Improvement of adjoint based methods for efficient computation of response surfaces

Size: px
Start display at page:

Download "Improvement of adjoint based methods for efficient computation of response surfaces"

Transcription

1 Center for Turbulence Researc Annual Researc Briefs Improvement of adjoint based metods for efficient computation of response surfaces By F. Palacios, K. Duraisamy AND J. J. Alonso 1. Motivation and objectives In computational fluid dynamics, te solution of te dual or adjoint) equations in conjunction wit te primal or flow equations) offers valuable information tat can be used in design, sensitivity analysis, error estimation, mes adaptation, uncertainty quantification, etc. From te view point of implementation and application of te adjoint tecnique, one of te following two approaces may be pursued: te continuous metod and te discrete one. In te continuous approac Jameson 1988; Anderson & Venkatakrisnan 1997; Kim et al. 2002; Castro et al. 2007), te adjoint equations are derived from te continuous flow equations and ten subsequently discretized using a numerical metod of coice, wereas in te discrete approac Griewank 2000; Nadaraja & Jameson 2000; Giles et al. 2003; Mavriplis 2006; Mader et al. 2008), te adjoint equations are directly derived from te discretized governing equations. Tese approaces offer distinct advantages in different scenarios and ave bot been implemented for purposes of Verification and Validation V&V) and Uncertainty Quantification UQ) in ypersonic flow applications. In tis work, we outline preliminary work in two areas of researc tat involve te use of adjoints in uncertainty quantification: Uncertainty quantification using non-intrusive metods suc as stocastic collocation Eldred 2009) typically involves te construction of a response surface of te quantity of interest as a function of te uncertain parameters. To acieve te level of desired accuracy, a large number of Computational Fluid Dynamics CFD) simulations is typically required. To improve te efficiency of tis process, we propose a new robust grid adaptation tecnique tat is aimed at minimizing te numerical error over a small variation of te parameters around a nominal state. In tis approac, computational grids are generated wit te knowledge tat small variations of te uncertain parameters can locally cange te solution and ence, te error distribution). Tis is in contrast to classical adjoint tecniques Pierce & Giles 2002; Fidkowski & Darmofal 2009), wic seek to adapt te grid wit te aim of minimizing numerical errors for a nominal flow condition. In te discrete sensitivity framework suc as discrete adjoints, complex step, and finite differences), te grid convergence of nonlinear flow calculations does not guarantee te convergence of linear sensitivities Giles et al. 2003). Tis attribute as a special relevance in problems involving strong socks, were te gradient of an output functional of interest may exibit noise, and ence, cannot be directly utilized. To counter tis difficulty, we utilize a bi-grid filtering strategy Glowinski 1992; Zuazua 2005) tat offers te promise of converged linear sensitivities using a discrete computation tecnique. In summary, te pursuit of UQ science in ypersonic flow applications, togeter wit sensitivity computations using adjoints, as brougt to te table some interesting limitations of te state-of-te-art in adjoint metodologies. In tis report, our early work FuSim-E project, AIRBUS Spain.

2 396 F. Palacios, K. Duraisamy and J. J. Alonso in two novel metodologies is introduced and some preliminary numerical results are presented. 2. Application of adjoint tecniques to enance te computation of response surfaces In tis section, we present a brief introduction to te teoretical basis of te metods tat ave been developed to improve and accelerate te computation of response surfaces, taking advantage of te adjoint metodology to minimize te numerical error incurred wen variations of uncertain input parameters are considered New robust grid metodology A response surface represents te explicit dependence of a desired output of a simulation model as a function of all possible values of te decision/input variables. Te response surface Cung & Alonso 2001; Cantrasmi & Iaccarino 2009) is a powerful tool in UQ, altoug, in practice, a large number of samples is required. Typically tese samples constitute small variations of te parameters of interest in some local neigborood. We seek to develop a new grid adaptation strategy for creating wat we call robust computational grids. Flow computations on tese grids must be able to result in low numerical error even if we sligtly cange te nominal flow conditions suc as te Mac number, geometry, pysical constants, etc.) tat were used in generating te original adapted grid. Te objective is to perform a single grid adaptation about a nominal value of te parameters and to ten use tis grid for many additional simulations wile guaranteeing tat te numerical error is eld constant. A baseline grid tat exibits tis property under reasonable variations of te input parameters is called a robust grid. In te following discussion, te basis for error estimation in integral outputs functionals) of partial differential equations Giles & Pierce 2001; Venditti & Darmofal 2002; Park 2003; Fidkowski & Darmofal 2009) are reviewed. It is well known tat tis local error estimate could be used as an adaptation indicator in a grid adaptive strategy for producing specially tuned grids for accurately estimating some functional. Let us consider a fine grid Ω and a coarse grid Ω H. Te goal is to obtain an accurate estimate of some functional f U ) in te fine grid witout solving te flow or adjoint equations on te fine grid. An expansion of f U ) about te coarse-grid solution is f U ) = f ) + f U U H ) + H.O.T., 2.1) were U H represents te coarse-grid solution expressed on te fine-grid via some consistent interpolation operation. Te nonlinear residual operator representing te flow system of equations is denoted by R U U ) = ) Linearizing te residual about te coarse-grid solution yields R U U ) = R U UH ) + RU U ) + H.O.T., 2.3)

3 Adjoint based metods for computation of response surfaces 397 were, neglecting te ig order terms, we can write U U H ) R U = Now we define te linear adjoint residual operator R Ψ Ψ) as ) T R Ψ R U Ψ) Ψ f ) 1 R U ). 2.4), 2.5) were it is important to note tat in tis last expression we are using a prolongated solution U H to evaluate te Jacobian. By definition RΨ Ψ ) = 0, and ence, ) T R U Ψ = f. 2.6) Introducing 2.6) and 2.4) into 2.1) we finally obtain te following expression for te estimate of te numerical error: ǫ U, ) = f ) f U ) = Ψ ) T R U ). 2.7) Unfortunately, to evaluate tis last expression it is necessary to solve te adjoint problem 2.6) on te fine grid. Te next step is to look for a more appropriate way for estimating te error. Defining Ψ H as te coarse-grid adjoint solution expressed on te fine grid via some consistent projection operation, te error can be written as ǫ U, ) = Ψ H ) T R U ) + Ψ Ψ H ) T R U ), 2.8) were te difference Ψ Ψ H ) can be estimated using 2.6), and using RΨ Ψ ) = 0 we finally obtain ǫ U, ) = Ψ H )T R U UH ) + ) 1 R Ψ ΨH )) T R U R U UH ). 2.9) Once again, using 2.4) it is possible to write te error estimation in te following useful form: ǫ U, ) = Ψ H ) T R U ) + R Ψ Ψ H ) ) T U ). 2.10) Tis expression can be used to generate specially tuned grids for accurately estimating te error in calculating a functional. Note tat te first term on te rigt and side of te above equation is known via coarse-grid computations and fine-grid interpolations. Previous work Fidkowski & Darmofal 2009; Venditti & Darmofal 2002) as successfully used tis expression to elp guide te mes adaptation process. Our aim is to introduce an adaptation tecnique tat produces low numerical error for a range of variation in some flow or geometric parameters. Te first step in te new metodology is to define a robust grid as a grid in wic, for small variations of te solution, te numerical error is nearly constant. In oter words, te variation of te numerical error wit respect to variations of a parameter k geometric or flow properties of te problem), must be small or zero in te ideal scenario. Te derivative of te numerical error 2.10) owing to a variation of te parameter k can be evaluated as dǫ dk = dψ H dk ) T R U ) + Ψ H ) T dru UH ) dk

4 398 F. Palacios, K. Duraisamy and J. J. Alonso dr Ψ + Ψ H ) ) T U ) + R Ψ Ψ H ) ) T du ), 2.11) dk dk were te adjoint and te flow solutions are functions of te parameter k. Te value of U ) can be estimated using 2.4) to obtain dǫ dk = ΨH ) T dru UH ) ) + dψ H T dr Ψ Ψ H ) ) T ) 1 R U R U ) dk dk dk du + ) ) T R Ψ Ψ H ). 2.12) dk In te limit of an extremely well adapted mes for te direct and adjoint problems, we expect tat R ΨΨH ) 0 and RU UH ) 0. In tat situation, te variation of te error due to a cange in te parameter k can be computed as dǫ dk = ΨH ) T dru UH ). 2.13) dk Te key idea of tis development is tat if we adapt te computational grid for solving te direct and te adjoint problems independently, te last expression gives us a very easy way to compute te numerical error wen we cange te flow conditions to determine te complete response surface. Furtermore, supposing tat V H is te new solution on te coarse grid, and Ψ H te adjoint solution in te original problem, te correction to te functional value can be computed as ǫv H ) ǫu, ) = Ψ H )T R V V H ) RU UH )). 2.14) Using 2.10), te error of te new solution can be computed as ǫv H ) = Ψ H ) T R V V H ) + R Ψ Ψ H ) ) T U ) Ψ H ) T R V V H ), 2.15) were it is important to note tat tis error does not depend on te finest grid solution. In principle, for a reconstruction-based finite volume sceme, te numerical error is likely to cange even wen te solution is sligtly perturbed by a cange in input conditions. Fortunately, tis error can be estimated using coarse-grid calculations. On te oter and, if we ave Galerkin ortogonality, te first term in 2.10) will be zero and te adapted grid for te adjoint and flow problem) will provide excellent results even wen some flow parameter is sligtly canged. Te adaptation tecnique tat we ave developed is based in obtaining R ΨΨH ) 0 and R UUH ) 0 for te baseline flow solution. To acieve tis objective, it is natural to employ a residual-based indicator to drive te flow and adjoint adaptation process. Eiter of te following approaces can be pursued: A direct approac, wic consists of computing te value of R ΨΨH ) and RU UH ) in a very fine grid and subsequently adapting te original grid in zones were tese values are large. An indirect strategy, in wic te residual is driven to zero in te dual control volumes of te grid. Note tat te original numerical sceme is designed to ave a zero flow residual on te primal control volumes. Once we ave an estimation of te residual for te direct and adjoint problems, it can be used as an indicator for mes adaptation. Wit regard to te refinement process itself Boroucaki et al. 2005), tere is only one element subdivision tat can preserve

5 Adjoint based metods for computation of response surfaces 399 te element sape quality: omotetic subdivision, wic for a planar triangle, splits it into four by adding tree new vertices at te middle of te edges. It as to be noted tat elements tat sare tese edges must also be divided if anging nodes are to be avoided. In te residual based strategy we typically define te error as e = Ω α R 2 ρ + R2 ρ v + R2 ρe, 2.16) were Ω is te area of te control volume, and R i are te four two-dimensional) or five tree-dimensional) residuals at eac node. A widely used formulation utilizes α = 1. Te advantage of scaling te indicator wit a positive value of α is tat te adaptation stops automatically in te corresponding area after several cycles even wen discontinuities are present in te flowfield. Finally an indicator of te error is computed as e lim = e m + c lim e s, 2.17) were e lim is te error limit, e m is te mean of te error indicator, e s is te standard deviation of te error indicator, and c lim is a constant. Typically a value near unity is used for te constant. Te omotetic element refinement is used in grid elements were te error indicator is greater tan a corresponding error tolerance Sensitivity computation on small pressure sensor bi-grid filtering tecnique) It is known tat wit te discrete sensitivity approac automatic differentiation, complex step, and finite differences), te convergence of nonlinear flow equations to macine precision does not guarantee te convergence of functional sensitivities Giles et al. 2003). In particular, we ave observed tat in te presence of strong socks and for functionals tat extend across only a small number of mes cells, tere are significant differences between te gradient of a functional computed wit te continuous and te discrete adjoints or exact finite differences), in tat te latter can be expected to be noisy mainly for functionals wit a small spatial footprint). Te use of automatic differentiation Griewank 2000; Martins et al. 2006; Gauger et al. 2007; Mader et al. 2008) is a very attractive tecnique to tackle derivative computations wen dealing wit complex multi-pysics systems. However, as we will see, in te presence of strong socks te computed gradient can be noisy, and tis situation as critical relevance if we are interested in computing second derivatives using te same metodology. To obtain a useful expression to process te discrete sensitivities, te starting assumption is tat te continuous adjoint metodology will provide a gradient approximation tat will be closer to te analytical gradient in fact, in te limit of a very fine grid it must be te analytical one). Tis as been confirmed by performing numerical experiments in flows wit strong socks. Tus, an attempt will be made at building a discrete adjoint tat mimics te beavior of te continuous adjoint. Once suc a discrete adjoint is constructed, te next step will be to compare tis new discrete adjoint wit te classical one. Consider te stationary Euler equations in a domain Ω: RU) = FU) = 0, in Ω, 2.18) were R is te residual, U is te vector of conservative variables, and F is te convective flux vector. Te Euler equations ave to be completed wit te flow tangency boundary condition at all solid walls, and at te inlet and outlet boundary conditions are specified

6 400 F. Palacios, K. Duraisamy and J. J. Alonso for incoming waves, wile outgoing waves are determined by te solution inside te fluid domain. In a numerical practice, a modified version of tese equations is solved. Tis modified governing equation can be written as RU) + DU) = FU) + ǫ GU) U) = 0, in Ω, 2.19) were te te stabilizing viscosity term can be interpreted as te discretization of a second derivative Gu) u)) for some viscosity function G depending on te particular sceme under consideration, and ǫ contains powers of spatial coordinates. It is important to note tat ǫ sould approac zero as te grid spacing approaces zero. If we define a functional JU, k) tat depends on te conservative variables and a parameter k tat depends on te flow or geometry, te complete variation of te functional is δj = J J δu + δk. 2.20) k Typically, te discrete adjoint problem will be introduced troug te Lagrange multipliers to remove te explicit dependence on δu. It is important to note tat our objective is to compare two discrete adjoints, te classical one and a new discrete adjoint tat mimics te continuous adjoint beavior: In te classical discrete adjoint approac, te modified state equation 2.19) is fulfilled, and we will build te Lagrangian: LU, k, λ D ) = JU) + λ D [RU, k) + DU)]. On te oter and, using ideas inspired in te continuous adjoint problem, we also define te following Lagrangian: LU, k, λ C ) = JU) + λ C [RU, k) + C], were C is a constant wic includes ig-order terms of te discretization tat do not depend on U in te continuous adjoint procedure). Te variation of te objective function J in tese approaces is, terefore δj = J δj = J δu + J k R δk λd + D R J δu + δk λc k ) R δu λ D k ) δu λ C R k ) δk, 2.21) ) δk. 2.22) Te final expressions for te sensitivities old if λ D and λ C satisfy te following adjoint equations: J R = λd + D ) J R, = λc, 2.23) were te classical discrete adjoint can be written as λ D = J ) 1 R λ D D ) 1 R. 2.24) Te last step of our formulation involves computing te difference between te discrete adjoint, wic mimics te continuous adjoint beavior, and te classical discrete adjoint: ) 1 λ C = λ D D R + I), 2.25) were it must be possible to ceck tat D R ) 1 is noting more tan a smooter of te continuous adjoint solution. Te main issue appears wen we are looking for solutions wit strong socks. In tat

7 Adjoint based metods for computation of response surfaces 401 R ) 1. Te analogy of tis problem wit oter similar problems Glowinski situation we ypotesize tat tis operator produces a strong damping in ig frequencies of te continuous adjoint beyond a critical wave number. In oter words, te ig frequency numerics of te continuous adjoint solution are limited by te dispersion produced by D 1992; Zuazua 2005) suggests a filtering of te ig frequencies to eliminate te armful effect of te smooter described above. Tis may be done by a bi-grid algoritm. Te bi-grid strategy is a powerful numerical tool tat can be used to damp or filter te spurious numerics in control problems. To acive tis radical filtering it suffices to define a finite grid of step size 2. Following tis, te flow solution is approximated in tat grid and re-interpolated again to te fine mes. Wit tis procedure, te sort wavelengts are eliminated from te adjoint solution. Te major component of te noise was introduced at tese scales. Finally, an iterative process is proposed to compute te new discrete adjoint solution λ D n+1 = λc Filter { λ D n D ) )} 1 R, were λ D 0 = λc. 2.26) It is important to note tat tis filtering strategy will be especially useful if we are interested in computing sensitivities on small sensors or for second derivatives Continuous Euler adjoint formulation for two-dimensional ypersonic problems Te sensitivity computation in internal aerodynamics problems is formulated on a fluid domain Ω, enclosed by boundaries divided into a inlet S in, outlet S out, and solid wall boundaries S wall. Te objective is to compute te variation of J = jp, n S )ds, 2.27) S S wall wit respect to canges in te geometry and inlet flow. It is important to note tat jp, n S ) is a function tat depends on n S inward-pointing unit normal to S) and te pressure P. Hypersonic inviscid flows are caracterized by te appearance of strong sock discontinuities Σ. If te te discontinuity touces te surface S and x b = Σ S we ave tat, in order to account for discontinuities Bardos & Pironneau 2002) in te variation of J, δj = + S [ j P np + t tg S\x b j n S ) κ j + j n S n S )] δs ds j δp ds [j] x b [δσx b ) n S n Σ )δsx b )]. 2.28) P n S t Σ Ideal fluids are governed by te Euler s equations introduced in te previous section) tat [ can] develop strong sock waves. Wen tis occurs, te Rankine-Hugoniot conditions F nσ = 0 relate te flow variables on bot sides of te discontinuity. Σ In te presence of sock discontinuities te formal linearization of te state equations fails to be true Baeza et al. 2009) and te sensitivity of te model needs to take into account perturbations of te solution and perturbations of te location of te sock. In tis case, δu stands for te infinitesimal cange in te state on bot sides of te discontinuity and results from te solution of te linearized Euler equations, wereas δσ describes te infinitesimal normal deformation of te discontinuity and it results from a linearization of te Rankine-Hugoniot conditions:

