23.01.2015 Views

Download - Wolfram Research

Download - Wolfram Research

Download - Wolfram Research

SHOW MORE
SHOW LESS
  • No tags were found...

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

328 3. Reference Manual<br />

If you set GEPSolver −> LREigenpair, note that the ProjectionVectors setting given for<br />

ApproximateDeterminant is not passed to LREigenpair. To choose the projection vectors<br />

for the GEP solver, change the value of GEPSolverOptions.<br />

SingularityTest<br />

After each approximation step, ApproximateDeterminant applies a singularity test to the matrix<br />

A s B to ensure that removing a matrix entry has not rendered the GEP singular. The singularity<br />

test is performed by numerical computation of the rank of A s B for some s ∈ C which is not<br />

an eigenvalue of the GEP. However, there is no guarantee that numerical rank computation always<br />

yields a mathematically correct result. This may cause singularity to remain undetected in some<br />

situations, particularly when the GEP is ill-conditioned. With the option SingularityTest, you<br />

can select the function ApproximateDeterminant uses to determine whether the GEP is singular. If<br />

you encounter singularity problems, i.e. if detA s B ⩵ after approximating a system of circuit<br />

equations, then you should change the value of SingularityTest and rerun the approximation.<br />

The possible values for SingularityTest are:<br />

SingularityTestByLU<br />

SingularityTestByQR<br />

Function[{M}, expr]<br />

perform singularity tests by rank computation using<br />

LUDecomposition<br />

perform singularity tests by rank computation using<br />

QRDecomposition<br />

specify a user-defined function which returns True if the<br />

complex-valued floating-point matrix M is singular<br />

Values for the SingularityTest option.<br />

TestFrequency<br />

With the option TestFrequency, you can specify the value of the complex frequency variable s<br />

which is used for testing whether the numerical matrix A s B is singular. For best numerical<br />

accuracy and computing performance, it is recommended that you choose a real value of the<br />

same order of magnitude as the modulus of the target eigenvalue. For example, if you wish to<br />

approximate a GEP with respect to an eigenvalue s ⩵ ⨯ j ⨯ , then the option setting<br />

TestFrequency −> 1.0*^8 constitutes an appropriate choice.<br />

The point in the complex plane represented by the value of TestFrequency should not lie<br />

in the neighborhood of any eigenvalue of the GEP to be approximated. An inappropriate<br />

choice of the test frequency may result in ill-conditioned rank computation problems.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!