Matrix Computations or Numerical Linear Algebra

[c1adde]

I'll follow a Scientific Computing course next semester and they gave us a choice between the books Matrix Computations (Golub, Gene H) and Numerical Linear Algebra (Trefethen, Lloyd H). Does anyone know the two books and can recommend one of the two?

[Rick_SAS]

I learned from "Matrix Computation" (taught by Van Loan himself), but for a one-semester course I'd recommend Trefethan's book. It is shorter, more modern, and stresses the main ideas. Golub/Van Loan is more comprehensive, but I haven't had the need to use more than half the material in that book. For a good intro that covers the basic, use Trefethan. 

[MargaretWhite]

Both are excellent books, but your choice depends on your focus: if you’re looking for more practical algorithms and detailed implementation for large-scale computations, go with “Matrix Computations” by Golub & Van Loan. If you prefer a modern, concept-driven approach to numerical methods in linear algebra, then “Numerical Linear Algebra” by Trefethen & Bau is your best bet.