|
Packit |
67cb25 |
* What is new in gsl-2.5:
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** doc bug fix in binomial distribution figure (Damien Desfontaines)
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** added Wishart distribution (Timothée Flutre)
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** added new module for digital filtering (gsl_filter); current filters include:
|
|
Packit |
67cb25 |
Gaussian filter
|
|
Packit |
67cb25 |
median filter
|
|
Packit |
67cb25 |
recursive median filter
|
|
Packit |
67cb25 |
impulse detection filter
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** added new module for moving window statistics (gsl_movstat)
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** added statistics functions:
|
|
Packit |
67cb25 |
gsl_stats_median()
|
|
Packit |
67cb25 |
gsl_stats_select()
|
|
Packit |
67cb25 |
gsl_stats_mad()
|
|
Packit |
67cb25 |
gsl_stats_mad0()
|
|
Packit |
67cb25 |
gsl_stats_Sn_from_sorted_data()
|
|
Packit |
67cb25 |
gsl_stats_Qn_from_sorted_data()
|
|
Packit |
67cb25 |
gsl_stats_gastwirth_from_sorted_data()
|
|
Packit |
67cb25 |
gsl_stats_trmean_from_sorted_data()
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** added Romberg integration (gsl_integration_romberg)
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** bug fix in deprecated functions gsl_multifit_wlinear_svd and
|
|
Packit |
67cb25 |
gsl_multifit_wlinear_usvd (reported by Vlad Koli)
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** documention corrected to state that gsl_sf_legendre functions do
|
|
Packit |
67cb25 |
not include Condon-Shortley phase by default
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** bug fix in exponential fitting example when using larger number
|
|
Packit |
67cb25 |
of points (reported by Anna Russo)
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** changed internal workspace inside gsl_spmatrix to a union to
|
|
Packit |
67cb25 |
avoid casting (suggested by Manuel Schmitz)
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** bug fixes in ode-initval2 for very rare solver crashing cases:
|
|
Packit |
67cb25 |
#52230 in msadams (reported by Michael Kaufman) and
|
|
Packit |
67cb25 |
#52336 in msbdf (reported by Andrew Benson). As a fix,
|
|
Packit |
67cb25 |
the maximum scaling of controlled step length was decreased
|
|
Packit |
67cb25 |
from 5.0 to 4.9.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** add histogram2d figure to manual (was missing in 2.4)
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** bug fix in gsl_spmatrix_add for duplicate input arguments
|
|
Packit |
67cb25 |
(reported by Alfredo Correa)
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** add support for negative arguments nu in gsl_sf_bessel_Jnu and
|
|
Packit |
67cb25 |
gsl_sf_bessel_Ynu (Konrad Griessinger)
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** better texinfo documentation for gsl_sf_hyperg functions
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** fix vector and matrix fread/fwrite testing on windows systems
|
|
Packit |
67cb25 |
when tmpfile() fails
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** fix for rstat/test.c on PPC64 (reported by Adam Majer)
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
* What was new in gsl-2.4:
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** migrated documentation to Sphinx software, which has built-in
|
|
Packit |
67cb25 |
support for latex equations and figures in HTML output
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** add const to declaration of appropriate gsl_rstat routines
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** bug fix for #45730: change gsl_sf_sin/cos to libm sin/cos
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** fix Cholesky documentation regarding upper triangle on output
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** added routines to compute integrals with fixed-point quadrature,
|
|
Packit |
67cb25 |
based on IQPACK (Konrad Griessinger)
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** added routines for Hermite polynomials, gsl_sf_hermite_*
|
|
Packit |
67cb25 |
(Konrad Griessinger)
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** added new nonlinear least squares example for fitting
|
|
Packit |
67cb25 |
a Gaussian to data
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** deprecated routines:
|
|
Packit |
67cb25 |
gsl_sf_coupling_6j_INCORRECT
|
|
Packit |
67cb25 |
gsl_sf_coupling_6j_INCORRECT_e
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** deprecated routine 'gsl_linalg_hessenberg' (replaced
|
|
Packit |
67cb25 |
by gsl_linalg_hessenberg_decomp)
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** removed routines which were deprecated in v2.1:
|
|
Packit |
67cb25 |
gsl_bspline_deriv_alloc
|
|
Packit |
67cb25 |
gsl_bspline_deriv_free
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** changed COD expression to Q R Z^T instead of Q R Z to
|
|
Packit |
67cb25 |
be consistent with standard texts
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** added check for nz == 0 in gsl_spmatrix_get
|
|
Packit |
67cb25 |
(reported by Manuel Schmitz)
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** permit zero-dimension blocks, vectors, matrics, subvectors,
|
|
Packit |
67cb25 |
submatrices, and views of the above (bug #49988)
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** added routine gsl_linalg_COD_lssolve2 for regularized
|
|
Packit |
67cb25 |
least squares problems
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
* What was new in gsl-2.3:
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** bug fix in documentation for gsl_linalg_LU_refine
|
|
Packit |
67cb25 |
(bug #49728, Joey De Pauw)
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** added gsl_multifit_linear_tsvd and gsl_multifit_wlinear_tsvd
|
|
Packit |
67cb25 |
to give user more control over cutoff for truncated SVD
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** added routines for Generalized Cross Validation for
|
|
Packit |
67cb25 |
regularized linear least squares
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** improved rstat example program and added documentation for
|
|
Packit |
67cb25 |
gsl_rstat_sd_mean (Jonathan Leto)
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** added function gsl_multifit_linear_rank
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** bug fix in nonlinear least squares when using data weights with
|
|
Packit |
67cb25 |
finite-difference Jacobian
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** add 2D subspace method for large systems (multilarge_nlinear)
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** bug fix in gsl_ran_beta for small parameters
|
|
Packit |
67cb25 |
(bug #47646, Yu Liu)
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** bug fix in gsl_complex_tan for negative imaginary arguments
|
|
Packit |
67cb25 |
(bug #47347, Yu Liu)
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** doc bug fix: value of golden ratio
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** fixed scaling issue in 2D subspace nonlinear least squares
|
|
Packit |
67cb25 |
method
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** optimize dogleg methods to calculate Gauss-Newton point
|
|
Packit |
67cb25 |
only when needed
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
* What was new in gsl-2.2.1:
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** reverted gsl_linalg_cholesky_decomp to its previous behavior
|
|
Packit |
67cb25 |
so it is backward compatible; new cholesky routine is
|
|
Packit |
67cb25 |
gsl_linalg_cholesky_decomp1
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
* What was new in gsl-2.2:
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** updated gsl_linalg_cholesky_invert to use Level-2 BLAS
|
|
Packit |
67cb25 |
and added function gsl_linalg_pcholesky_invert
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** added functions gsl_linalg_tri_*_invert for inverting
|
|
Packit |
67cb25 |
triangular matrices
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** fix GSL_EIGEN_SORT_VAL_{ASC,DESC} for nonsymmetric
|
|
Packit |
67cb25 |
eigensystems (Victor Zverovich)
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** added complete orthogonal decomposition routines
|
|
Packit |
67cb25 |
(gsl_linalg_COD)
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** bug fix where median calculation wasn't reset in
|
|
Packit |
67cb25 |
gsl_rstat_reset(); added gsl_rstat_quantile_reset() function
|
|
Packit |
67cb25 |
(reported by Pedro Donato)
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** added multivariate Gaussian random distribution
|
|
Packit |
67cb25 |
gsl_ran_multivariate_gaussian (Timothée Flutre)
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** added functions to estimate the 1-norm reciprocal condition
|
|
Packit |
67cb25 |
number for various matrix factorizations:
|
|
Packit |
67cb25 |
* gsl_linalg_cholesky_rcond
|
|
Packit |
67cb25 |
* gsl_linalg_QRPT_rcond
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** added functions gsl_linalg_QRPT_{lssolve,lssolve2} to
|
|
Packit |
67cb25 |
compute least squares solutions with the QRPT decomposition
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** added function gsl_permute_matrix()
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** added modified Cholesky factorization (gsl_linalg_mcholesky)
|
|
Packit |
67cb25 |
to handle symmetric indefinite matrices
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** added pivoted Cholesky factorization (gsl_linalg_pcholesky)
|
|
Packit |
67cb25 |
for ill-conditioned matrices
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** rewrote (real) Cholesky decomposition to use
|
|
Packit |
67cb25 |
a Level-2 blas algorithm instead of Level-1. Flop
|
|
Packit |
67cb25 |
count is about the same but the code is much simpler
|
|
Packit |
67cb25 |
and easier to follow
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** completely rewritten nonlinear least squares module,
|
|
Packit |
67cb25 |
including support for large problems; the user may
|
|
Packit |
67cb25 |
now control the linear solver used, the trust region
|
|
Packit |
67cb25 |
updating strategy, and the scaling method. In addition,
|
|
Packit |
67cb25 |
support has been added for the geodesic acceleration
|
|
Packit |
67cb25 |
step (Transtrum 2011) which can speed up convergence
|
|
Packit |
67cb25 |
on a wide class of problems.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** added gsl_rstat_rms() for root mean square
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** optimized lmniel nonlinear least squares solver
|
|
Packit |
67cb25 |
(bug #46369)
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** improved precision in Bessel K0/K1 near x = 2
|
|
Packit |
67cb25 |
(Pavel Holoborodko, bug #47401)
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** added support for compressed row storage sparse
|
|
Packit |
67cb25 |
matrices (Alexis Tantet)
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** bug fix in convergence check of hypergeometric 2F1
|
|
Packit |
67cb25 |
function (bug #45926)
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** added gsl_multilarge_linear_lcurve() to compute
|
|
Packit |
67cb25 |
the L-curve for large linear systems
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** updated multilarge normal equations method to use
|
|
Packit |
67cb25 |
new Cholesky scaling for better numerical stability
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** added scaling to Cholesky routines to reduce the
|
|
Packit |
67cb25 |
condition number prior to factorization
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
* What was new in gsl-2.1:
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** added test suite for example programs
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** bug fix when compiling with #undef GSL_DISABLE_DEPRECATED
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** bug fix in setting libtool age versioning
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** bug fix in gsl_multifit_wlinear()
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** added gsl_multifit_linear_rcond() to compute reciprocal
|
|
Packit |
67cb25 |
condition number of least squares matrix
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** added gsl_multilarge module for large linear least squares
|
|
Packit |
67cb25 |
systems
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
* What was new in gsl-2.0:
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** fixed bug #43258 for hypergeometric functions (Raymond Rogers)
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** added L-curve analysis routines for linear Tikhonov regression
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** add running statistics module
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** added bilinear and bicubic interpolation (David Zaslavsky)
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** added function gsl_multifit_robust_residuals to compute robust
|
|
Packit |
67cb25 |
fit residuals
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** added Steffen monotonic interpolation method (Jean-François Caron)
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** added new nonlinear least squares solver 'lmniel' suitable for
|
|
Packit |
67cb25 |
systems with large numbers of data
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** nonlinear least squares solver now tracks the number of function
|
|
Packit |
67cb25 |
and Jacobian evaluations, see example program for details
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** the 'fdf' field of gsl_multifit_function_fdf is now deprecated
|
|
Packit |
67cb25 |
and does not need to be specified for nonlinear least squares
|
|
Packit |
67cb25 |
problems
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** added extensive test suite to nonlinear least squares module,
|
|
Packit |
67cb25 |
resulting in a few minor bug fixes; the routine
|
|
Packit |
67cb25 |
gsl_multifit_fdfsolver_driver has been rewritten (with API change)
|
|
Packit |
67cb25 |
to handle the various error codes of the lmsder iterate
|
|
Packit |
67cb25 |
routine, resulting in a high level caller which is highly robust
|
|
Packit |
67cb25 |
for a wide class of problems
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** added support for sparse matrices, including a GMRES
|
|
Packit |
67cb25 |
iterative linear solver
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** added routines gsl_linalg_givens and gsl_linalg_givens_gv
|
|
Packit |
67cb25 |
for Givens rotations
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** added Tikhonov (ridge) regularization to least squares module
|
|
Packit |
67cb25 |
(linear and nonlinear)
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** removed unused argument 'n' from gsl_sf_ellint_D
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** merged bspline_deriv_workspace into bspline_workspace to simplify
|
|
Packit |
67cb25 |
bspline API; the functions
|
|
Packit |
67cb25 |
gsl_bspline_deriv_alloc
|
|
Packit |
67cb25 |
gsl_bspline_deriv_free
|
|
Packit |
67cb25 |
are now deprecated and will be removed in a future release.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** merged ALF extension into GSL for associated Legendre functions;
|
|
Packit |
67cb25 |
api has changed; consequently the functions:
|
|
Packit |
67cb25 |
gsl_sf_legendre_Plm_array
|
|
Packit |
67cb25 |
gsl_sf_legendre_Plm_deriv_array
|
|
Packit |
67cb25 |
gsl_sf_legendre_sphPlm_array
|
|
Packit |
67cb25 |
gsl_sf_legendre_sphPlm_deriv_array
|
|
Packit |
67cb25 |
gsl_sf_legendre_array_size
|
|
Packit |
67cb25 |
are now deprecated and will be removed in a future release.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** added function gsl_multifit_robust_weights to allow user to
|
|
Packit |
67cb25 |
access the various weighting functions
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
* What was new in gsl-1.16:
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** fixed error in gsl_rng_fwrite where uninitialized padding
|
|
Packit |
67cb25 |
bytes were being written (bug #39104)
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** fixed error in gsl_block_alloc where padding bytes were not
|
|
Packit |
67cb25 |
properly initialized (bugs #39101,#39102,#39103)
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** fixed error in ntuple/test.c where padding bytes were not
|
|
Packit |
67cb25 |
properly initialized (bug #39105)
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** fixed triangle selection bug in gsl_sf_coupling_6j_e and
|
|
Packit |
67cb25 |
gsl_sf_coupling_9j_e (bugs #39466 and #29606) (HÃ¥kan Johansson and
|
|
Packit |
67cb25 |
Alexey Illarionov)
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** added higher level wrapper routine gsl_multifit_fdfsolver_driver
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** converted gsl_multifit_linear_residuals to use dgemv to improve
|
|
Packit |
67cb25 |
efficiency (bug #39153)
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** added functions gsl_stats_spearman and gsl_sort_vector2 to compute
|
|
Packit |
67cb25 |
Spearman rank correlation
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** added function gsl_poly_dd_hermite_init for Hermite interpolation
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added support for robust linear least squares
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added function gsl_linalg_SV_leverage for computing statistical
|
|
Packit |
67cb25 |
leverages from SVD decomposition
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added support for approximating the Jacobian of nonlinear least
|
|
Packit |
67cb25 |
squares fits using forward finite differences
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Extended gsl_sf_coupling_3j to allow larger range and to handle
|
|
Packit |
67cb25 |
the special case (ja jb jc; 0 0 0)=0 when ja+jb+jc is odd
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed gsl_sf_mathieu_se_array to return zero when the order is zero
|
|
Packit |
67cb25 |
[bug #33679].
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed overflow in gsl_sf_lncosh for large negative x (x<-354).
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Improved gsl_ran_negative_binomial_pdf to avoid underflow/overflow
|
|
Packit |
67cb25 |
for large arguments.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Multisets now allow k strictly greater than n.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed gsl_matrix_complex_fwrite/fread failure for noncontiguous
|
|
Packit |
67cb25 |
matrices (Matthias Sitte).
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
* What was new in gsl-1.15:
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added Tuomo Keskitalo's new ode branch ode-initval2 with a
|
|
Packit |
67cb25 |
gsl_odeiv2 prefix. This provides proper support for implicit
|
|
Packit |
67cb25 |
solvers. It is intended to be the new default for differential
|
|
Packit |
67cb25 |
equations. The existing gsl_odeiv routines will be retained for
|
|
Packit |
67cb25 |
binary compatibility but their interface will be deprecated.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added new gsl_integrate_cquad routines for robust integration of
|
|
Packit |
67cb25 |
difficult functions using the doubly-adaptive CQUAD algorithm
|
|
Packit |
67cb25 |
(Pedro Gonnet).
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added error checking to CBLAS functions (José Luis GarcÃa Pallero)
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added a new function gsl_integration_glfixed_point to return
|
|
Packit |
67cb25 |
ordered Gauss-Legendre points and weights contained within a
|
|
Packit |
67cb25 |
gsl_integration_glfixed_table [bug #32237].
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added a new function gsl_interp_type_min_size to return the size of
|
|
Packit |
67cb25 |
an interpolation type.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added a function gsl_pow_uint(x,n) to compute x^n for unsigned
|
|
Packit |
67cb25 |
exponents (needed when n exceeds the range of signed integers).
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added new routine gsl_linalg_complex_cholesky_invert to handle the
|
|
Packit |
67cb25 |
matrix inversion for complex Cholesky decompositions (Huan Wu).
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added the functions gsl_vector_equal(x,y) and gsl_matrix_equal(x,y)
|
|
Packit |
67cb25 |
for testing equality of two vectors or matrices.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added function gsl_eigen_nonsymmv_params to control the balancing
|
|
Packit |
67cb25 |
transformation for eigenvector calculations. Balancing is now
|
|
Packit |
67cb25 |
turned off by default for gsl_eigen_nonsymmv.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** It is now possible to choose an alternative cblas library via
|
|
Packit |
67cb25 |
pkg-config using the GSL_CBLAS_LIB environment variable or
|
|
Packit |
67cb25 |
the pkg-config --define-variable option.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The jacobi method gsl_eigen_jacobi now uses the norm of the
|
|
Packit |
67cb25 |
off-diagonal elements for its convergence criterion, as in
|
|
Packit |
67cb25 |
algorithm 8.4.3 of Golub and van Loan.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The newton multiroot solvers now return an error when a singular
|
|
Packit |
67cb25 |
jacobian is detected.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The interpolation functions now return NaN and when x is out of range,
|
|
Packit |
67cb25 |
instead of extrapolating.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The gsl_multimin_fdfsolver multidimensional minimisers now return
|
|
Packit |
67cb25 |
GSL_ENOPROG immediately if the generated trial point does not
|
|
Packit |
67cb25 |
differ from the initial point (to machine precision), avoiding
|
|
Packit |
67cb25 |
unnecessary further iterations.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Extended the range of gsl_sf_bessel_lnKnu_e by rescaling
|
|
Packit |
67cb25 |
intermediate results to avoid internal overflows [bug #31528].