8 402 F. Palacios, K. Duraisamy and J. J. Alonso ) AδU = 0, in Ω \ Σ, δ v n S = δs n v n S + tg δs) v t S, on S wall \ x b, δw) + = δw in, on S in, δw) [ + = 0, [ ] on S out, AδΣ n U + δu)] Σ + F Σ = 0 on Σ, Σ Σ 2.29) wit δw) + representing te incoming caracteristics and were F/ = A is te Jacobian matrix. In order to eliminate δp and δσ from 2.28), te adjoint problem is introduced troug Lagrange multipliers Ψ T A T Ψ = 0, in Ω \ Σ, ϕ n S = j, on S wall \ x b, Ψ T A n S ) = 0, on S in S out, [ ] [ ] Ψ T Σ = 0, tgψ T F t Σ = 0, on Σ, [ ] Ψ T x b ) F t Σ = [j] xb, at x x n b. b S t Σ 2.30) Te procedure is as follows: first te linearized Euler equations and Rankine-Hugoniot conditions are multiplied by te adjoint variables. Ten, te result is integrated over te domain taking care of te discontinuities), and finally te result is added to 2.28). Upon identifying te corresponding terms in tat final expression wit te adjoint equations, we can write te variation of te functional as: [ ) j j δj = S P np + t tg κ n S [ ) )] + n v n S )ϑ + tg v t S ϑ δs ds S\x b + j + j )] n S δs ds n S S in Ψ T A n S )δu in ds + [jp)] xb n S n Σ n S t Σ δs x b ), 2.31) were κ is te surface curvature, ϕ = ψ 2, ψ 3 ) and ϑ = ρψ 1 + ρ v S ϕ + ρhψ 4. Using 2.30) and 2.31) we are able to compute any variation of J due to a geometrical or inlet flow cange, even wen te computation of J includes a sock discontinuity impinging on te surface of integration S. Note tat te importance of some of tese terms depends on te actual flow condition and, in particular, on te relative angle of impingement of te discontinuity on te surface S. Wile not explicitly used in te results presented ere, tis derivation is meant to suggest tat care must be exercised to ensure tat terms in adjoint solutions tat include discontinuities are accounted for carefully. Witin te context of actual flow discretizations, discontinuities are represented by regions of ig gradients. Terefore, te importance of all terms must be carefully assessed. 3. Numerical experiments In tis section a selection of preliminary numerical results are sown. Two geometries involving single and multiple sock reflections in a constant area cannel) ave been selected as baseline configurations. Inviscid simulations at ig Mac number and

9 Adjoint based metods for computation of response surfaces 403 Figure 1. Density contours for nominal conditions Mac number 4.0 and ramp angle 6.0 ) Figure 2. Adjoint density contours for nominal conditions and pressure sensor located in te lower wall Figure 3. Grid adaptation using a gradient-based metod upper), adjoint-based computable error middle) and te new robust metod lower) functionals defined over small patces of te solid wall will be common aspects of te simulations tat we ave performed Robust grid adaptation To study te applicability of tis metod, a two-dimensional Euler supersonic test case as been selected. In particular, we are interested in creating a robust grid for te Clemens configuration Wagner et al. 2007). Te nominal conditions as in te experiments) are an inlet Mac number of 5.0 and a ramp angle 6.0 Fig.1). Te sensitivity computation Fig.2) as been performed using a second-order continuous adjoint solution. Te objective of tis test is to verify if te grid adapted using te new robust tecnique) for tose nominal conditions also provides a good value of te objective function for te complete response surface. Te baseline grid as a total of 1, 356 nodes and 2, 450 triangular elements. Te total lengt of te setup is mm, and te pressure sensor is located on te lower wall at mm from te inlet te size of te pressure sensor is 5.08 mm). In order to compare te performance of different grid adaptation tecniques Fig.3)

10 404 F. Palacios, K. Duraisamy and J. J. Alonso Figure 4. Comparison between different adaptation strategies grey color signifies an error < 4.0%) Figure 5. Direct comparison between computable error based adaptation and robust grid adaptation we ave used tree different error indicators: gradient based, classical adjoint based computable error), and te new robust adaptation metod. Once te adaptation is completed, te final step consists of computing te entire response surface Fig.4) and comparing te results wit te response surface tat as been computed using a very fine reference grid Fig.5). In conclusion, for te complete response surface computation, te robust grid strategy provides a lower error compared wit oter tecniques, despite te fact tat only one adaptation around Mac 5.0 and ramp angle 6.0 is used.

11 Adjoint based metods for computation of response surfaces 405 Figure 6. Density contours of te supersonic ramp Figure 7. Adjoint density contours of te supersonic ramp; note tat te sensor is located were te adjoint solution reaces it maximum value at te solid wall Figure 8. Comparison between continuous adjoint sensitivity and finite-difference tecnique on a baseline grid continuous adjoint ; finite differences ; finite differences wit bi-grid ) Figure 9. Comparison between continuous adjoint sensitivity and finite-differences tecnique on a very fine grid continuous adjoint ; finite differences bi-grid ) ; finite differences wit 3.2. Bi-grid filtering tecnique on a supersonic ramp To study te applicability of te bi-grid filtering tecnique, a two-dimensional Euler supersonic ramp of 25 Fig.6) is selected. In particular, we are interested in te derivative, wit respect to te inlet Mac number, of te pressure on a small sensor located approximately in te middle of te ramp. Wit te use of a fourt-order dissipation-based discretization, te continuous adjoint Fig.7) sensitivity was confirmed to be witin 0.1% of te analytical sensitivity. Similarly, te discrete adjoint sensitivity was confirmed to exactly matc te finite-difference value for a small enoug finite difference step size. For comparison purposes, we use finite differences to approximate te discrete sensitivity, and te bi-grid tecnique are applied to te solution of te direct problem. Te convergence of te finite-difference sensitivity is sown for two different grids: baseline grid Fig.8) wit 5, 680 nodes, Mac number 4.3, and a very fine grid Fig.9) wit 22, 472 nodes, Mac number 4.1. It is interesting to note tat te finite-difference approac sows a strange beavior near te analytical value of te gradient, beyond wic it does not converge to te analytical value of te gradient represented by te continuous adjoint solution). Once te bi-grid tecnique is applied to te flow solution, we obtain a value for te discrete sensitivity tat is very similar to te analytic one.

12 406 F. Palacios, K. Duraisamy and J. J. Alonso 4. Future plans and conclusions In tis work we proposed and sowed preliminary results for two new adjoint-based tecniques tat were aimed at enancing te computation of response surfaces and flow sensitivities. We ave demonstrated tat by adapting te computational grid using a residualbased strategy for te flow and te adjoint solutions, one can obtain a robust grid wit respect to small variations of te flow parameters. Furter work will be required to incorporate anisotropic adaptation, develop sensors to estimate te strengt of adaptation, and obtain rigorous upper bounds for te error estimation. We ave also proposed a metod to filter te flow solution wit te aim of obtaining noise-free sensitivities using discrete tecniques. Tis bi-grid strategy is expected to be especially useful for flows wit strong socks and for objective functions tat are defined over a small portion of te computational boundary. Furter work must be done in order to understand te beavior of te filtering operator. Current tests are underway in testing tis witin te framework of Automatic Differentiation AD). Acknowledgments Tis work was supported by te U.S. Department of Energy under te Predictive Science Academic Alliance Program PSAAP). Te first autor was visiting te Aero/Astro Department at Stanford University wile tis work was performed. He also tanks Prof. E. Zuazua for elpful discussions. REFERENCES Anderson, W. K. & Venkatakrisnan, V Design optimization on unstructured grids wit a continuous adjoint formulation. AIAA Paper Baeza, A., Castro, C., Palacios, F. & Zuazua, E D Euler sape design on nonregular flows using adjoint Rankine-Hugoniot relations. AIAA J. 47 3), Bardos, C. & Pironneau, O A formalism for te differentiation of conservation laws. C. R. Acad. Sci., Paris I 335), Boroucaki, H., Villard, J., Laug, P. & George, P. L Surface mes enancement wit geometric singularities identification. Comput. Metods Appl. Mec. Engrg. 194, Castro, C., Lozano, C., Palacios, F. & Zuazua, E A systematic continuous adjoint approac to viscous aerodynamic design on unstructured grids. AIAA J. 45 9), Cantrasmi, T. & Iaccarino, G Computing sock interactions under uncertainty. AIAA Paper Cung, H. S. & Alonso, J. J Using gradients to construct response surface models for ig-dimensional design optimization problems. AIAA Paper Eldred, M. S Recent advances in non-intrusive polynomial caos and stocastic collocation metods for uncertainty analysis and design. AIAA Paper Fidkowski, K. J. & Darmofal, D. L Output-based error estimation and mes adaptation in computational fluid dynamics: overview and recent results. AIAA Paper Gauger, N. R., Walter, A., Moldenauer, C. & Widalm, M Automatic

13 Adjoint based metods for computation of response surfaces 407 differentiation of an entire design cain for aerodynamic sape optimization. New Res. in Num. and Exp. Fluid Mec. 96, Giles, M. B., Duta, M. C., Müller, J.-D. & Pierce, N. A Algoritm developments for discrete adjoint metods. AIAA J. 41 2), Giles, M. B. & Pierce, N. A Adjoint error correction for integral outputs. Tec. Rep. 01/18. Oxford University Computing Laboratory, Numerical Analysis Group, Oxford, England. Glowinski, R Ensuring well-posedness by analogy; Stokes problem and boundary control for te wave equation. J. of Comp. Pys. 103, Griewank, A Evaluating derivatives, principles and tecniques of algoritmic diferentiation. Society for Industrial and Applied Matematics, Piladelpia. Jameson, A Aerodynamic design via control teory. J. Sci. Comput. 3, Kim, S., Alonso, J. J. & Jameson, A Design optimization of ig-lift configurations using a viscous continuous adjoint metod. AIAA Paper Mader, C. A., Martins, J. R. R. A., Alonso, J. J. & van der Weide, E ADjoint: an approac for rapid development of discrete adjoint solvers. AIAA J. 46 4), Martins, J. R. R. A., Mader, C. A. & Alonso, J. J ADjoint: an approac for rapid development of discrete adjoint solvers. AIAA Paper Mavriplis, D. J Multigrid solution of te discrete adjoint for optimization problems on unstructured meses. AIAA J. 44 1), Nadaraja, S. K. & Jameson, A A comparison of te continuous and discrete adjoint approac to automatic aerodynamic optimization. AIAA Paper Park, M. A Tree-dimensional turbulent rans adjointbased error correction. AIAA Paper Pierce, N. A. & Giles, M. B Adjoint and defect error bounding and correction for functional estimates. J. of Comp. Pys. 200, Venditti, D. A. & Darmofal, D. L Grid adaptation for functional outputs: application to two-dimensional inviscid flows. J. of Comp. Pys. 176, Wagner, J. L., Valdivia, A., Yuceil, K. B., Clemens, N. T. & Dolling, D. S An experimental investigation of supersonic inlet unstart. AIAA Paper Zuazua, E Propagation, observation, control and numerical approximation of waves approximated by finite difference metod. SIAM Review 47 2),

Error Estimation for High Speed Flows Using Continuous and Discrete Adjoints

Error Estimation for High Speed Flows Using Continuous and Discrete Adjoints Error Estimation for High Speed Flows Using and Adjoints Karthikeyan Duraisamy, Juan Alonso Stanford University, Stanford CA 94305, USA Francisco Palacios Madrid Polytechnic University, Madrid 8040, Spain

More information

Chapter 5 FINITE DIFFERENCE METHOD (FDM)

Chapter 5 FINITE DIFFERENCE METHOD (FDM) MEE7 Computer Modeling Tecniques in Engineering Capter 5 FINITE DIFFERENCE METHOD (FDM) 5. Introduction to FDM Te finite difference tecniques are based upon approximations wic permit replacing differential

More information

Numerical Differentiation

Numerical Differentiation Numerical Differentiation Finite Difference Formulas for te first derivative (Using Taylor Expansion tecnique) (section 8.3.) Suppose tat f() = g() is a function of te variable, and tat as 0 te function

More information

GRID CONVERGENCE ERROR ANALYSIS FOR MIXED-ORDER NUMERICAL SCHEMES

GRID CONVERGENCE ERROR ANALYSIS FOR MIXED-ORDER NUMERICAL SCHEMES GRID CONVERGENCE ERROR ANALYSIS FOR MIXED-ORDER NUMERICAL SCHEMES Cristoper J. Roy Sandia National Laboratories* P. O. Box 5800, MS 085 Albuquerque, NM 8785-085 AIAA Paper 00-606 Abstract New developments

More information

Jian-Guo Liu 1 and Chi-Wang Shu 2

Jian-Guo Liu 1 and Chi-Wang Shu 2 Journal of Computational Pysics 60, 577 596 (000) doi:0.006/jcp.000.6475, available online at ttp://www.idealibrary.com on Jian-Guo Liu and Ci-Wang Su Institute for Pysical Science and Tecnology and Department

More information

A First-Order System Approach for Diffusion Equation. I. Second-Order Residual-Distribution Schemes

A First-Order System Approach for Diffusion Equation. I. Second-Order Residual-Distribution Schemes A First-Order System Approac for Diffusion Equation. I. Second-Order Residual-Distribution Scemes Hiroaki Nisikawa W. M. Keck Foundation Laboratory for Computational Fluid Dynamics, Department of Aerospace

More information

Preconditioning in H(div) and Applications

Preconditioning in H(div) and Applications 1 Preconditioning in H(div) and Applications Douglas N. Arnold 1, Ricard S. Falk 2 and Ragnar Winter 3 4 Abstract. Summarizing te work of [AFW97], we sow ow to construct preconditioners using domain decomposition

More information

Optimal Shape Design of a Two-dimensional Asymmetric Diffsuer in Turbulent Flow

Optimal Shape Design of a Two-dimensional Asymmetric Diffsuer in Turbulent Flow THE 5 TH ASIAN COMPUTAITIONAL FLUID DYNAMICS BUSAN, KOREA, OCTOBER 7 ~ OCTOBER 30, 003 Optimal Sape Design of a Two-dimensional Asymmetric Diffsuer in Turbulent Flow Seokyun Lim and Haeceon Coi. Center

More information

1. Introduction. We consider the model problem: seeking an unknown function u satisfying

1. Introduction. We consider the model problem: seeking an unknown function u satisfying A DISCONTINUOUS LEAST-SQUARES FINITE ELEMENT METHOD FOR SECOND ORDER ELLIPTIC EQUATIONS XIU YE AND SHANGYOU ZHANG Abstract In tis paper, a discontinuous least-squares (DLS) finite element metod is introduced

More information

Arbitrary order exactly divergence-free central discontinuous Galerkin methods for ideal MHD equations

Arbitrary order exactly divergence-free central discontinuous Galerkin methods for ideal MHD equations Arbitrary order exactly divergence-free central discontinuous Galerkin metods for ideal MHD equations Fengyan Li, Liwei Xu Department of Matematical Sciences, Rensselaer Polytecnic Institute, Troy, NY

More information

Variational Localizations of the Dual Weighted Residual Estimator

Variational Localizations of the Dual Weighted Residual Estimator Publised in Journal for Computational and Applied Matematics, pp. 192-208, 2015 Variational Localizations of te Dual Weigted Residual Estimator Tomas Ricter Tomas Wick Te dual weigted residual metod (DWR)

More information

Parameter Fitted Scheme for Singularly Perturbed Delay Differential Equations

Parameter Fitted Scheme for Singularly Perturbed Delay Differential Equations International Journal of Applied Science and Engineering 2013. 11, 4: 361-373 Parameter Fitted Sceme for Singularly Perturbed Delay Differential Equations Awoke Andargiea* and Y. N. Reddyb a b Department

More information

New Streamfunction Approach for Magnetohydrodynamics

New Streamfunction Approach for Magnetohydrodynamics New Streamfunction Approac for Magnetoydrodynamics Kab Seo Kang Brooaven National Laboratory, Computational Science Center, Building 63, Room, Upton NY 973, USA. sang@bnl.gov Summary. We apply te finite

More information

5 Ordinary Differential Equations: Finite Difference Methods for Boundary Problems

5 Ordinary Differential Equations: Finite Difference Methods for Boundary Problems 5 Ordinary Differential Equations: Finite Difference Metods for Boundary Problems Read sections 10.1, 10.2, 10.4 Review questions 10.1 10.4, 10.8 10.9, 10.13 5.1 Introduction In te previous capters we

More information

How to Find the Derivative of a Function: Calculus 1

How to Find the Derivative of a Function: Calculus 1 Introduction How to Find te Derivative of a Function: Calculus 1 Calculus is not an easy matematics course Te fact tat you ave enrolled in suc a difficult subject indicates tat you are interested in te

More information

A = h w (1) Error Analysis Physics 141

A = h w (1) Error Analysis Physics 141 Introduction In all brances of pysical science and engineering one deals constantly wit numbers wic results more or less directly from experimental observations. Experimental observations always ave inaccuracies.

