48.jpg)
Kriging – From a Bayesian PerspectiveIncorporating observations
48
![Page 49: Applications of an Adjoint Capabilityflowhead.sems.qmul.ac.uk/events/conference2012/downloads/prese… · Current best (NASA-Cart3d) 32 With permission, from: M. Nemec, M. Aftosmis,](https://reader033.vdocuments.mx/reader033/viewer/2022060606/605b6e51f9e4604740056a2e/html5/thumbnails/49.jpg)
Posterior distribution = Kriging surfaceMean +- 1 sd
49
![Page 50: Applications of an Adjoint Capabilityflowhead.sems.qmul.ac.uk/events/conference2012/downloads/prese… · Current best (NASA-Cart3d) 32 With permission, from: M. Nemec, M. Aftosmis,](https://reader033.vdocuments.mx/reader033/viewer/2022060606/605b6e51f9e4604740056a2e/html5/thumbnails/50.jpg)
Incorporating gradients
50
![Page 51: Applications of an Adjoint Capabilityflowhead.sems.qmul.ac.uk/events/conference2012/downloads/prese… · Current best (NASA-Cart3d) 32 With permission, from: M. Nemec, M. Aftosmis,](https://reader033.vdocuments.mx/reader033/viewer/2022060606/605b6e51f9e4604740056a2e/html5/thumbnails/51.jpg)
Gradient informationPrior on the gradients
• Augment unknown value vector with unknown gradients:
• Prior is still a multivariate Gaussian (i.e. discrete random
field)
51
• So:
![Page 52: Applications of an Adjoint Capabilityflowhead.sems.qmul.ac.uk/events/conference2012/downloads/prese… · Current best (NASA-Cart3d) 32 With permission, from: M. Nemec, M. Aftosmis,](https://reader033.vdocuments.mx/reader033/viewer/2022060606/605b6e51f9e4604740056a2e/html5/thumbnails/52.jpg)
Gradient informationDefining the covariance matrix
• Value and gradient are not correlated at a point
⇒ can not apply Gaussian
52
![Page 53: Applications of an Adjoint Capabilityflowhead.sems.qmul.ac.uk/events/conference2012/downloads/prese… · Current best (NASA-Cart3d) 32 With permission, from: M. Nemec, M. Aftosmis,](https://reader033.vdocuments.mx/reader033/viewer/2022060606/605b6e51f9e4604740056a2e/html5/thumbnails/53.jpg)
Apply Bayes to this priorExactly as in previous case
53
![Page 54: Applications of an Adjoint Capabilityflowhead.sems.qmul.ac.uk/events/conference2012/downloads/prese… · Current best (NASA-Cart3d) 32 With permission, from: M. Nemec, M. Aftosmis,](https://reader033.vdocuments.mx/reader033/viewer/2022060606/605b6e51f9e4604740056a2e/html5/thumbnails/54.jpg)
Posterior = Kriging surface
54
![Page 55: Applications of an Adjoint Capabilityflowhead.sems.qmul.ac.uk/events/conference2012/downloads/prese… · Current best (NASA-Cart3d) 32 With permission, from: M. Nemec, M. Aftosmis,](https://reader033.vdocuments.mx/reader033/viewer/2022060606/605b6e51f9e4604740056a2e/html5/thumbnails/55.jpg)
Estimating gradient errorMLE for gradient error
• We don’t know the gradient error.
• What do we do when we don’t know something?
55
• We apply Bayes and hope that what we do know informs
what we don’t know!
![Page 56: Applications of an Adjoint Capabilityflowhead.sems.qmul.ac.uk/events/conference2012/downloads/prese… · Current best (NASA-Cart3d) 32 With permission, from: M. Nemec, M. Aftosmis,](https://reader033.vdocuments.mx/reader033/viewer/2022060606/605b6e51f9e4604740056a2e/html5/thumbnails/56.jpg)
Estimating gradient error
56
de Baar, Dwight, Bijl (2012), Improvements to Gradient-Enhanced Kriging using a Bayesian Interpretation. Technometrics (submitted).
![Page 57: Applications of an Adjoint Capabilityflowhead.sems.qmul.ac.uk/events/conference2012/downloads/prese… · Current best (NASA-Cart3d) 32 With permission, from: M. Nemec, M. Aftosmis,](https://reader033.vdocuments.mx/reader033/viewer/2022060606/605b6e51f9e4604740056a2e/html5/thumbnails/57.jpg)
Uncertainty quantification accuracy
57
de Baar, Dwight, Bijl (2012), Improvements to Gradient-Enhanced Kriging using a Bayesian Interpretation. Technometrics (submitted).
![Page 58: Applications of an Adjoint Capabilityflowhead.sems.qmul.ac.uk/events/conference2012/downloads/prese… · Current best (NASA-Cart3d) 32 With permission, from: M. Nemec, M. Aftosmis,](https://reader033.vdocuments.mx/reader033/viewer/2022060606/605b6e51f9e4604740056a2e/html5/thumbnails/58.jpg)
Thank you
58
Thank youfor your attention