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Improved the result of gsl_sf_atanint_e for large arguments by
|
|
Packit |
67cb25 |
adding the first order 1/x correction term. [bug #29562]
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed the gsl_rng_ranlxs generators to enforce a maximum seed value
|
|
Packit |
67cb25 |
of 2^31-1. Larger seed values corrupted the random number state
|
|
Packit |
67cb25 |
and caused out of range values to be returned.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed gsl_ran_chisq_pdf(x,nu) to return correct result of 1/2
|
|
Packit |
67cb25 |
instead of 0 when x=0 and nu=2, and +inf when x=0 and nu<2.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed gsl_pow_int(x,n) to avoid an infinite loop when n=INT_MIN due
|
|
Packit |
67cb25 |
to wrapping of signed integers.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed gsl_sf_hyperg_2F1(a,b,c,x) to avoid returning NaN for
|
|
Packit |
67cb25 |
arguments |a|>10. [bug #24812]
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Avoid spurious underflow return code in gsl_sf_beta_inc_e when
|
|
Packit |
67cb25 |
intermediate underflow does not affect the result. [bug #30933]
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Avoid segfault in Chebyshev series derivatives gsl_cheb_calc_deriv
|
|
Packit |
67cb25 |
for n=1. [bug #29139]
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
* What was new in gsl-1.14:
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Upgraded to latest libtool, autoconf and automake (libtool-2.2.6b,
|
|
Packit |
67cb25 |
autoconf-2.65, automake-1.11.1). Fixes security hole in 'make
|
|
Packit |
67cb25 |
dist' (see Automake CVE-2009-4029).
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added support for "multisets", which are similar to permutations
|
|
Packit |
67cb25 |
and combinations. A multiset is an array of k integers in the
|
|
Packit |
67cb25 |
range 0 to n-1 where each value may occur more than once. Multisets
|
|
Packit |
67cb25 |
can be used to iterate over the indices of a k-th order symmetric
|
|
Packit |
67cb25 |
tensor in n-space. (Rhys Ulerich)
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added gsl_integrate_glfixed routines for performing fixed order
|
|
Packit |
67cb25 |
Gauss-Legendre integration
|
|
Packit |
67cb25 |
(Pavel Holoborodko, Konstantin Holoborodko, Rhys Ulerich)
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** In the LMDER multi-dimensional fitting routines, the return code
|
|
Packit |
67cb25 |
GSL_ENOPROG is now used instead of GSL_CONTINUE for the case where
|
|
Packit |
67cb25 |
10 or more attempts to find a suitable trial step have been made
|
|
Packit |
67cb25 |
without success. [bug #25383]
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The confluent hypergeometric function gsl_sf_hyperg_U (a,b,x) has
|
|
Packit |
67cb25 |
been extended to support x < 0 for cases not involving
|
|
Packit |
67cb25 |
singularities in 1F1(a,b,x). [bug #27859]
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The F-distribution gsl_ran_fdist_pdf now avoids unnecessary
|
|
Packit |
67cb25 |
underflow and overflow and handles cases where n > 248. [bug
|
|
Packit |
67cb25 |
#28500]
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The SVD routines now use a scaled version of the implicit QR method
|
|
Packit |
67cb25 |
and compute the shift more reliably for values at the limit of
|
|
Packit |
67cb25 |
double precision, avoiding potential NaNs. [bug #28767]
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** A compile time error is emitted if the configuration stage prevents
|
|
Packit |
67cb25 |
the functions gsl_isnan and gsl_finite from being defined.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added missing dereference in GSL_VECTOR_COMPLEX when not using
|
|
Packit |
67cb25 |
GSL_RANGE_CHECK [bug #28017]
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Improved the range of gsl_sf_hyperg1F1(a,b,x) when a<0,b>0. [bug
|
|
Packit |
67cb25 |
#28718]
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added macros GSL_MAJOR_VERSION and GSL_MINOR_VERSION in
|
|
Packit |
67cb25 |
<gsl/gsl_version.h>
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Improved gsl_eigen_symmv and gsl_eigen_symm to avoid a potential
|
|
Packit |
67cb25 |
infinite loop when the tridiagonalised matrix contains very small
|
|
Packit |
67cb25 |
values. [bug #28096]
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added functions gsl_multifit_linear_usvd and
|
|
Packit |
67cb25 |
gsl_multifit_wlinear_usvd for multilinear fitting without
|
|
Packit |
67cb25 |
column-scaling of the fit matrix.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
* What was new in gsl-1.13:
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Upgraded to latest autoconf and automake (autoconf-2.64,
|
|
Packit |
67cb25 |
automake-1.11)
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed the rk4 and bspline allocators to avoid invalid free()
|
|
Packit |
67cb25 |
calls under out of memory conditions. [bug #27194, #27236]
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed a bug in gsl_multimin_fminimizer_nmsimplex2 where the center
|
|
Packit |
67cb25 |
and size of the simplex were not updated on contract-by-best steps,
|
|
Packit |
67cb25 |
causing failures in convergence. [bug #27180]
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added new functions to set MISER and VEGAS Monte Carlo integration
|
|
Packit |
67cb25 |
parameters, and to examine VEGAS chi-squared value and intermediate
|
|
Packit |
67cb25 |
results.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added the function gsl_bspline_greville_abscissa to compute
|
|
Packit |
67cb25 |
Greville abscissae for B-splines.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The cumulative distribution functions gsl_cdf_gumbel1_{P,Q} should
|
|
Packit |
67cb25 |
now handle a larger range of parameters without underflow and
|
|
Packit |
67cb25 |
overflow.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The header file gsl_const_cgs.h no longer defines values for
|
|
Packit |
67cb25 |
electromagnetic units. Applications should use gsl_const_cgsm.h
|
|
Packit |
67cb25 |
instead to obtain the values in the CGS-Magnetic system. The
|
|
Packit |
67cb25 |
previous values for these units given in gsl_const_cgs.h were
|
|
Packit |
67cb25 |
ill-defined as the type of CGS electromagnetic system was
|
|
Packit |
67cb25 |
unspecified (the values were a mixture of CGS units with the Ampere
|
|
Packit |
67cb25 |
of the MSKA system). The affected constants are
|
|
Packit |
67cb25 |
GSL_CONST_CGS_BOHR_MAGNETON, GSL_CONST_CGS_ELECTRON_CHARGE,
|
|
Packit |
67cb25 |
GSL_CONST_CGS_ELECTRON_MAGNETIC_MOMENT, GSL_CONST_CGS_FARADAY,
|
|
Packit |
67cb25 |
GSL_CONST_CGS_GAUSS, GSL_CONST_CGS_NUCLEAR_MAGNETON,
|
|
Packit |
67cb25 |
GSL_CONST_CGS_PROTON_MAGNETIC_MOMENT, and GSL_CONST_CGS_ROENTGEN.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The Pochhammer functions gsl_sf_poch(a,x) and gsl_sf_lnpoch(a,x) now
|
|
Packit |
67cb25 |
handle the special cases where a and a+x are zero or negative
|
|
Packit |
67cb25 |
integers.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The confluent hypergeometric function gsl_sf_hyperg_U (a,b,x) now
|
|
Packit |
67cb25 |
handles some cases where x=0. The case where 1+a-b is a negative
|
|
Packit |
67cb25 |
integer no longer returns an error [bug #22859] and the incorrect
|
|
Packit |
67cb25 |
termination of the series in certain cases is fixed [bug #26706].
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added a new function gsl_poly_eval_derivs to evaluate a polynomial
|
|
Packit |
67cb25 |
and its derivatives simultaneously.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added a new univariate minimisation algorithm
|
|
Packit |
67cb25 |
gsl_min_fminimizer_quad_golden which is a variant of Brent's
|
|
Packit |
67cb25 |
algorithm with safeguarded step-length adjustment.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added a new Nelder-Mead minimiser gsl_multimin_fminimizer_nmsimplex2rand
|
|
Packit |
67cb25 |
which uses a randomly oriented simplex rather than one fixed on
|
|
Packit |
67cb25 |
the coordinate axes [bug #25077]
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The texinfo file now uses the dircategory "Software libraries" from
|
|
Packit |
67cb25 |
the Free Software Directory, as recommended in the Texinfo manual.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The function gsl_ran_exponential now includes zero in its output
|
|
Packit |
67cb25 |
range. [bug #25039]
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** All functions for freeing allocated memory now accept a NULL
|
|
Packit |
67cb25 |
pointer, following the standard C convention for free(). [bug
|
|
Packit |
67cb25 |
#25319]
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The function gsl_sum_levin_u_accel now handles the special case
|
|
Packit |
67cb25 |
c_0 + 0 + 0 + 0 + .... that occurs when summing power series
|
|
Packit |
67cb25 |
c_n*x^n with x=0. [bug #26807]
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The functions gsl_linalg_LU_solve, gsl_linalg_LU_svx,
|
|
Packit |
67cb25 |
gsl_linalg_LU_refine, gsl_linalg_LU_invert and their complex
|
|
Packit |
67cb25 |
equivalents now return an error for singular matrices.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The multifit LMDER hybrid solvers now check the return code of the
|
|
Packit |
67cb25 |
user-supplied function in the gsl_multifit_fdfsolver_set
|
|
Packit |
67cb25 |
method. [bug #26871]
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Improved the implementation of gsl_ran_discrete_preproc to avoid
|
|
Packit |
67cb25 |
internal errors due to inconsistencies from excess precision on
|
|
Packit |
67cb25 |
some platforms. [bug #26502]
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Corrected gsl_sf_hyperg_2F1(a,b,c,x) to not give a domain error in
|
|
Packit |
67cb25 |
the case where c is a negative integer and the series terminates
|
|
Packit |
67cb25 |
with a finite result.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The C99 inline keyword is now supported, in addition to the
|
|
Packit |
67cb25 |
previously supported GNU-style inline.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Modified gsl_poly_complex_solve_cubic and gsl_poly_solve_cubic to
|
|
Packit |
67cb25 |
avoid returning NaNs in cases where excess precision causes a
|
|
Packit |
67cb25 |
change in the number of roots.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed incorrect length check in gsl_blas_drotm. [bug #26503]
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed gsl_odeiv_step_gear2 to restore y on step failure
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** gsl_odeiv_evolve_apply now restores the correct value of t on step
|
|
Packit |
67cb25 |
failures [bug #26255].
