You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Our current solver for linear equations computes a matrix inverse using Gaussian elimination. Especially for validated matrices, this leads to large errors due to the dependency effect. We should instead implement direct solution for approximate matrices, and a preconditioned solver for validated matrices.
We should also implement a solver using a-posteriori error analysis as proposed by Rump.
Finally, we should provide a LinearSolver interface to abstract details of linear algebra solver methods.
The text was updated successfully, but these errors were encountered:
Our current solver for linear equations computes a matrix inverse using Gaussian elimination. Especially for validated matrices, this leads to large errors due to the dependency effect. We should instead implement direct solution for approximate matrices, and a preconditioned solver for validated matrices.
We should also implement a solver using a-posteriori error analysis as proposed by Rump.
Finally, we should provide a
LinearSolver
interface to abstract details of linear algebra solver methods.The text was updated successfully, but these errors were encountered: