Towards efficient backward-in-time adjoint computations using data compression techniques
Abstract
In the context of a posteriori error estimation for nonlinear time-dependent partial differential equations, the state-of-the-practice is to use adjoint approaches which require the solution of a backward-in-time problem defined by a linearization of the forward problem. One of the major obstacles in the practical application of these approaches, we found, is the need to store, or recompute, the forward solution to define the adjoint problem and to evaluate the error representation. Our study considers the use of data compression techniques to approximate forward solutions employed in the backward-in-time integration. The development derives an error representation that accounts for the difference between the standard-approach and the compressed approximation of the forward solution. This representation is algorithmically similar to the standard representation and only requires the computation of the quantity of interest for the forward solution and the data-compressed reconstructed solution (i.e. scalar quantities that can be evaluated as the forward problem is integrated). This approach is then compared with existing techniques, such as checkpointing and time-averaged adjoints. Lastly, we provide numerical results indicating the potential efficiency of our approach on a transient diffusion–reaction equation and on the Navier–Stokes equations. These results demonstrate memory compression ratios up to 450×450× while maintainingmore »
- Authors:
-
- Sandia National Lab. (SNL-NM), Albuquerque, NM (United States)
- Publication Date:
- Research Org.:
- Sandia National Lab. (SNL-NM), Albuquerque, NM (United States)
- Sponsoring Org.:
- USDOE National Nuclear Security Administration (NNSA)
- OSTI Identifier:
- 1141564
- Report Number(s):
- SAND-2014-2541J
Journal ID: ISSN 0045-7825; PII: S0045782514004800
- Grant/Contract Number:
- AC04-94AL85000
- Resource Type:
- Accepted Manuscript
- Journal Name:
- Computer Methods in Applied Mechanics and Engineering
- Additional Journal Information:
- Journal Volume: 288; Journal Issue: C; Journal ID: ISSN 0045-7825
- Publisher:
- Elsevier
- Country of Publication:
- United States
- Language:
- English
- Subject:
- 97 MATHEMATICS AND COMPUTING; 42 ENGINEERING; data compression; adjoint problem; error analysis; navier-stokes
Citation Formats
Cyr, E. C., Shadid, J. N., and Wildey, T. Towards efficient backward-in-time adjoint computations using data compression techniques. United States: N. p., 2014.
Web. doi:10.1016/j.cma.2014.12.001.
Cyr, E. C., Shadid, J. N., & Wildey, T. Towards efficient backward-in-time adjoint computations using data compression techniques. United States. https://doi.org/10.1016/j.cma.2014.12.001
Cyr, E. C., Shadid, J. N., and Wildey, T. Tue .
"Towards efficient backward-in-time adjoint computations using data compression techniques". United States. https://doi.org/10.1016/j.cma.2014.12.001. https://www.osti.gov/servlets/purl/1141564.
@article{osti_1141564,
title = {Towards efficient backward-in-time adjoint computations using data compression techniques},
author = {Cyr, E. C. and Shadid, J. N. and Wildey, T.},
abstractNote = {In the context of a posteriori error estimation for nonlinear time-dependent partial differential equations, the state-of-the-practice is to use adjoint approaches which require the solution of a backward-in-time problem defined by a linearization of the forward problem. One of the major obstacles in the practical application of these approaches, we found, is the need to store, or recompute, the forward solution to define the adjoint problem and to evaluate the error representation. Our study considers the use of data compression techniques to approximate forward solutions employed in the backward-in-time integration. The development derives an error representation that accounts for the difference between the standard-approach and the compressed approximation of the forward solution. This representation is algorithmically similar to the standard representation and only requires the computation of the quantity of interest for the forward solution and the data-compressed reconstructed solution (i.e. scalar quantities that can be evaluated as the forward problem is integrated). This approach is then compared with existing techniques, such as checkpointing and time-averaged adjoints. Lastly, we provide numerical results indicating the potential efficiency of our approach on a transient diffusion–reaction equation and on the Navier–Stokes equations. These results demonstrate memory compression ratios up to 450×450× while maintaining reasonable accuracy in the error-estimates.},
doi = {10.1016/j.cma.2014.12.001},
journal = {Computer Methods in Applied Mechanics and Engineering},
number = C,
volume = 288,
place = {United States},
year = {Tue Dec 16 00:00:00 EST 2014},
month = {Tue Dec 16 00:00:00 EST 2014}
}
Web of Science
Works referenced in this record:
A Posteriori Error Bounds and Global Error Control for Approximation of Ordinary Differential Equations
journal, February 1995
- Estep, Donald
- SIAM Journal on Numerical Analysis, Vol. 32, Issue 1
An A Posteriori–A Priori Analysis of Multiscale Operator Splitting
journal, January 2008
- Estep, D.; Ginting, V.; Ropp, D.