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Using make install prefix=DIR now puts correct paths in package
|
|
Packit |
67cb25 |
config files gsl-config and gsl.pc
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Modified gsl_monte_vegas to work around pow() function inaccuracies
|
|
Packit |
67cb25 |
on MinGW [bug #25413].
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Increased the number of terms in gsl_sf_mathieu_a and
|
|
Packit |
67cb25 |
gsl_sf_mathieu_b to improve convergence in difficult regions [bug
|
|
Packit |
67cb25 |
#25075]
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
* What was new in gsl-1.12:
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Upgraded to latest libtool, autoconf and automake (libtool-2.2.6,
|
|
Packit |
67cb25 |
autoconf-2.63, automake-1.10.2)
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Improved the convergence of gsl_sf_gamma_inc_P for x/a ~=~ 1 and
|
|
Packit |
67cb25 |
large x,a. Fixes problems with large arguments in cdf functions
|
|
Packit |
67cb25 |
such as gsl_cdf_chisq_Pinv(x,nu) [bug 24704].
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed gsl_ran_gamma_knuth to handle the case of a >= UINT_MAX [bug
|
|
Packit |
67cb25 |
#24897]
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added gsl_bspline_eval_deriv to compute bspline derivatives
|
|
Packit |
67cb25 |
(Rhys Ulerich)
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added a faster simplex mininimser gsl_multimin_fminimizer_nmsimplex2
|
|
Packit |
67cb25 |
which is O(N) instead of O(N^2) [bug #24418]
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Improved the original chi-squared formula in gsl_monte_vegas to
|
|
Packit |
67cb25 |
avoid catastrophic cancellation [bug #24510]. The previous formula
|
|
Packit |
67cb25 |
could return incorrect or negative values for relative errors <
|
|
Packit |
67cb25 |
1e-8, which could occur when integrating very smooth functions.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added new auxiliary functions gsl_cheb_order, gsl_cheb_size,
|
|
Packit |
67cb25 |
gsl_cheb_coeffs for Chebyshev series [bug #21830]
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Updated license of the reference manual to GNU FDL version 1.3.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed a bug where the gsl_isinf function would return +1 for -Inf
|
|
Packit |
67cb25 |
on systems where isinf(-Inf) returns the non-standard value +1.
|
|
Packit |
67cb25 |
[bug #24489]
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added missing functions gsl_vector_complex_{isnonneg,add,sub,mul,
|
|
Packit |
67cb25 |
div,scale,add_constant} and gsl_matrix_complex_float_isnonneg [bug
|
|
Packit |
67cb25 |
#22478]
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Cross compilation should now work for x86 hosts.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed a bug in gsl_interp_accel_find() where values lying on the
|
|
Packit |
67cb25 |
upper boundary between interpolation points could return the index
|
|
Packit |
67cb25 |
from the lower side. [bug #24211]
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed gsl_linalg_solve_cyc_tridiag so that its output respects the
|
|
Packit |
67cb25 |
solution vector's stride. Previously the x_stride value was ignored
|
|
Packit |
67cb25 |
causing the output to be incorrect for non-unit stride. [bug #24162]
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Corrected a bug in the series calculation of gsl_sf_ellint_Kcomp
|
|
Packit |
67cb25 |
for k close to 1. [bug #24146]
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Extended gsl_linalg_QRPT_update to handle rectangular matrices.
|
|
Packit |
67cb25 |
Corrected definition of the update formula in the manual for
|
|
Packit |
67cb25 |
both gsl_linalg_QR_update and gsl_linalg_QRPT_update.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added routine gsl_linalg_cholesky_invert
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed a bug the simplex algorithm which caused the second highest
|
|
Packit |
67cb25 |
point to be incorrectly equal to the first when the first value was
|
|
Packit |
67cb25 |
the highest, which could cause suboptimal convergence. [bug #23192]
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed a problem with convergence for inverse gamma and chisq
|
|
Packit |
67cb25 |
distribitions, gsl_cdf_gamma_{P,Q}inv and gsl_cdf_chisq_{P,Q}inv.
|
|
Packit |
67cb25 |
[bug #23101]
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Improved the handling of constant regions in Vegas by eliminating
|
|
Packit |
67cb25 |
spurious excess precision when computing box variances.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed a potential division by zero in gsl_monte_miser when the
|
|
Packit |
67cb25 |
left/right weight factors decrease below 1.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed incorrect dimensions check in gsl_matrix_sub{row,column}
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
* What was new in gsl-1.11:
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The GSL repository and bug database are now hosted at Savannah
|
|
Packit |
67cb25 |
http://savannah.gnu.org/projects/gsl/
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Upgraded to latest libtool, autoconf and automake (libtool-2.2,
|
|
Packit |
67cb25 |
autoconf-2.61, automake-1.10.1)
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed underflow in ODE adaptive step size controller that could
|
|
Packit |
67cb25 |
cause step size to decrease to zero (bug #21933).
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Improved the handling of the asymptotic regime in gsl_sf_bessel_jl.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Improved the handling of large arguments in cumulative distribution
|
|
Packit |
67cb25 |
functions using the incomplete beta function, such as gsl_cdf_fdist_P.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed overflow bug in gsl_cdf_hypergeometric_{P,Q} for large
|
|
Packit |
67cb25 |
arguments (bug #22293).
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** gsl_ran_gaussian_ziggurat now handles generators with different
|
|
Packit |
67cb25 |
ranges explicitly, to minimise the number of function calls
|
|
Packit |
67cb25 |
required (bug #21820). Also fixes bug #22446 (rng limit in
|
|
Packit |
67cb25 |
gsl_ran_chisq()).
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added missing error terms in gsl_sf_exp_mult_e10_e to prevent
|
|
Packit |
67cb25 |
the error being underestimated (bug #22041).
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Updated some constants to the CODATA 2006 values.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The hypergeometric function gsl_sf_hyperg_2F1 now handles the case
|
|
Packit |
67cb25 |
where x==1.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed a bug in the brent minimiser which prevented optimal convergence.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added functions for evaluating complex polynomials
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The convergence condition for gsl_multiroots_test_delta now accepts
|
|
Packit |
67cb25 |
dxi == 0.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Improved functions gsl_ldexp and gsl_frexp to handle the full range
|
|
Packit |
67cb25 |
of double precision numbers in all cases.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added new quasi random generators gsl_qrng_halton and
|
|
Packit |
67cb25 |
gsl_qrng_reversehalton which support dimensions up to 1229.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added function gsl_multifit_linear_residuals for computing the
|
|
Packit |
67cb25 |
residuals of the fit
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
* What was new in gsl-1.10:
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** License updated to GNU GPL version 3.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added support for generalized eigensystems
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added function gsl_stats_correlation to compute Pearson correlation
|
|
Packit |
67cb25 |
of two datasets
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added the new function gsl_sf_expint(n,x) for computing the n-th
|
|
Packit |
67cb25 |
order exponential integral.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added functions gsl_vector_isnonneg and gsl_matrix_isnonneg.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added functions gsl_matrix_subrow and gsl_matrix_subcolumn
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Extended Cholesky routines to complex matrices
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added support in gsl_ieee_set_mode for controlling SSE exceptions
|
|
Packit |
67cb25 |
and rounding through the MXCSR control word on x86 processors.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The autoconf macro AM_PATH_GSL has been renamed to AX_PATH_GSL, to
|
|
Packit |
67cb25 |
avoid conflicts with the autoconf namespace.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Improved handling of underflow in gsl_eigen_symm.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The function gsl_multiroot_fdjacobian now returns the error code
|
|
Packit |
67cb25 |
GSL_ESING if any of the columns of the computed jacobian matrix are
|
|
Packit |
67cb25 |
zero. This may occur if the step size of the derivative is too small.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Extended the function gsl_sf_beta_inc(a,b,x) to handle cases where
|
|
Packit |
67cb25 |
a<0 or b<0.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed the round-off error estimate in gsl_deriv_{central,backwards,
|
|
Packit |
67cb25 |
forward} to correctly account for numerical error in the step-size h.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed gsl_cdf_beta_Pinv, gsl_cdf_gamma_Pinv, gsl_cdf_beta_Pinv to
|
|
Packit |
67cb25 |
avoid returning spurious values for large parameters when the
|
|
Packit |
67cb25 |
iteration did not converge. If the iteration cannot converge, GSL_NAN
|
|
Packit |
67cb25 |
is returned.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** gsl_ran_dirichlet now handles smaller values of alpha[] without
|
|
Packit |
67cb25 |
underflow, avoiding a NaN in the returned value.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The SVD routines now avoid underflow in the Schur decomposition for
|
|
Packit |
67cb25 |
matrices with extremely small values
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** gsl_complex_pow now returns 0^0=1 (instead of zero) to match the usual
|
|
Packit |
67cb25 |
pow function, and handles z^(1,0) and z^(-1,0) as special cases.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed a bug in the set function for multifit lmder solver so that
|
|
Packit |
67cb25 |
previous values are cleared correctly.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Improved the function gsl_log1p to prevent possible loss of
|
|
Packit |
67cb25 |
accuracy caused by optimisations.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Improved the convergence test in the Lambert functions to take
|
|
Packit |
67cb25 |
account of finite precision and avoid possible failure to converge.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The configure script no longer redefines finite() to isfinite() as
|
|
Packit |
67cb25 |
a workaround for missing finite(), as this caused problems on Solaris.
