The Schur decomposition is a representation of matrix *A* as *A = S ^{ T} T S*, where

This algorithm performs the Schur decomposition of the matrix in upper Hessenberg form by using a QR algorithm with multiple shifts. This algorithm is a block-matrix analog of the ordinary QR algorithm with double shift. As all other block-matrix algorithms, this algorithm requires adjustment to achieve optimal performance.

You can adjust a value of **NS** (internal parameter of the subroutine **InternalSchurDecomposition**) by defining a number of shifts in one iteration. By increasing the number of shifts, the algorithm performance rises, reaching its maximum between NS=4 and NS=16. If NS is more than 16, algorithm slows down considerably. Interval ranges can differ according to different systems, but the situation is mostly same. The default value of NS provides a good performance on most systems, but if the performance is critical, it is worth calibrating this parameter. It should be noted that the optimal parameter value depends both on the system characteristics and on the properties of the matrices processed.

The subroutine **UpperHessenbergSchurDecomposition** performs the Schur decomposition. It returns matrices *T* and *S*. Commonly, we need to know the block structure of matrix T. It could be easily recognized: since all the elements below the blocks are zeros, elements a[i+1,i] which are equal to 0 show the block border.

*Algorithm implemented on the basis of the subroutine DHSEQR (LAPACK 3.0 library).*

*This article is licensed for personal use only.*

ALGLIB Project offers you two editions of ALGLIB:

**ALGLIB Free Edition**:

delivered for free

offers full set of numerical functionality

extensive algorithmic optimizations

no low level optimizations

non-commercial license

**ALGLIB Commercial Edition**:

flexible pricing

offers full set of numerical functionality

extensive algorithmic optimizations

high performance (SMP, SIMD)

commercial license with support plan

Links to download sections for Free and Commercial editions can be found below:

C++ library.

Delivered with sources.

Monolithic design.

Extreme portability.

Delivered with sources.

Monolithic design.

Extreme portability.

C# library with native kernels.

Delivered with sources.

VB.NET and IronPython wrappers.

Extreme portability.

Delivered with sources.

VB.NET and IronPython wrappers.

Extreme portability.

Delphi wrapper around C core.

Delivered as precompiled binary.

Compatible with FreePascal.

Delivered as precompiled binary.

Compatible with FreePascal.

CPython wrapper around C core.

Delivered as precompiled binary.

Delivered as precompiled binary.