- SIAM Journal on Numerical Analysis, Vol. 46, Issue 3
A posteriori error analysis of IMEX multi-step time integration methods for advection–diffusion–reaction equations
journal, March 2015
- Chaudhry, Jehanzeb H.; Estep, Donald; Ginting, Victor
- Computer Methods in Applied Mechanics and Engineering, Vol. 285
A Posteriori Analysis and Adaptive Error Control for Multiscale Operator Decomposition Solution of Elliptic Systems I: Triangular Systems
journal, January 2009
- Carey, V.; Estep, D.; Tavener, S.
- SIAM Journal on Numerical Analysis, Vol. 47, Issue 1
A Posteriori Analysis and Improved Accuracy for an Operator Decomposition Solution of a Conjugate Heat Transfer Problem
journal, January 2008
- Estep, D.; Tavener, S.; Wildey, T.
- SIAM Journal on Numerical Analysis, Vol. 46, Issue 4
A posteriori error estimation and adaptive mesh refinement for a multiscale operator decomposition approach to fluid–solid heat transfer
journal, June 2010
- Estep, Donald; Tavener, Simon; Wildey, Tim
- Journal of Computational Physics, Vol. 229, Issue 11
Achieving logarithmic growth of temporal and spatial complexity in reverse automatic differentiation
journal, January 1992
- Griewank, Andreas
- Optimization Methods and Software, Vol. 1, Issue 1
Checkpointing Schemes for Adjoint Codes: Application to the Meteorological Model Meso-NH
journal, January 2001
- Charpentier, I.
- SIAM Journal on Scientific Computing, Vol. 22, Issue 6
Algorithm 799: revolve: an implementation of checkpointing for the reverse or adjoint mode of computational differentiation
journal, March 2000
- Griewank, Andreas; Walther, Andrea
- ACM Transactions on Mathematical Software, Vol. 26, Issue 1
Minimal Repetition Dynamic Checkpointing Algorithm for Unsteady Adjoint Calculation
journal, January 2009
- Wang, Qiqi; Moin, Parviz; Iaccarino, Gianluca
- SIAM Journal on Scientific Computing, Vol. 31, Issue 4
Blockwise Adaptivity for Time Dependent Problems Based on Coarse Scale Adjoint Solutions
journal, January 2010
- Carey, V.; Estep, D.; Johansson, A.
- SIAM Journal on Scientific Computing, Vol. 32, Issue 4
Approaches for Adjoint-Based A Posteriori Analysis of Stabilized Finite Element Methods
journal, January 2014
- Cyr, Eric C.; Shadid, John; Wildey, Tim
- SIAM Journal on Scientific Computing, Vol. 36, Issue 2
An optimal control approach to a posteriori error estimation in finite element methods
journal, May 2001
- Becker, Roland; Rannacher, Rolf
- Acta Numerica, Vol. 10
Preserving Symmetries in the Proper Orthogonal Decomposition
journal, March 1993
- Aubry, Nadine; Lian, Wen-Yu; Titi, Edriss S.
- SIAM Journal on Scientific Computing, Vol. 14, Issue 2
The Proper Orthogonal Decomposition in the Analysis of Turbulent Flows
journal, January 1993
- Berkooz, G.; Holmes, P.; Lumley, J. L.
- Annual Review of Fluid Mechanics, Vol. 25, Issue 1
POD and CVT-based reduced-order modeling of Navier–Stokes flows
journal, December 2006
- Burkardt, John; Gunzburger, Max; Lee, Hyung-Chun
- Computer Methods in Applied Mechanics and Engineering, Vol. 196, Issue 1-3
Model Reduction for Large-Scale Systems with High-Dimensional Parametric Input Space
journal, January 2008
- Bui-Thanh, T.; Willcox, K.; Ghattas, O.
- SIAM Journal on Scientific Computing, Vol. 30, Issue 6
Reduced-order modeling of parameterized PDEs using time-space-parameter principal component analysis
journal, November 2009
- Audouze, C.; De Vuyst, F.; Nair, P. B.
- International Journal for Numerical Methods in Engineering, Vol. 80, Issue 8
Duality Based A Posteriori Error Estimation for Quasi-Periodic Solutions Using Time Averages
journal, January 2011
- Braack, M.; Burman, E.; Taschenberger, N.
- SIAM Journal on Scientific Computing, Vol. 33, Issue 5
Works referencing / citing this record:
Compression Challenges in Large Scale Partial Differential Equation Solvers
journal, September 2019
- Götschel, Sebastian; Weiser, Martin
- Algorithms, Vol. 12, Issue 9