|
|
Packit |
67cb25 |
If finite() is not present gsl_finite() will be used instead.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Improved the accuracy of the generalised laguerre polynomials for
|
|
Packit |
67cb25 |
large n when alpha=0.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The functions gsl_{isnan,isinf,finite} will now always use the
|
|
Packit |
67cb25 |
system versions of isnan, isinf and finite if they are available.
|
|
Packit |
67cb25 |
Previously the portable GSL implementations were used whenever the
|
|
Packit |
67cb25 |
platform supported IEEE comparisons. The portable gsl_isinf function
|
|
Packit |
67cb25 |
now returns -1 instead of +1 for -Inf, in line with recent versions of
|
|
Packit |
67cb25 |
the GNU C Library.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
* What was new in gsl-1.9:
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed the elliptic integrals F,E,P,D so that they have the correct
|
|
Packit |
67cb25 |
behavior for phi > pi/2 and phi < 0. The angular argument is now
|
|
Packit |
67cb25 |
valid for all phi. Also added the complete elliptic integral
|
|
Packit |
67cb25 |
gsl_sf_ellint_Pcomp.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added a new BFGS minimisation method gsl_multimin_fdfminimizer_vector_bfgs2
|
|
Packit |
67cb25 |
based on the algorithm given by R.Fletcher in "Practical Methods of
|
|
Packit |
67cb25 |
Optimisation" (Second edition). This requires substantially fewer
|
|
Packit |
67cb25 |
function and gradient evaluations, and supercedes the existing BFGS
|
|
Packit |
67cb25 |
minimiser.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The beta functions gsl_sf_beta_e(a,b) and gsl_sf_lnbeta_e(a,b) now
|
|
Packit |
67cb25 |
handle negative arguments a,b. Added new function gsl_sf_lnbeta_sgn_e
|
|
Packit |
67cb25 |
for computing magnitude and sign of negative beta values, analagous to
|
|
Packit |
67cb25 |
gsl_sf_lngamma_sgn_e.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** gsl_cheb_eval_mode now uses the same error estimate as
|
|
Packit |
67cb25 |
gsl_cheb_eval_err.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Improved gsl_sf_legendre_sphPlm_e to avoid underflow with large
|
|
Packit |
67cb25 |
arguments.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added updated Knuth generator, gsl_rng_knuthran2002, from 9th
|
|
Packit |
67cb25 |
printing of "The Art of Computer Programming". Fixes various
|
|
Packit |
67cb25 |
weaknesses in the earlier version gsl_rng_knuthran. See
|
|
Packit |
67cb25 |
http://www-cs-faculty.stanford.edu/~knuth/news02.htm
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The functions gsl_multifit_fsolver_set, gsl_multifit_fdfsolver_set
|
|
Packit |
67cb25 |
and gsl_multiroot_fsolver_set, gsl_multiroot_fdfsolver_set now have a
|
|
Packit |
67cb25 |
const qualifier for the input vector x, reflecting their actual usage.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** gsl_sf_expint_E2(x) now returns the correct value 1 for x==0,
|
|
Packit |
67cb25 |
instead of NaN.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The gsl_ran_gamma function now uses the Marsaglia-Tsang fast gamma
|
|
Packit |
67cb25 |
method of gsl_ran_gamma_mt by default.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The matrix and vector min/max functions now always propagate any
|
|
Packit |
67cb25 |
NaNs in their input.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Prevented NaN occuring for extreme parameters in
|
|
Packit |
67cb25 |
gsl_cdf_fdist_{P,Q}inv and gsl_cdf_beta_{P,Q}inv
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Corrected error estimates for the angular reduction functions
|
|
Packit |
67cb25 |
gsl_sf_angle_restrict_symm_err and gsl_sf_angle_restrict_pos_err.
|
|
Packit |
67cb25 |
Fixed gsl_sf_angle_restrict_pos to avoid possibility of returning
|
|
Packit |
67cb25 |
small negative values. Errors are now reported for out of range
|
|
Packit |
67cb25 |
negative arguments as well as positive. These functions now return
|
|
Packit |
67cb25 |
NaN when there would be significant loss of precision.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Corrected an error in the higher digits of M_PI_4 (this was beyond
|
|
Packit |
67cb25 |
the limit of double precision, so double precision results are not
|
|
Packit |
67cb25 |
affected).
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** gsl_root_test_delta now always returns success if two iterates are
|
|
Packit |
67cb25 |
the same, x1==x0.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** A Japanese translation of the reference manual is now available
|
|
Packit |
67cb25 |
from the GSL webpage at http://www.gnu.org/software/gsl/ thanks to
|
|
Packit |
67cb25 |
Daisuke TOMINAGA.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added new functions for basis splines, see the "Basis Splines"
|
|
Packit |
67cb25 |
chapter in the GSL Reference Manual for details.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added new functions for testing the sign of vectors and matrices,
|
|
Packit |
67cb25 |
gsl_vector_ispos, gsl_vector_isneg, gsl_matrix_ispos and
|
|
Packit |
67cb25 |
gsl_matrix_isneg.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed a bug in gsl_sf_lnpoch_e and gsl_sf_lnpoch_sgn_e which caused
|
|
Packit |
67cb25 |
the incorrect value 1.0 instead of 0.0 to be returned for x==0.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed cancellation error in gsl_sf_laguerre_n for n > 1e7 so that
|
|
Packit |
67cb25 |
larger arguments can be calculated without loss of precision.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Improved gsl_sf_zeta_e to return exactly zero for negative even
|
|
Packit |
67cb25 |
integers, avoiding less accurate trigonometric reduction.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed a bug in gsl_sf_zetam1_int_e where 0 was returned instead of
|
|
Packit |
67cb25 |
-1 for negative even integer arguments.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** When the differential equation solver gsl_odeiv_apply encounters a
|
|
Packit |
67cb25 |
singularity it returns the step-size which caused the error code from
|
|
Packit |
67cb25 |
the user-defined function, as opposed to leaving the step-size
|
|
Packit |
67cb25 |
unchanged.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added support for nonsymmetric eigensystems
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added Mathieu functions
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
* What was new in gsl-1.8:
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added an error check to trap multifit calls with fewer observations
|
|
Packit |
67cb25 |
than parameters. Previously calling the multifit routines with n
|
|
Packit |
67cb25 |
would cause invalid memory access.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added the Debye unit to physical constants.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added cumulative distribution functions for the discrete
|
|
Packit |
67cb25 |
distributions, including binomial, poisson, geometric, negative
|
|
Packit |
67cb25 |
binomial, pascal and hypergeometric.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added the functions gsl_cdf_beta_{Pinv,Qinv} and
|
|
Packit |
67cb25 |
gsl_cdf_fdist_{Pinv,Qinv} for computing the inverse of the cumulative
|
|
Packit |
67cb25 |
beta and F distributions.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added the multilinear fit estimator function gsl_multifit_linear_est
|
|
Packit |
67cb25 |
for computing model values and their errors.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Avoid division by zero in gsl_multimin_fdfminimizer_vector_bfgs
|
|
Packit |
67cb25 |
if the step-size becomes too small.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Users on DEC Alpha systems will need to specify their desired IEEE
|
|
Packit |
67cb25 |
arithmetic options via CFLAGS when building the library, as these are
|
|
Packit |
67cb25 |
no longer added automatically.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added new random variate generators gsl_ran_gaussian_ziggurat
|
|
Packit |
67cb25 |
and gsl_ran_gamma_mt for the Gaussian and Gamma distributions based on
|
|
Packit |
67cb25 |
the Marsaglia-Tsang ziggurat and fast gamma methods.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Improved the speed of the exponential power distribution
|
|
Packit |
67cb25 |
gsl_ran_exppow.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Improved the speed of the Gaussian ratio method by adding quadratic
|
|
Packit |
67cb25 |
bounds in gsl_ran_gaussian_ratio_method.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added an extra term to the taylor series of the synchrotron
|
|
Packit |
67cb25 |
functions gsl_sf_synchrotron_1 and gsl_sf_synchrotron_2 for small x to
|
|
Packit |
67cb25 |
ensure smooth matching with the chebyshev expansion.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The binomial pdf gsl_ran_binomial_pdf now handles the cases p=0
|
|
Packit |
67cb25 |
and p=1 and is more accurate for the case of small p with k=0.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed the spherical bessel function gsl_sf_bessel_jl_e) to limit
|
|
Packit |
67cb25 |
the use of gsl_sf_bessel_Jnu_asympx_e to the range x>100*l*l to
|
|
Packit |
67cb25 |
satisfy he requirement x>>l*l in the asymptotic expansion.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The scaled bessel function gsl_sf_bessel_In_scaled now handles
|
|
Packit |
67cb25 |
larger arguments x > 1e7 correctly for n < 150 using the uniform
|
|
Packit |
67cb25 |
asymptotic expansion instead of the continued fraction expansion.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The functions gsl_stats_min/max now return NaN if the data contains
|
|
Packit |
67cb25 |
NaN. Similarly, the functions gsl_stats_min/max_index return the index
|
|
Packit |
67cb25 |
of the first occurring NaN in the data when it contains a NaN.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed an invalid memory access that caused incorrect results for
|
|
Packit |
67cb25 |
the special case in periodic cubic spline interpolation of 3 points.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added Debye functions for n=5 and n=6
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added the missing functions gsl_spline_name() and
|
|
Packit |
67cb25 |
gsl_spline_min_size()
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The function gsl_rng_uniform_int(r,n) now returns an error for n=0,
|
|
Packit |
67cb25 |
which can occur when passing an unsigned integer value of 2^32.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
* What was new in gsl-1.7:
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Switched gsl_randist_binomial to use the faster binomial random
|
|
Packit |
67cb25 |
variate TPE algorithm by default. The previous binomial variate
|
|
Packit |
67cb25 |
algorithm is available as gsl_randist_binomial_knuth. This will
|
|
Packit |
67cb25 |
result in a different sequence of binomial variates in programs using
|
|
Packit |
67cb25 |
this function.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Improved the algorithm for gsl_sf_elljac_e to avoid cancellation
|
|
Packit |
67cb25 |
errors near quarter periods.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed the branch selection in gsl_sf_gamma_inc_Q_e to avoid
|
|
Packit |
67cb25 |
inaccurate results for large a,x where x~=~a.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The multilinear fitting functions now have forms which accept a
|
|
Packit |
67cb25 |
user-specified tolerance for the SVD cutoff and return the
|
|
Packit |
67cb25 |
corresponding effective rank of the design matrix.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The quadratic solvers in poly/ now handle linear equations
|
|
Packit |
67cb25 |
gracefully (i.e. quadratrics with a leading coefficient of zero).