More information

A h u h = f h. 4.1 The CoarseGrid SystemandtheResidual Equation

A h u h = f h. 4.1 The CoarseGrid SystemandtheResidual Equation Capter Grid Transfer Remark. Contents of tis capter. Consider a grid wit grid size and te corresponding linear system of equations A u = f. Te summary given in Section 3. leads to te idea tat tere migt

More information

LIMITS AND DERIVATIVES CONDITIONS FOR THE EXISTENCE OF A LIMIT

LIMITS AND DERIVATIVES CONDITIONS FOR THE EXISTENCE OF A LIMIT LIMITS AND DERIVATIVES Te limit of a function is defined as te value of y tat te curve approaces, as x approaces a particular value. Te limit of f (x) as x approaces a is written as f (x) approaces, as

More information

Chapter 4: Numerical Methods for Common Mathematical Problems

Chapter 4: Numerical Methods for Common Mathematical Problems 1 Capter 4: Numerical Metods for Common Matematical Problems Interpolation Problem: Suppose we ave data defined at a discrete set of points (x i, y i ), i = 0, 1,..., N. Often it is useful to ave a smoot

More information

A Polynomial Adaptive LCP Scheme for Viscous Compressible Flows

A Polynomial Adaptive LCP Scheme for Viscous Compressible Flows A Polynomial Adaptive LCP Sceme for Viscous Compressible Flows J.S. Cagnone, B.C. Vermeire, and S.. Nadaraja Department of Mecanical Engineering, McGill University, Montreal, Canada, H3A 2S6 Email: jean-sebastien.cagnone@mail.mcgill.ca

More information

lecture 26: Richardson extrapolation

lecture 26: Richardson extrapolation 43 lecture 26: Ricardson extrapolation 35 Ricardson extrapolation, Romberg integration Trougout numerical analysis, one encounters procedures tat apply some simple approximation (eg, linear interpolation)

More information

1 Calculus. 1.1 Gradients and the Derivative. Q f(x+h) f(x)

1 Calculus. 1.1 Gradients and the Derivative. Q f(x+h) f(x) Calculus. Gradients and te Derivative Q f(x+) δy P T δx R f(x) 0 x x+ Let P (x, f(x)) and Q(x+, f(x+)) denote two points on te curve of te function y = f(x) and let R denote te point of intersection of

More information

Discontinuous Galerkin Methods for Relativistic Vlasov-Maxwell System

Discontinuous Galerkin Methods for Relativistic Vlasov-Maxwell System Discontinuous Galerkin Metods for Relativistic Vlasov-Maxwell System He Yang and Fengyan Li December 1, 16 Abstract e relativistic Vlasov-Maxwell (RVM) system is a kinetic model tat describes te dynamics

More information

Multigrid Methods for Discretized PDE Problems

Multigrid Methods for Discretized PDE Problems Towards Metods for Discretized PDE Problems Institute for Applied Matematics University of Heidelberg Feb 1-5, 2010 Towards Outline A model problem Solution of very large linear systems Iterative Metods

More information

Convergence and Descent Properties for a Class of Multilevel Optimization Algorithms

Convergence and Descent Properties for a Class of Multilevel Optimization Algorithms Convergence and Descent Properties for a Class of Multilevel Optimization Algoritms Stepen G. Nas April 28, 2010 Abstract I present a multilevel optimization approac (termed MG/Opt) for te solution of

More information

FEM solution of the ψ-ω equations with explicit viscous diffusion 1

FEM solution of the ψ-ω equations with explicit viscous diffusion 1 FEM solution of te ψ-ω equations wit explicit viscous diffusion J.-L. Guermond and L. Quartapelle 3 Abstract. Tis paper describes a variational formulation for solving te D time-dependent incompressible

More information

Order of Accuracy. ũ h u Ch p, (1)

Order of Accuracy. ũ h u Ch p, (1) Order of Accuracy 1 Terminology We consider a numerical approximation of an exact value u. Te approximation depends on a small parameter, wic can be for instance te grid size or time step in a numerical

More information

Flavius Guiaş. X(t + h) = X(t) + F (X(s)) ds.

Flavius Guiaş. X(t + h) = X(t) + F (X(s)) ds. Numerical solvers for large systems of ordinary differential equations based on te stocastic direct simulation metod improved by te and Runge Kutta principles Flavius Guiaş Abstract We present a numerical

More information

A method of Lagrange Galerkin of second order in time. Une méthode de Lagrange Galerkin d ordre deux en temps

A method of Lagrange Galerkin of second order in time. Une méthode de Lagrange Galerkin d ordre deux en temps A metod of Lagrange Galerkin of second order in time Une métode de Lagrange Galerkin d ordre deux en temps Jocelyn Étienne a a DAMTP, University of Cambridge, Wilberforce Road, Cambridge CB3 0WA, Great-Britain.

More information

Consider a function f we ll specify which assumptions we need to make about it in a minute. Let us reformulate the integral. 1 f(x) dx.

Consider a function f we ll specify which assumptions we need to make about it in a minute. Let us reformulate the integral. 1 f(x) dx. Capter 2 Integrals as sums and derivatives as differences We now switc to te simplest metods for integrating or differentiating a function from its function samples. A careful study of Taylor expansions

More information

Journal of Computational and Applied Mathematics

Journal of Computational and Applied Mathematics Journal of Computational and Applied Matematics 94 (6) 75 96 Contents lists available at ScienceDirect Journal of Computational and Applied Matematics journal omepage: www.elsevier.com/locate/cam Smootness-Increasing

More information

Copyright c 2008 Kevin Long

Copyright c 2008 Kevin Long Lecture 4 Numerical solution of initial value problems Te metods you ve learned so far ave obtained closed-form solutions to initial value problems. A closedform solution is an explicit algebriac formula

More information

LIMITATIONS OF EULER S METHOD FOR NUMERICAL INTEGRATION

LIMITATIONS OF EULER S METHOD FOR NUMERICAL INTEGRATION LIMITATIONS OF EULER S METHOD FOR NUMERICAL INTEGRATION LAURA EVANS.. Introduction Not all differential equations can be explicitly solved for y. Tis can be problematic if we need to know te value of y

More information

Numerical analysis of a free piston problem

Numerical analysis of a free piston problem MATHEMATICAL COMMUNICATIONS 573 Mat. Commun., Vol. 15, No. 2, pp. 573-585 (2010) Numerical analysis of a free piston problem Boris Mua 1 and Zvonimir Tutek 1, 1 Department of Matematics, University of

More information

Computers and Mathematics with Applications. A nonlinear weighted least-squares finite element method for Stokes equations

Computers and Mathematics with Applications. A nonlinear weighted least-squares finite element method for Stokes equations Computers Matematics wit Applications 59 () 5 4 Contents lists available at ScienceDirect Computers Matematics wit Applications journal omepage: www.elsevier.com/locate/camwa A nonlinear weigted least-squares

More information

arxiv: v1 [math.na] 7 Mar 2019

arxiv: v1 [math.na] 7 Mar 2019 Local Fourier analysis for mixed finite-element metods for te Stokes equations Yunui He a,, Scott P. MacLaclan a a Department of Matematics and Statistics, Memorial University of Newfoundland, St. Jon

More information

HOMEWORK HELP 2 FOR MATH 151

HOMEWORK HELP 2 FOR MATH 151 HOMEWORK HELP 2 FOR MATH 151 Here we go; te second round of omework elp. If tere are oters you would like to see, let me know! 2.4, 43 and 44 At wat points are te functions f(x) and g(x) = xf(x)continuous,

