---
- branch: MAIN
  date: Wed Jan  4 18:31:36 UTC 2023
  files:
  - new: '1.34'
    old: '1.33'
    path: pkgsrc/math/lapack/distinfo
    pathrev: pkgsrc/math/lapack/distinfo@1.34
    type: modified
  - new: '1.5'
    old: '1.4'
    path: pkgsrc/math/lapack/version.mk
    pathrev: pkgsrc/math/lapack/version.mk@1.5
    type: modified
  id: 20230104T183136Z.d78edae63b0bd3501fc8e027001e2f7c66072e93
  log: "lapack & friends: update to 3.11.0\n\n3.11.0\n\nThis is a major release and
    also addressing multiple bug fixes.\n\n2.1. Normwise criterion for infinite eigenvalues
    in the QZ algorithm\nby @thijssteel\n\nThe normwise criterion is more robust at
    detecting infinite\neigenvalues than the elementwise criterion (PR #698). See
    also\nhttps://arxiv.org/abs/2208.02057.\n\n2.2 level-3 BLAS solvers for the triangular
    system and triangular\nSylvester equations by @angsch\n\nThe triangular Sylvester
    equation has been recognized to be prone\nto overflow. For that purpose, *TRSYL
    utilizes a scaling factor to\nrepresent the solution as (s^{-1} X) and solve the
    scaled equation\nAX + XB = s C. Due to the scaling factor, there is some flexibility\nin
    the representation of the solution. The proposed level-3 BLAS\nversion computes
    the scaling factors based on the upper bounds of\nblocks to enable level-3 BLAS.
    The scaling is typically slightly\nmore aggressive so that an alternatively scaled
    final solution is\ncomputed. This is no problem as long as the scaling factor
    does\nnot get flushed to zero (PR #651). The same upper bound calculation\nwas
    used to write the level-3 BLAS solver for the triangular system,\n*LATRS3.\n\n2.3
    Givens rotations generated with less accumulation error by\n@weslleyspereira\n\nNew
    algorithms for computing Givens rotations in complex arithmetic\nthat reduce the
    accumulation errors for computing each of the\noutputs, c, s, r. The new algorithms
    are, on average, more accurate\nthan both the algorithms from LAPACK 3.9.1 and
    LAPACK 3.10.0 (PR\n#631). See also https://arxiv.org/abs/2211.04010.\n\n2.4 Faster
    algorithms for Least Squares by @scr2016\n\nThe new algorithms, *GELST, are similar
    to *GELS. *GELST avoids\ncomputing triangular blocks twice as in *GELS, which
    means *GELST\nruns faster (PR #739).\n\n2.5 Improvements on the building system
    and the Continuous Testing\nframework\n\n    Regression test for illegal modification
    of Y in *GEMV by\n    @matcross in PR #622.\n\n    Test the Fortran intrinsic
    ABS and complex divisions and report\n    failures during build by @weslleyspereira
    in PR #623.\n\n    Appveyor is working in the Continuous Testing once more by\n
    \   @martin-frbg in PR #627.\n\n    More tests and fixes for type mismatches by
    @angsch in PR #675.\n\n    Solve build issues with IBM XLF by @friedc in PR #677.\n\n
    \   Explicit type casts to INT by @angsch in PR #684.\n\n    Fix an out-of-bounds
    write in the tests by @angsch in PR #685.\n\n    Adding support for NAG Fortran
    compiler (nagfor) by @ACSimon33\n    in PR #686.\n\n    Fix time check flag propagation
    and default value by @ACSimon33\n    in PR #696.\n\n    Fixed format warnings
    in 64 bit integer builds by @ACSimon33\n    in PR #700.\n\n    Explicit type casts
    and more rigorous CI by @weslleyspereira\n    in PR #703.\n\n    Fix type mismatches
    in function calls of testing code by\n    @mjacobse in PR #706.\n\n    Fix lower
    triangular band matrix generation by @angsch in PR\n    #728.\n\n2.6 Other improvements\n\n
    \   Improves the computation of norms in {S,D}ORBDB6 and {C,Z}UNBDB6\n    by @Christoph
    Conrads in PR #647.\n\n    Uses a Newton step to compute the initial guess for
    ETA in\n    {S,D}LAED4 to reduce the number of iterations by @weslleyspereira\n
    \   in PR #655.\n\n    Return immediately when scaling with one by @angsch in
    PR #674.\n\n    Rearrange the application of the Householder reflectors in\n    *LAQR5,
    which lowers the instruction count when FMA is available\n    by @angsch in PR
    #681.\n\n    Set SCALE early for robust triangular solvers by @angsch in PR\n
    \   #712.\n\n    Add {C,Z}ROTG, {CS,ZD}ROT, {S,D}CABS1 to CBLAS by @angsch in\n
    \   PR #721.\n\n    Add *LANGB to LAPACKE by @ACSimon33 in PR #725.\n\n    NaN
    check for trapezoidal matrices on LAPACKE by @ACSimon33 in\n    PR #738 and PR
    #742.\n\n2.7 Bug fixes\n\n    Fix LAPACKE_*tpmqrt_work for row-major matrices
    by @weslleyspereira\n    in PR 540.\n\n    Fix leading dimension check in LAPACKE窶å\x86±
    *geesv[x] and *gges[x]\n    by @angsch in PR 665.\n\n    Fix the left-looking
    variant of GEQRF by @weslleyspereira in\n    PR 690.\n\n    Fix workspace query
    for *SYEVD and *HEEVD routines by\n    neil-lindquist@neil-lindquist in PR 691.\n\n
    \   Fix behavior of SCALE in *LATBS and *LATRS, and avoids NaN\n    generation
    if entries in CNORM exceed the overflow threshold\n    by @angsch in PR 712.\n"
  module: pkgsrc
  subject: 'CVS commit: pkgsrc/math/lapack'
  unixtime: '1672857096'
  user: wiz