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The output of "make check" now only shows test failures by default,
|
|
Packit |
67cb25 |
to reduce the amount of output. Set the environment variable
|
|
Packit |
67cb25 |
GSL_TEST_VERBOSE=1 to display all the output. To assist debugging,
|
|
Packit |
67cb25 |
the test number of each failure is shown in square brackets at the
|
|
Packit |
67cb25 |
line-end [NNNN].
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed bugs in gsl_linalg_SV_decomp_jacobi which caused
|
|
Packit |
67cb25 |
incorrect results for some input matrices.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Bessel, coulomb, dilogarithm and legendre_H3d functions now use
|
|
Packit |
67cb25 |
hypot internally to avoid overflow when computing terms like
|
|
Packit |
67cb25 |
sqrt(1+x*x).
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The 'Usage' chapter of the reference manual now explains how to
|
|
Packit |
67cb25 |
handle deprecated functions using the GSL_DISABLE_DEPRECATED macro.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The conflicting enum definitions for 'forward' and 'backward' in
|
|
Packit |
67cb25 |
gsl_ftt.h and gsl_wavelet.h are deprecated. User code should switch
|
|
Packit |
67cb25 |
to the new definitions gsl_fft_forward, gsl_fft_backward,
|
|
Packit |
67cb25 |
gsl_wavelet_forward and gsl_wavelet_backward. Selectively define
|
|
Packit |
67cb25 |
GSL_DISABLE_DEPRECATED before including the headers to use the new
|
|
Packit |
67cb25 |
definitions on either or both modules.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed an error in the the brent minimisation algorithm. Iterations
|
|
Packit |
67cb25 |
should now follow Brent's original description correctly.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The bound coulomb function gsl_sf_hydrogenicR_e no longer reports
|
|
Packit |
67cb25 |
an underflow for exact zeroes of the wavefunction.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** gsl_linalg_SV_decomp_jacobi now reports an error for the
|
|
Packit |
67cb25 |
unimplemented case M
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed conformance test for the SYRK and HERK blas functions
|
|
Packit |
67cb25 |
gsl_blas_{s,d,c,z}syrk and gsl_blas_{c,z}herk for non-square matrices.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Configure now checks for presence of ieeefp.h if needed.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Differential equation solvers now propagate error codes returned
|
|
Packit |
67cb25 |
from user-defined functions to the top-level in all cases.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Sort functions now avoid an infinite loop if Nans are present in
|
|
Packit |
67cb25 |
the input vector. The order of nans in the output is undefined,
|
|
Packit |
67cb25 |
although other elements will be sorted correctly.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
* What was new in gsl-1.6:
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added a new wavelet directory, with 1-dimensional and 2-dimensional
|
|
Packit |
67cb25 |
discrete wavelet transforms.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added support for LQ and P^T LQ decompositions. To find the QR
|
|
Packit |
67cb25 |
decomposition of large systems (M>>N) use the LQ decomposition,
|
|
Packit |
67cb25 |
solving the transpose of the original system. This allows more
|
|
Packit |
67cb25 |
efficient memory access, and is useful for solving large least-squares
|
|
Packit |
67cb25 |
problems.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed a bug in the SYRK and HERK blas functions gsl_blas_{s,d,c,z}syrk
|
|
Packit |
67cb25 |
and gsl_blas_{c,z}herk which caused invalid memory access for non-square
|
|
Packit |
67cb25 |
matrices.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed a bug in gsl_swap_vectors which caused it to return incorrect
|
|
Packit |
67cb25 |
results when swapping vectors with different strides.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Corrected the error estimate for gsl_cheb_eval_n_err to use
|
|
Packit |
67cb25 |
evaluation order instead of the approximation order.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Improved the reliability of the gsl_sf_gamma_inc family of
|
|
Packit |
67cb25 |
functions.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Equal abscissae are now handled gracefully in the cspline and
|
|
Packit |
67cb25 |
periodic cspline interpolations.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Removed potential cancellation error in calculation of uniform
|
|
Packit |
67cb25 |
histogram ranges.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Improved numerical stability of integration for akima and cspline
|
|
Packit |
67cb25 |
interpolation.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Differential equation solvers now handle error codes returned from
|
|
Packit |
67cb25 |
user-defined functions.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Improved error estimates in ode-initval solvers, and provide exact
|
|
Packit |
67cb25 |
derivatives on output. Added new semi-implicit ode-initval solver,
|
|
Packit |
67cb25 |
gsl_odeiv_step_rk2simp.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added missing function definition for gsl_sf_psi_1.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed the function gsl_sf_expint_Ei_scaled to call
|
|
Packit |
67cb25 |
gsl_sf_expint_Ei_scaled_e instead of gsl_sf_expint_Ei_e.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added cumulative distribution function for exponential power
|
|
Packit |
67cb25 |
distribution.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The functions gsl_cdf_beta_P and gsl_cdf_beta_Q now return
|
|
Packit |
67cb25 |
consistent results of 0 or 1 for out of range values, x<0 and x>1,
|
|
Packit |
67cb25 |
rather than 0 for left and right tails simultaneously.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The Jacobi eigensolvers gsl_eigen_jacobi and gsl_eigen_jacobi_invert
|
|
Packit |
67cb25 |
have new implementations from Golub and Van Loan.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The standard output and standard error streams are now flushed by
|
|
Packit |
67cb25 |
the default error handler before the program aborts, in order to
|
|
Packit |
67cb25 |
ensure that error messages are properly displayed on some platforms.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
* What was new in gsl-1.5:
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Multifit routines now handle iterations where |f| is already
|
|
Packit |
67cb25 |
minimised to zero, without division by zero.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed the singular value tolerance test in the multifit covariance
|
|
Packit |
67cb25 |
calculation from < to <= to match the original MINPACK code.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The macro HAVE_INLINE is now tested with #ifdef instead of #if as
|
|
Packit |
67cb25 |
in versions prior to 1.4, to match the documentation, and the macro
|
|
Packit |
67cb25 |
GSL_RANGE_CHECK_OFF now works correctly. An alternative macro
|
|
Packit |
67cb25 |
GSL_RANGE_CHECK={0,1} can be used to control range-checking.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed a potential array overflow in gsl_ran_landau.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed a small discrepancy in the tolerance calculation of the
|
|
Packit |
67cb25 |
one-dimensional brent minimiser.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Numerical derivatives should now be calculated using the
|
|
Packit |
67cb25 |
gsl_deriv_forward, gsl_deriv_central and gsl_deriv_backward functions,
|
|
Packit |
67cb25 |
which accept a step-size argument in addition to the position x. The
|
|
Packit |
67cb25 |
original gsl_diff functions (without the step-size) are deprecated.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Corrected documentation for gsl_ran_hypergeometric_pdf()
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The tridiagonal matrix solvers gsl_linalg_solve_symm_tridiag,
|
|
Packit |
67cb25 |
gsl_linalg_solve_tridiag, gsl_linalg_solve_symm_cyc_tridiag,
|
|
Packit |
67cb25 |
gsl_linalg_solve_cyc_tridiag now use the GSL_ERROR macro to report
|
|
Packit |
67cb25 |
errors, instead of simply returning an error code. The arguments to
|
|
Packit |
67cb25 |
these functions must now use exact lengths with no additional
|
|
Packit |
67cb25 |
elements. For cyclic systems all vectors must be of length N, for
|
|
Packit |
67cb25 |
tridiagonal systems the offdiagonal elements must be of length N-1.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The singular value decomposition routines gsl_linalg_SV_decomp and
|
|
Packit |
67cb25 |
gsl_linalg_SV_decomp_mod now handle the SVD of a column vector (N=1,
|
|
Packit |
67cb25 |
arbitrary M), which can occur in linear fitting.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Restored missing header files gsl_const_mks.h and gsl_const_cgs.h.