More information

Dedicated to the 70th birthday of Professor Lin Qun

Dedicated to the 70th birthday of Professor Lin Qun Journal of Computational Matematics, Vol.4, No.3, 6, 4 44. ACCELERATION METHODS OF NONLINEAR ITERATION FOR NONLINEAR PARABOLIC EQUATIONS Guang-wei Yuan Xu-deng Hang Laboratory of Computational Pysics,

More information

Investigating Euler s Method and Differential Equations to Approximate π. Lindsay Crowl August 2, 2001

Investigating Euler s Method and Differential Equations to Approximate π. Lindsay Crowl August 2, 2001 Investigating Euler s Metod and Differential Equations to Approximate π Lindsa Crowl August 2, 2001 Tis researc paper focuses on finding a more efficient and accurate wa to approximate π. Suppose tat x

More information

NON STANDARD FITTED FINITE DIFFERENCE METHOD FOR SINGULAR PERTURBATION PROBLEMS USING CUBIC SPLINE

NON STANDARD FITTED FINITE DIFFERENCE METHOD FOR SINGULAR PERTURBATION PROBLEMS USING CUBIC SPLINE Global and Stocastic Analysis Vol. 4 No. 1, January 2017, 1-10 NON STANDARD FITTED FINITE DIFFERENCE METHOD FOR SINGULAR PERTURBATION PROBLEMS USING CUBIC SPLINE K. PHANEENDRA AND E. SIVA PRASAD Abstract.

More information

arxiv: v1 [math.na] 3 Nov 2011

arxiv: v1 [math.na] 3 Nov 2011 arxiv:.983v [mat.na] 3 Nov 2 A Finite Difference Gost-cell Multigrid approac for Poisson Equation wit mixed Boundary Conditions in Arbitrary Domain Armando Coco, Giovanni Russo November 7, 2 Abstract In

More information

AN ANALYSIS OF THE EMBEDDED DISCONTINUOUS GALERKIN METHOD FOR SECOND ORDER ELLIPTIC PROBLEMS

AN ANALYSIS OF THE EMBEDDED DISCONTINUOUS GALERKIN METHOD FOR SECOND ORDER ELLIPTIC PROBLEMS AN ANALYSIS OF THE EMBEDDED DISCONTINUOUS GALERKIN METHOD FOR SECOND ORDER ELLIPTIC PROBLEMS BERNARDO COCKBURN, JOHNNY GUZMÁN, SEE-CHEW SOON, AND HENRYK K. STOLARSKI Abstract. Te embedded discontinuous

More information

The derivative function

The derivative function Roberto s Notes on Differential Calculus Capter : Definition of derivative Section Te derivative function Wat you need to know already: f is at a point on its grap and ow to compute it. Wat te derivative

More information

Large eddy simulation of turbulent flow downstream of a backward-facing step

Large eddy simulation of turbulent flow downstream of a backward-facing step Available online at www.sciencedirect.com Procedia Engineering 31 (01) 16 International Conference on Advances in Computational Modeling and Simulation Large eddy simulation of turbulent flow downstream

More information

Different Approaches to a Posteriori Error Analysis of the Discontinuous Galerkin Method

Different Approaches to a Posteriori Error Analysis of the Discontinuous Galerkin Method WDS'10 Proceedings of Contributed Papers, Part I, 151 156, 2010. ISBN 978-80-7378-139-2 MATFYZPRESS Different Approaces to a Posteriori Error Analysis of te Discontinuous Galerkin Metod I. Šebestová Carles

More information

Average Rate of Change

Average Rate of Change Te Derivative Tis can be tougt of as an attempt to draw a parallel (pysically and metaporically) between a line and a curve, applying te concept of slope to someting tat isn't actually straigt. Te slope

More information

NUMERICAL DIFFERENTIATION. James T. Smith San Francisco State University. In calculus classes, you compute derivatives algebraically: for example,

NUMERICAL DIFFERENTIATION. James T. Smith San Francisco State University. In calculus classes, you compute derivatives algebraically: for example, NUMERICAL DIFFERENTIATION James T Smit San Francisco State University In calculus classes, you compute derivatives algebraically: for example, f( x) = x + x f ( x) = x x Tis tecnique requires your knowing

More information

EXTENSION OF A POSTPROCESSING TECHNIQUE FOR THE DISCONTINUOUS GALERKIN METHOD FOR HYPERBOLIC EQUATIONS WITH APPLICATION TO AN AEROACOUSTIC PROBLEM

EXTENSION OF A POSTPROCESSING TECHNIQUE FOR THE DISCONTINUOUS GALERKIN METHOD FOR HYPERBOLIC EQUATIONS WITH APPLICATION TO AN AEROACOUSTIC PROBLEM SIAM J. SCI. COMPUT. Vol. 26, No. 3, pp. 821 843 c 2005 Society for Industrial and Applied Matematics ETENSION OF A POSTPROCESSING TECHNIQUE FOR THE DISCONTINUOUS GALERKIN METHOD FOR HYPERBOLIC EQUATIONS

More information

Finite Difference Methods Assignments

Finite Difference Methods Assignments Finite Difference Metods Assignments Anders Söberg and Aay Saxena, Micael Tuné, and Maria Westermarck Revised: Jarmo Rantakokko June 6, 1999 Teknisk databeandling Assignment 1: A one-dimensional eat equation

More information

Efficient, unconditionally stable, and optimally accurate FE algorithms for approximate deconvolution models of fluid flow

Efficient, unconditionally stable, and optimally accurate FE algorithms for approximate deconvolution models of fluid flow Efficient, unconditionally stable, and optimally accurate FE algoritms for approximate deconvolution models of fluid flow Leo G. Rebolz Abstract Tis paper addresses an open question of ow to devise numerical

More information

ERROR BOUNDS FOR THE METHODS OF GLIMM, GODUNOV AND LEVEQUE BRADLEY J. LUCIER*

ERROR BOUNDS FOR THE METHODS OF GLIMM, GODUNOV AND LEVEQUE BRADLEY J. LUCIER* EO BOUNDS FO THE METHODS OF GLIMM, GODUNOV AND LEVEQUE BADLEY J. LUCIE* Abstract. Te expected error in L ) attimet for Glimm s sceme wen applied to a scalar conservation law is bounded by + 2 ) ) /2 T

More information

A Modified Distributed Lagrange Multiplier/Fictitious Domain Method for Particulate Flows with Collisions

A Modified Distributed Lagrange Multiplier/Fictitious Domain Method for Particulate Flows with Collisions A Modified Distributed Lagrange Multiplier/Fictitious Domain Metod for Particulate Flows wit Collisions P. Sing Department of Mecanical Engineering New Jersey Institute of Tecnology University Heigts Newark,

More information

Derivation Of The Schwarzschild Radius Without General Relativity

Derivation Of The Schwarzschild Radius Without General Relativity Derivation Of Te Scwarzscild Radius Witout General Relativity In tis paper I present an alternative metod of deriving te Scwarzscild radius of a black ole. Te metod uses tree of te Planck units formulas:

More information

Exercise 19 - OLD EXAM, FDTD

Exercise 19 - OLD EXAM, FDTD Exercise 19 - OLD EXAM, FDTD A 1D wave propagation may be considered by te coupled differential equations u x + a v t v x + b u t a) 2 points: Derive te decoupled differential equation and give c in terms

More information

AN EFFICIENT AND ROBUST METHOD FOR SIMULATING TWO-PHASE GEL DYNAMICS

AN EFFICIENT AND ROBUST METHOD FOR SIMULATING TWO-PHASE GEL DYNAMICS AN EFFICIENT AND ROBUST METHOD FOR SIMULATING TWO-PHASE GEL DYNAMICS GRADY B. WRIGHT, ROBERT D. GUY, AND AARON L. FOGELSON Abstract. We develop a computational metod for simulating models of gel dynamics

More information

Entropy and the numerical integration of conservation laws

Entropy and the numerical integration of conservation laws Pysics Procedia Pysics Procedia 00 2011) 1 28 Entropy and te numerical integration of conservation laws Gabriella Puppo Dipartimento di Matematica, Politecnico di Torino Italy) Matteo Semplice Dipartimento

More information

Introduction to Derivatives

Introduction to Derivatives Introduction to Derivatives 5-Minute Review: Instantaneous Rates and Tangent Slope Recall te analogy tat we developed earlier First we saw tat te secant slope of te line troug te two points (a, f (a))

More information

Practice Problem Solutions: Exam 1

Practice Problem Solutions: Exam 1 Practice Problem Solutions: Exam 1 1. (a) Algebraic Solution: Te largest term in te numerator is 3x 2, wile te largest term in te denominator is 5x 2 3x 2 + 5. Tus lim x 5x 2 2x 3x 2 x 5x 2 = 3 5 Numerical

