Skip to content

Official Release for LAPACK 3.11.0
Compare
Choose a tag to compare
@weslleyspereira weslleyspereira released this 11 Nov 18:24
· 476 commits to master since this release
v3.11.0
7866626
This commit was created on GitHub.com and signed with GitHub’s verified signature. The key has expired.
GPG key ID: 4AEE18F83AFDEB23
Expired
Learn about vigilant mode.

This material is based upon work supported by the National Science Foundation and the Department of Energy.
LAPACK is a software package provided by Univ. of Tennessee, Univ. of California, Berkeley, Univ. of Colorado Denver and NAG Ltd..

The following notes are also at http://netlib.org/lapack/lapack-3.11.0.html

Normwise criterion for infinite eigenvalues in the QZ algorithm by @thijssteel

The normwise criterion is more robust at detecting infinite eigenvalues than the elementwise criterion (PR #698). See also https://arxiv.org/abs/2208.02057.

level-3 BLAS solvers for the triangular system and triangular Sylvester equation by @angsch

The triangular Sylvester equation has been recognized to be prone to overflow. For that purpose, *TRSYL utilizes a scaling factor to represent the solution as $(s^{-1} X)$ and solve the scaled equation $AX + XB = s C$. Due to the scaling factor, there is some flexibility in the representation of the solution. The proposed level-3 BLAS version, *TRSYL3, computes the scaling factors based on the upper bounds of blocks to enable level-3 BLAS. The scaling is typically slightly more aggressive so that an alternatively scaled final solution is computed. This is no problem as long as the scaling factor does not get flushed to zero (PR #651). The same upper bound calculation was used to write the level-3 BLAS solver for the triangular system, *LATRS3.

Givens rotations generated with less accumulation error by @weslleyspereira

New algorithms for computing Givens rotations in complex arithmetic that reduce the accumulation errors for computing each of the outputs, c, s, r. The new algorithms are, on average, more accurate than both the algorithms from LAPACK 3.9.1 and LAPACK 3.10.0 (PR #631). See also https://arxiv.org/abs/2211.04010.

Faster algorithms for Least Squares by @scr2016

The new algorithms, *GELST, are similar to *GELS. *GELST avoids computing triangular blocks twice as in *GELS, which means *GELST runs faster (PR #739).

Improvements on the building system and the Continuous Testing framework

  • Regression test for illegal modification of Y in *GEMV by @matcross in #622.
  • Test the Fortran intrinsic ABS and complex divisions and report failures during build by @weslleyspereira in #623.
  • Appveyor is working in the Continuous Testing once more by @martin-frbg in #627.
  • More tests and fixes for type mismatches by @angsch in #675.
  • Solve build issues with IBM XLF by @friedc in #677.
  • Explicit type casts to INT by @angsch in #684.
  • Fix an out-of-bounds write in the tests by @angsch in #685.
  • Adding support for NAG Fortran compiler (nagfor) by @ACSimon33 in #686.
  • Fix time check flag propagation and default value by @ACSimon33 in #696.
  • Fixed format warnings in 64 bit integer builds by @ACSimon33 in #700.
  • Explicit type casts and more rigorous CI by @weslleyspereira in #703.
  • Fix type mismatches in function calls of testing code by @mjacobse in #706.
  • Fix lower triangular band matrix generation by @angsch in #728.

Other improvements

  • Improves the computation of norms in {S,D}ORBDB6 and {C,Z}UNBDB6 by @christoph-conrads in #647.
  • Uses a Newton step to compute the initial guess for ETA in {S,D}LAED4 to reduce the number of iterations by @weslleyspereira in #655.
  • Return immediately when scaling with one by @angsch in #674.
  • Rearrange the application of the Householder reflectors in *LAQR5, which lowers the instruction count when FMA is available by @angsch in #681.
  • Add {C,Z}ROTG, {CS,ZD}ROT, {S,D}CABS1 to CBLAS by @angsch in #721.
  • Add *LANGB to LAPACKE by @ACSimon33 in #725.
  • NaN check for trapezoidal matrices on LAPACKE by @ACSimon33 in #738 and #742.

Bug fixes

  • Fix LAPACKE_*tpmqrt_work for row-major matrices by @weslleyspereira in #540.
  • Fix leading dimension check in LAPACKE's *geesv[x] and *gges[x] by @angsch in #665.
  • Fix the left-looking variant of GEQRF by @weslleyspereira in #690.
  • Fix workspace query for *SYEVD and *HEEVD routines by @neil-lindquist in #691.
  • Fix behavior of SCALE in *LATBS and *LATRS, and avoids NaN generation if entries in CNORM exceed the overflow threshold by @angsch in #712.

New Contributors

Full Changelog: v3.10.1...v3.11

Thanks to all our contributors!
Thanks to the Mathworks team: Penny Anderson, Mary Ann Freeman, Bobby Cheng, Pat Quillen, Christine Tobler, Heiko Weichelt.

The LAPACK team

Contributors

scr2016, matcross, and 15 other contributors
Assets 2