|
|
Packit |
67cb25 |
The incorrect values of the electrical units for gsl_const_cgs
|
|
Packit |
67cb25 |
(VACUUM_PERMEABILITY and VACUUM_PERMITTIVITY) have been removed.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed gsl_linalg_SV_decomp() to avoid an infinite loop when
|
|
Packit |
67cb25 |
computing the SVD of matrices containing Inf and Nan.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed gsl_linalg_balance_columns() to avoid an infinite loop when
|
|
Packit |
67cb25 |
rescaling matrices containing Inf and NaN.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed header file <gsl/gsl_sf_log.h> to include declarations for
|
|
Packit |
67cb25 |
error codes in inline versions of gsl_sf_log functions
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed header file <gsl/gsl_const.h> to include new MKSA and CGSM
|
|
Packit |
67cb25 |
header files.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added Stefan-Boltzmann constant and Thomson cross section to
|
|
Packit |
67cb25 |
physical constants
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
* What was new in gsl-1.4:
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added cumulative distribution functions and their inverses for the
|
|
Packit |
67cb25 |
continuous random distributions including: gaussian, lognormal, gamma,
|
|
Packit |
67cb25 |
beta, cauchy, laplace, chisq, exponential, gumbel, weibull,
|
|
Packit |
67cb25 |
F-distribution, t-distribution, logistic, pareto and rayleigh.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added faster binomial random variates using the TPE rejection
|
|
Packit |
67cb25 |
algorithm, in the function gsl_randist_binomial_tpe.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added new functions gsl_rng_fwrite and gsl_rnd_fread for storing
|
|
Packit |
67cb25 |
the state of random number generators in a file.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added a new function gsl_combination_memcpy()
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Corrected values of electrical constants in CGS units. To take
|
|
Packit |
67cb25 |
account of different electrical systems of units the values are now
|
|
Packit |
67cb25 |
prefixed by GSL_CONST_MKSA (for the SI Metre, Kilogram, Second, Ampere
|
|
Packit |
67cb25 |
system) or GSL_CONST_CGSM (for the Centimetre, Gram, Second, Magnetic
|
|
Packit |
67cb25 |
system with the Gauss as the fundamental unit of magnetic field
|
|
Packit |
67cb25 |
strength). The previous GSL_CONST_MKS and GSL_CONST_CGS prefixes have
|
|
Packit |
67cb25 |
been removed, as have the permeability and permittivity constants in
|
|
Packit |
67cb25 |
the CGS system since this uses different defining equations.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed bugs in the random number generators gsl_rng_fishman18,
|
|
Packit |
67cb25 |
gsl_rng_fishman2x, and gsl_rng_knuthran2 which caused them to return
|
|
Packit |
67cb25 |
incorrect results. Minor corrections were made to the parameters in
|
|
Packit |
67cb25 |
the other Knuth generators borosh13, coveyou, fishman20, lecuyer21,
|
|
Packit |
67cb25 |
and waterman14.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed a missing transpose bug in the gsl_linalg_QR_QRsolve
|
|
Packit |
67cb25 |
and gsl_linalg_QRPT_QRsolve routines which were computing the
|
|
Packit |
67cb25 |
solution to Q^T R x = b instead of Q R x = b.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed gsl_sf_gammainv to return zero instead of a domain
|
|
Packit |
67cb25 |
error for arguments corresponding to singularities in gamma.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed a bug in the simplex minimization algorithm which
|
|
Packit |
67cb25 |
caused it to fail to find the second highest point correctly
|
|
Packit |
67cb25 |
when searching the set of simplex points.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed a bug in the conjugate gradient minimizers conjugate_pr,
|
|
Packit |
67cb25 |
conjugate_fr and vector_bgfs which caused the search
|
|
Packit |
67cb25 |
directions to be updated incorrectly.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed a bug in gsl_sf_psi_1_int(1) which caused it to
|
|
Packit |
67cb25 |
return the incorrect sign for psi(1,1).
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed the simulated annealing routine gsl_siman_solve to use the
|
|
Packit |
67cb25 |
parameter iters_fixed_T for the number of iterations at fixed
|
|
Packit |
67cb25 |
temperature instead of n_tries.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed a bug in gsl_combination_valid which caused it to return the
|
|
Packit |
67cb25 |
incorrect status.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed a bug in gsl_permutation_canonical_to_linear which caused the
|
|
Packit |
67cb25 |
output to always be zero, and the input permutation to be incorrectly
|
|
Packit |
67cb25 |
replaced by the output.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed a bug is gsl_ran_discrete which could cause uninitialised
|
|
Packit |
67cb25 |
data to be returned for some distributions.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed the dependencies for gsl_chebyshev.h to include gsl_math.h.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed a bug in gsl_complex_arccsc_real which caused it to return
|
|
Packit |
67cb25 |
the incorrect sign for the imaginary part when -1
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed a bug in the QAWC Cauchy integration routine which could
|
|
Packit |
67cb25 |
allow the singularity to fall on an interval boundary, leading to
|
|
Packit |
67cb25 |
division by zero.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Improved gsl_sf_gamma_inc_P(a,x) to avoid a domain error for x<
|
|
Packit |
67cb25 |
when a>10.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Improved the accuracy of gsl_sf_coupling_3j for large arguments.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Improved the performance of gsl_sf_choose(m,n) by separating the
|
|
Packit |
67cb25 |
calculations for small and large arguments.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** On platforms without IEEE comparisons gsl_{isnan,isinf,finite} will
|
|
Packit |
67cb25 |
fall back to the system versions of isnan, isinf and finite if
|
|
Packit |
67cb25 |
available.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** gsl_linalg_householder_hv now uses BLAS routines internally
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The script configure.in is now compatible with autoconf-2.50 and
|
|
Packit |
67cb25 |
later.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Reduced the memory usage of the multifit algorithms from MxM to MxN
|
|
Packit |
67cb25 |
for large M by performing the QR decomposition of the Jacobian
|
|
Packit |
67cb25 |
in-place.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** IEEE modes now use the C99 fenv.h functions when platform spectific
|
|
Packit |
67cb25 |
functions are not available.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
* What was new in gsl-1.3:
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Changed interface for gsl_sf_coupling_6j...(...). The old functions
|
|
Packit |
67cb25 |
actually calculated 6j for a permutation of the arguments (that
|
|
Packit |
67cb25 |
related to Racah W). This was incorrect and not consistent with
|
|
Packit |
67cb25 |
the documentation. The new versions calculate < {a,b,c}, {d,e,f} >,
|
|
Packit |
67cb25 |
as stated in the documentation. The old versions are still available
|
|
Packit |
67cb25 |
as gsl_sf_coupling_6j_INCORRECT...(...), though they are deprecated
|
|
Packit |
67cb25 |
and will be removed at some point in the future.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added new functions for computing Em(x)=exp(-x)*Ei(x), the modified
|
|
Packit |
67cb25 |
(scaled) form of the exponential integral, gsl_sf_expint_E1_scaled,
|
|
Packit |
67cb25 |
gsl_sf_expint_E2_scaled, gsl_sf_expint_Ei_scaled.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed compilation problems with gcc -ansi and other ANSI compilers.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed uninitialized memory access in the Niederreiter quasi-random
|
|
Packit |
67cb25 |
number generator.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed the eigenvalue routines to prevent an infinite loop for Inf
|
|
Packit |
67cb25 |
or NaN entries in matrix.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed a bug in the multifit and multiroots allocation routines
|
|
Packit |
67cb25 |
which cause them to fail to report some out of memory conditions.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed a bug in the seeding for the random number generator
|
|
Packit |
67cb25 |
gsl_rng_taus2 which affected a small number of seeds.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Modified the complex householder transforms to avoid division by
|
|
Packit |
67cb25 |
zero, which could cause NaNs to be returned by the gsl_eigen_hermv
|
|
Packit |
67cb25 |
eigenvalue decomposition.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The Nelder-Mead simplex algorithm for multidimensional
|
|
Packit |
67cb25 |
minimisation has been added.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The random number distributions now include the Dirichlet and
|
|
Packit |
67cb25 |
Multinomial distributions.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added a new function gsl_fcmp for approximate comparison of
|
|
Packit |
67cb25 |
floating point numbers using Knuth's algorithm.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added new functions gsl_ldexp and gsl_frexp as portable
|
|
Packit |
67cb25 |
alternatives to ldexp() and frexp().
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed a bug in gsl_linalg_bidiag_unpack_B which was returning
|
|
Packit |
67cb25 |
incorrect results for the superdiagonal.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed a bug in the acceptance condition for simulated annealing
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Ordinary differential equations can now be solved using a different
|
|
Packit |
67cb25 |
absolute error for each component with gsl_odeiv_control_scaled_new().