More information

arxiv: v3 [math.na] 15 Dec 2009

arxiv: v3 [math.na] 15 Dec 2009 THE NAVIER-STOKES-VOIGHT MODEL FOR IMAGE INPAINTING M.A. EBRAHIMI, MICHAEL HOLST, AND EVELYN LUNASIN arxiv:91.4548v3 [mat.na] 15 Dec 9 ABSTRACT. In tis paper we investigate te use of te D Navier-Stokes-Voigt

More information

A Hybridizable Discontinuous Galerkin Method for the Compressible Euler and Navier-Stokes Equations

A Hybridizable Discontinuous Galerkin Method for the Compressible Euler and Navier-Stokes Equations A Hybridizable Discontinuous Galerkin Metod for te Compressible Euler and Navier-Stokes Equations J. Peraire and N. C. Nguyen Massacusetts Institute of Tecnology, Cambridge, MA 02139, USA B. Cockburn University

More information

arxiv: v1 [math.oc] 18 May 2018

arxiv: v1 [math.oc] 18 May 2018 Derivative-Free Optimization Algoritms based on Non-Commutative Maps * Jan Feiling,, Amelie Zeller, and Cristian Ebenbauer arxiv:805.0748v [mat.oc] 8 May 08 Institute for Systems Teory and Automatic Control,

More information

1 The concept of limits (p.217 p.229, p.242 p.249, p.255 p.256) 1.1 Limits Consider the function determined by the formula 3. x since at this point

1 The concept of limits (p.217 p.229, p.242 p.249, p.255 p.256) 1.1 Limits Consider the function determined by the formula 3. x since at this point MA00 Capter 6 Calculus and Basic Linear Algebra I Limits, Continuity and Differentiability Te concept of its (p.7 p.9, p.4 p.49, p.55 p.56). Limits Consider te function determined by te formula f Note

More information

LECTURE 14 NUMERICAL INTEGRATION. Find

LECTURE 14 NUMERICAL INTEGRATION. Find LECTURE 14 NUMERCAL NTEGRATON Find b a fxdx or b a vx ux fx ydy dx Often integration is required. However te form of fx may be suc tat analytical integration would be very difficult or impossible. Use

More information

Polynomial Interpolation

Polynomial Interpolation Capter 4 Polynomial Interpolation In tis capter, we consider te important problem of approximatinga function fx, wose values at a set of distinct points x, x, x,, x n are known, by a polynomial P x suc

More information

Department of Mathematical Sciences University of South Carolina Aiken Aiken, SC 29801

Department of Mathematical Sciences University of South Carolina Aiken Aiken, SC 29801 RESEARCH SUMMARY AND PERSPECTIVES KOFFI B. FADIMBA Department of Matematical Sciences University of Sout Carolina Aiken Aiken, SC 29801 Email: KoffiF@usca.edu 1. Introduction My researc program as focused

More information

MIXED DISCONTINUOUS GALERKIN APPROXIMATION OF THE MAXWELL OPERATOR. SIAM J. Numer. Anal., Vol. 42 (2004), pp

MIXED DISCONTINUOUS GALERKIN APPROXIMATION OF THE MAXWELL OPERATOR. SIAM J. Numer. Anal., Vol. 42 (2004), pp MIXED DISCONTINUOUS GALERIN APPROXIMATION OF THE MAXWELL OPERATOR PAUL HOUSTON, ILARIA PERUGIA, AND DOMINI SCHÖTZAU SIAM J. Numer. Anal., Vol. 4 (004), pp. 434 459 Abstract. We introduce and analyze a

More information

LEAST-SQUARES FINITE ELEMENT APPROXIMATIONS TO SOLUTIONS OF INTERFACE PROBLEMS

LEAST-SQUARES FINITE ELEMENT APPROXIMATIONS TO SOLUTIONS OF INTERFACE PROBLEMS SIAM J. NUMER. ANAL. c 998 Society for Industrial Applied Matematics Vol. 35, No., pp. 393 405, February 998 00 LEAST-SQUARES FINITE ELEMENT APPROXIMATIONS TO SOLUTIONS OF INTERFACE PROBLEMS YANZHAO CAO

More information

Effects of Radiation on Unsteady Couette Flow between Two Vertical Parallel Plates with Ramped Wall Temperature

Effects of Radiation on Unsteady Couette Flow between Two Vertical Parallel Plates with Ramped Wall Temperature Volume 39 No. February 01 Effects of Radiation on Unsteady Couette Flow between Two Vertical Parallel Plates wit Ramped Wall Temperature S. Das Department of Matematics University of Gour Banga Malda 73

More information

Click here to see an animation of the derivative

Click here to see an animation of the derivative Differentiation Massoud Malek Derivative Te concept of derivative is at te core of Calculus; It is a very powerful tool for understanding te beavior of matematical functions. It allows us to optimize functions,

More information

Theoretical Analysis of Flow Characteristics and Bearing Load for Mass-produced External Gear Pump

Theoretical Analysis of Flow Characteristics and Bearing Load for Mass-produced External Gear Pump TECHNICAL PAPE Teoretical Analysis of Flow Caracteristics and Bearing Load for Mass-produced External Gear Pump N. YOSHIDA Tis paper presents teoretical equations for calculating pump flow rate and bearing

More information

Derivatives. By: OpenStaxCollege

Derivatives. By: OpenStaxCollege By: OpenStaxCollege Te average teen in te United States opens a refrigerator door an estimated 25 times per day. Supposedly, tis average is up from 10 years ago wen te average teenager opened a refrigerator

More information

Quantum Numbers and Rules

Quantum Numbers and Rules OpenStax-CNX module: m42614 1 Quantum Numbers and Rules OpenStax College Tis work is produced by OpenStax-CNX and licensed under te Creative Commons Attribution License 3.0 Abstract Dene quantum number.

More information

Learning based super-resolution land cover mapping

Learning based super-resolution land cover mapping earning based super-resolution land cover mapping Feng ing, Yiang Zang, Giles M. Foody IEEE Fellow, Xiaodong Xiuua Zang, Siming Fang, Wenbo Yun Du is work was supported in part by te National Basic Researc

More information

Problem Solving. Problem Solving Process

Problem Solving. Problem Solving Process Problem Solving One of te primary tasks for engineers is often solving problems. It is wat tey are, or sould be, good at. Solving engineering problems requires more tan just learning new terms, ideas and

More information

CFD calculation of convective heat transfer coefficients and validation Part I: Laminar flow Neale, A.; Derome, D.; Blocken, B.; Carmeliet, J.E.

CFD calculation of convective heat transfer coefficients and validation Part I: Laminar flow Neale, A.; Derome, D.; Blocken, B.; Carmeliet, J.E. CFD calculation of convective eat transfer coefficients and validation Part I: Laminar flow Neale, A.; Derome, D.; Blocken, B.; Carmeliet, J.E. Publised in: IEA Annex 41 working meeting, Kyoto, Japan Publised:

More information

Material for Difference Quotient

Material for Difference Quotient Material for Difference Quotient Prepared by Stepanie Quintal, graduate student and Marvin Stick, professor Dept. of Matematical Sciences, UMass Lowell Summer 05 Preface Te following difference quotient

More information

SECTION 3.2: DERIVATIVE FUNCTIONS and DIFFERENTIABILITY

SECTION 3.2: DERIVATIVE FUNCTIONS and DIFFERENTIABILITY (Section 3.2: Derivative Functions and Differentiability) 3.2.1 SECTION 3.2: DERIVATIVE FUNCTIONS and DIFFERENTIABILITY LEARNING OBJECTIVES Know, understand, and apply te Limit Definition of te Derivative

More information

Lyapunov spectrum of scale-resolving turbulent simulations. Application to chaotic adjoints

Lyapunov spectrum of scale-resolving turbulent simulations. Application to chaotic adjoints Lyapunov spectrum of scale-resolving turbulent simulations. Application to caotic adjoints Te MIT Faculty as made tis article openly available. Please sare ow tis access benefits you. Your story matters.

