A saddle point least squares approach for primal mixed formulations of second order PDEs.
Constantin BacutaKlajdi QirkoPublished in: Comput. Math. Appl. (2017)
Keyphrases
- least squares
- saddle point
- primal dual
- valid inequalities
- partial differential equations
- numerical methods
- linear programming
- variational inequalities
- interior point
- penalty function
- higher order
- linear program
- convex optimization
- linear programming problems
- parameter estimation
- convergence rate
- interior point methods
- structured prediction
- optical flow
- discrete space
- level set
- image processing
- approximation algorithms
- semidefinite programming
- image denoising
- nonnegative matrix factorization
- generative model
- global constraints
- np hard
- maximum margin
- objective function
- high order
- multiscale
- bayesian networks