More information

Explicit Hyperbolic Reconstructed Discontinuous Galerkin Methods for Time-Dependent Problems

Explicit Hyperbolic Reconstructed Discontinuous Galerkin Methods for Time-Dependent Problems AIAA AVIATION Forum June 25-29 218 Atlanta Georgia 218 Fluid Dynamics Conference 1.2514/6.218-427 Explicit Hyperbolic Reconstructed Discontinuous Galerkin Metods for Time-Dependent Problems Jialin Lou

More information

Downloaded 01/08/14 to Redistribution subject to SIAM license or copyright; see

Downloaded 01/08/14 to Redistribution subject to SIAM license or copyright; see SIAM J. SCI. COMPUT. Vol. 3, No. 3, pp. 506 56 c 00 Society for Industrial and Applied Matematics NESTED ITERATION AND FIRST-ORDER SYSTEM LEAST SQUARES FOR INCOMPRESSIBLE, RESISTIVE MAGNETOHYDRODYNAMICS

More information

Continuity and Differentiability Worksheet

Continuity and Differentiability Worksheet Continuity and Differentiability Workseet (Be sure tat you can also do te grapical eercises from te tet- Tese were not included below! Typical problems are like problems -3, p. 6; -3, p. 7; 33-34, p. 7;

More information

The Laplace equation, cylindrically or spherically symmetric case

The Laplace equation, cylindrically or spherically symmetric case Numerisce Metoden II, 7 4, und Übungen, 7 5 Course Notes, Summer Term 7 Some material and exercises Te Laplace equation, cylindrically or sperically symmetric case Electric and gravitational potential,

More information

arxiv: v1 [math.na] 27 Jan 2014

arxiv: v1 [math.na] 27 Jan 2014 L 2 -ERROR ESTIMATES FOR FINITE ELEMENT APPROXIMATIONS OF BOUNDARY FLUXES MATS G. LARSON AND ANDRÉ MASSING arxiv:1401.6994v1 [mat.na] 27 Jan 2014 Abstract. We prove quasi-optimal a priori error estimates

More information

Advancements In Finite Element Methods For Newtonian And Non-Newtonian Flows

Advancements In Finite Element Methods For Newtonian And Non-Newtonian Flows Clemson University TigerPrints All Dissertations Dissertations 8-3 Advancements In Finite Element Metods For Newtonian And Non-Newtonian Flows Keit Galvin Clemson University, kjgalvi@clemson.edu Follow

More information

An Adaptive Model Switching and Discretization Algorithm for Gas Flow on Networks

An Adaptive Model Switching and Discretization Algorithm for Gas Flow on Networks Procedia Computer Science 1 (21) (212) 1 1 1331 134 Procedia Computer Science www.elsevier.com/locate/procedia International Conference on Computational Science, ICCS 21 An Adaptive Model Switcing and

More information

Quaternion Dynamics, Part 1 Functions, Derivatives, and Integrals. Gary D. Simpson. rev 01 Aug 08, 2016.

Quaternion Dynamics, Part 1 Functions, Derivatives, and Integrals. Gary D. Simpson. rev 01 Aug 08, 2016. Quaternion Dynamics, Part 1 Functions, Derivatives, and Integrals Gary D. Simpson gsim1887@aol.com rev 1 Aug 8, 216 Summary Definitions are presented for "quaternion functions" of a quaternion. Polynomial

More information

Simulation and verification of a plate heat exchanger with a built-in tap water accumulator

Simulation and verification of a plate heat exchanger with a built-in tap water accumulator Simulation and verification of a plate eat excanger wit a built-in tap water accumulator Anders Eriksson Abstract In order to test and verify a compact brazed eat excanger (CBE wit a built-in accumulation

More information

158 Calculus and Structures

158 Calculus and Structures 58 Calculus and Structures CHAPTER PROPERTIES OF DERIVATIVES AND DIFFERENTIATION BY THE EASY WAY. Calculus and Structures 59 Copyrigt Capter PROPERTIES OF DERIVATIVES. INTRODUCTION In te last capter you

More information

A Hybrid Mixed Discontinuous Galerkin Finite Element Method for Convection-Diffusion Problems

A Hybrid Mixed Discontinuous Galerkin Finite Element Method for Convection-Diffusion Problems A Hybrid Mixed Discontinuous Galerkin Finite Element Metod for Convection-Diffusion Problems Herbert Egger Joacim Scöberl We propose and analyse a new finite element metod for convection diffusion problems

More information

A Numerical Scheme for Particle-Laden Thin Film Flow in Two Dimensions

A Numerical Scheme for Particle-Laden Thin Film Flow in Two Dimensions A Numerical Sceme for Particle-Laden Tin Film Flow in Two Dimensions Mattew R. Mata a,, Andrea L. Bertozzi a a Department of Matematics, University of California Los Angeles, 520 Portola Plaza, Los Angeles,

More information

Mathematics 5 Worksheet 11 Geometry, Tangency, and the Derivative

Mathematics 5 Worksheet 11 Geometry, Tangency, and the Derivative Matematics 5 Workseet 11 Geometry, Tangency, and te Derivative Problem 1. Find te equation of a line wit slope m tat intersects te point (3, 9). Solution. Te equation for a line passing troug a point (x

More information

HT TURBULENT NATURAL CONVECTION IN A DIFFERENTIALLY HEATED VERTICAL CHANNEL. Proceedings of 2008 ASME Summer Heat Transfer Conference HT2008

HT TURBULENT NATURAL CONVECTION IN A DIFFERENTIALLY HEATED VERTICAL CHANNEL. Proceedings of 2008 ASME Summer Heat Transfer Conference HT2008 Proceedings of 2008 ASME Summer Heat Transfer Conference HT2008 August 10-14, 2008, Jacksonville, Florida USA Proceedings of HT2008 2008 ASME Summer Heat Transfer Conference August 10-14, 2008, Jacksonville,

More information

One-Sided Position-Dependent Smoothness-Increasing Accuracy-Conserving (SIAC) Filtering Over Uniform and Non-uniform Meshes

One-Sided Position-Dependent Smoothness-Increasing Accuracy-Conserving (SIAC) Filtering Over Uniform and Non-uniform Meshes DOI 10.1007/s10915-014-9946-6 One-Sided Position-Dependent Smootness-Increasing Accuracy-Conserving (SIAC) Filtering Over Uniform and Non-uniform Meses JenniferK.Ryan Xiaozou Li Robert M. Kirby Kees Vuik

More information

New Distribution Theory for the Estimation of Structural Break Point in Mean

New Distribution Theory for the Estimation of Structural Break Point in Mean New Distribution Teory for te Estimation of Structural Break Point in Mean Liang Jiang Singapore Management University Xiaou Wang Te Cinese University of Hong Kong Jun Yu Singapore Management University

More information

Implicit-explicit variational integration of highly oscillatory problems

Implicit-explicit variational integration of highly oscillatory problems Implicit-explicit variational integration of igly oscillatory problems Ari Stern Structured Integrators Worksop April 9, 9 Stern, A., and E. Grinspun. Multiscale Model. Simul., to appear. arxiv:88.39 [mat.na].

More information

MATH745 Fall MATH745 Fall

MATH745 Fall MATH745 Fall MATH745 Fall 5 MATH745 Fall 5 INTRODUCTION WELCOME TO MATH 745 TOPICS IN NUMERICAL ANALYSIS Instructor: Dr Bartosz Protas Department of Matematics & Statistics Email: bprotas@mcmasterca Office HH 36, Ext

More information

Simulations of the turbulent channel flow at Re τ = 180 with projection-based finite element variational multiscale methods

Simulations of the turbulent channel flow at Re τ = 180 with projection-based finite element variational multiscale methods INTERNATIONAL JOURNAL FOR NUMERICAL METHODS IN FLUIDS Int. J. Numer. Met. Fluids 7; 55:47 49 Publised online 4 Marc 7 in Wiley InterScience (www.interscience.wiley.com). DOI:./fld.46 Simulations of te

More information

arxiv: v2 [cs.na] 22 Dec 2016

arxiv: v2 [cs.na] 22 Dec 2016 MONOTONICITY-PRESERVING FINITE ELEMENT SCHEMES BASED ON DIFFERENTIABLE NONLINEAR STABILIZATION SANTIAGO BADIA AND JESÚS BONILLA arxiv:66.8743v2 [cs.na] 22 Dec 26 Abstract. In tis work, we propose a nonlinear

More information

Time (hours) Morphine sulfate (mg)

Time (hours) Morphine sulfate (mg) Mat Xa Fall 2002 Review Notes Limits and Definition of Derivative Important Information: 1 According to te most recent information from te Registrar, te Xa final exam will be eld from 9:15 am to 12:15

More information

arxiv: v2 [math.na] 5 Jul 2017

arxiv: v2 [math.na] 5 Jul 2017 Trace Finite Element Metods for PDEs on Surfaces Maxim A. Olsanskii and Arnold Reusken arxiv:1612.00054v2 [mat.na] 5 Jul 2017 Abstract In tis paper we consider a class of unfitted finite element metods

More information