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Upgraded to libtool-1.4.3
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
* What was new in gsl-1.2:
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added new functions for combining permutations, converting between
|
|
Packit |
67cb25 |
cyclic and linear representations, and counting cycles and inversions.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** New multiroot functions now allow access to the current values of f
|
|
Packit |
67cb25 |
and dx.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The default error handler now outputs a explanatory message before
|
|
Packit |
67cb25 |
aborting.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Extended gsl_linalg_SV_decomp to handle exact zeroes in the
|
|
Packit |
67cb25 |
singular values, and added tests for 3x3 matrices.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed a bug in gsl_linalg_SV_decomp which caused singular values to
|
|
Packit |
67cb25 |
be sorted incorrectly.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed a bug in gsl_linalg_solv_symm_cyc_tridiag which caused it to
|
|
Packit |
67cb25 |
produce incorrect results.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added nonsymmetric tridiagonal solvers gsl_linalg_solve_tridiag and
|
|
Packit |
67cb25 |
gsl_linalg_solve_cyc_tridiag.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The declarations used to export static objects can now be
|
|
Packit |
67cb25 |
controlled through a macro GSL_VAR and the header file
|
|
Packit |
67cb25 |
<gsl/gsl_types.h>.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The simulated annealing routine gsl_siman_solve now keeps track of
|
|
Packit |
67cb25 |
the best solution so far.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The values of the physical constants have been updated to the
|
|
Packit |
67cb25 |
CODATA 1998 recommendations.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added new physical constants, newton, dyne, joule, erg and
|
|
Packit |
67cb25 |
power-of-ten prefixes, Mega, Giga, Tera, etc.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The error estimate for the elliptic function gsl_sf_ellint_Kcomp_e
|
|
Packit |
67cb25 |
has been improved to take account of numerical cancellation for small
|
|
Packit |
67cb25 |
arguments.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The domain of gsl_sf_psi_1piy has been extended to negative y.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed memory leak in the Chebyshev module.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The seeding procedure of mt19937 has been updated to the latest
|
|
Packit |
67cb25 |
version from Makoto Matsumoto and Takuji Nishimura (Jan 2002). The
|
|
Packit |
67cb25 |
original seeding procedure is available through the generator
|
|
Packit |
67cb25 |
gsl_rng_mt19937_1999.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** A new random number generator gsl_rng_taus2 has been added to
|
|
Packit |
67cb25 |
correct flaws in the seeding procedure of gsl_rng_taus, as described
|
|
Packit |
67cb25 |
in an erratum to the original paper of P. L'Ecuyer.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added missing declaration for the generator gsl_rng_mt_19937_1998.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added missing quasi-random number generator function gsl_qrng_init.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Removed unnecessary endpoint subtraction in chebyshev-based
|
|
Packit |
67cb25 |
QUADPACK routines to avoid possible loss of precision.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed bug in gsl_interp_cspline_periodic which caused a
|
|
Packit |
67cb25 |
discontinuity in the derivative near the boundary.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The function gsl_min_fminimizer_minimum has been renamed to
|
|
Packit |
67cb25 |
gsl_min_fminimizer_x_minimum for consistency (the old function name is
|
|
Packit |
67cb25 |
still available but is deprecated). Additional functions have been
|
|
Packit |
67cb25 |
added for accessing the function values at the minimum and endpoints
|
|
Packit |
67cb25 |
of the bounding interval.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The KNOWN-PROBLEMS file of "make check" failures has been replaced
|
|
Packit |
67cb25 |
by a BUGS file, since we now require "make check" to work correctly
|
|
Packit |
67cb25 |
for stable releases.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
* What was new in gsl-1.1.1:
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixes to histogram2d stat functions
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added missing prototypes for complex LU determinant functions
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Improved error handling in multifit routines
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added check to avoid division by zero for rank-deficient matrix in
|
|
Packit |
67cb25 |
multifit iteration
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
* What was new in gsl-1.1:
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The permutation module now includes a copy function
|
|
Packit |
67cb25 |
gsl_permutation_memcpy
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The implementation of gsl_sf_gamma_inc has been improved and now
|
|
Packit |
67cb25 |
avoids problems caused by internal singularities which occurred in the
|
|
Packit |
67cb25 |
series expansion for some combinations of parameters.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** IEEE comparisons of infinities and NaNs are tested during the
|
|
Packit |
67cb25 |
configure stage and the functions gsl_isnan, gsl_isinf and gsl_finite
|
|
Packit |
67cb25 |
are only compiled on platforms which support the necessary tests.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The histogram routines now include a sum function,
|
|
Packit |
67cb25 |
gsl_histogram_sum for computing the total bin sum, and additional
|
|
Packit |
67cb25 |
statistics functions for 2d histograms.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Internal error checking of user-defined functions has been improved
|
|
Packit |
67cb25 |
in the multiroots functions.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Constants now include the Bohr Radius and Vacuum Permittivity.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Range checking is now turned off when building the library, but is
|
|
Packit |
67cb25 |
still on by default when compiling user applications.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** A combinations directory has been added for generating combinations (n,k).
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The gamma function now returns exact values for integer arguments.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed bugs in gsl_sf_hyperg_1F1_int and gsl_sf_hyperg_1F1.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed internal error handling in gsl_sf_laguerre_n to allow
|
|
Packit |
67cb25 |
recovery from overflow.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Several routines for handling divided difference polynomials have
|
|
Packit |
67cb25 |
been added to the poly/ directory.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The interpolation routines now include polynomial interpolation,
|
|
Packit |
67cb25 |
based on divided-differences.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added new random number generators from Knuth's Seminumerical
|
|
Packit |
67cb25 |
Algorithms, 3rd Edition: borosh13, coveyou, fishman18, fishman20,
|
|
Packit |
67cb25 |
fishman2x, knuthran, knuthran2, lecuyer21, waterman14.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Changed divisor in random number generator gfsr4 from 2^32-1 to
|
|
Packit |
67cb25 |
2^32 to prevent exact value of 1.0 from being returned, as specified
|
|
Packit |
67cb25 |
in the documentation.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
* What was new in gsl-1.0:
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** First general release.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Increased the maximum number of iterations in gsl_poly_complex_solve()
|
|
Packit |
67cb25 |
from 30 to 60.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
* What was new in gsl-0.9.4:
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Reorganized the multmin functions to use the same interface as the
|
|
Packit |
67cb25 |
other iterative solvers.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added histogram _alloc functions for consistency, in addition to the
|
|
Packit |
67cb25 |
existing _calloc functions.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Renamed all the gsl_multimin functions to be consistent with the
|
|
Packit |
67cb25 |
rest of the library. An underscore has been removed from _minimizer
|
|
Packit |
67cb25 |
in all the function names.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Renamed the function gsl_sf_coulomb_CL_list to gsl_sf_coulomb_CL_array
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** A bug in the multimin functions where the function parameters
|
|
Packit |
67cb25 |
(params) were omitted has been fixed.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** A bug in the nonlinear minimization routines has been fixed, which
|
|
Packit |
67cb25 |
could prevent the algorithms from converging. Additional tests from
|
|
Packit |
67cb25 |
the NIST reference datasets have been added and these now agree with
|
|
Packit |
67cb25 |
MINPACK.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** All the physical constants and conversion factors are now defined as
|
|
Packit |
67cb25 |
real numbers to avoid potential problems with integer arithmetic.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The ODE evolution routines now allow for negative step sizes, and
|
|
Packit |
67cb25 |
integrating backwards as well as forwards.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The implicit Burlisch-Stoer ODE algorithm 'bsimp' now detects
|
|
Packit |
67cb25 |
singularities and forces a reduction in step size, preventing runaway
|
|
Packit |
67cb25 |
instabilities.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed a bug in the ODE evolution function gsl_odeiv_evolve_apply
|
|
Packit |
67cb25 |
which could cause an erroneous value to be returned if the step size
|
|
Packit |
67cb25 |
is reduced on the last step.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
* What was new in gsl-0.9.3:
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Routines for complex LU decomposition are now available, allowing
|
|
Packit |
67cb25 |
the solution of systems of equations with complex coefficients.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Matrix views of vectors now correctly require a unit stride for the
|
|
Packit |
67cb25 |
original vector.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Permutations can now be applied to complex arrays and vectors.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** gsl_sf_pow_int now handles the case x = 0, n < 0
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The static versions of inline functions can now be hidden by
|
|
Packit |
67cb25 |
defining the preprocessor macro HIDE_INLINE_STATIC. This is needed
|
|
Packit |
67cb25 |
for some compilers.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The original seeding procedure of mt19937 is available through the
|
|
Packit |
67cb25 |
generator gsl_rng_mt19937_1998. The seeding procedure was flawed, but
|
|
Packit |
67cb25 |
is available for compatibility.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added missing functions gsl_complex_div_real and
|
|
Packit |
67cb25 |
gsl_complex_div_imag.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Missing functions for constant vector and matrix views have now been
|
|
Packit |
67cb25 |
added.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Statistical calculations for histograms are now available, and the
|
|
Packit |
67cb25 |
gsl-histogram command also displays the histogram mean and standard
|
|
Packit |
67cb25 |
deviation.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The behavior of GSL_IEEE_MODE for denormalized exceptions has been
|
|
Packit |
67cb25 |
fixed on Openbsd and Netbsd.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** A pkg-config file gsl.pc is included in the distribution
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The reference manual can now be printed in @smallbook format without
|
|
Packit |
67cb25 |
overflow.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
* What was new in gsl-0.9.2:
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Vector and matrix views are now compliant with the ANSI standard.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added Lambert functions gsl_sf_lambert_W0, gsl_sf_lambert_Wm1.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The reference manual now uses the GNU Free Documentation License.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Fixed a couple of bugs in the SVD routines.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Macros for Infinity and Nan now work correctly with Microsoft Visual
|
|
Packit |
67cb25 |
C++, and a bug in the config.h file for the finite() function has been
|
|
Packit |
67cb25 |
fixed.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Redundant entries in the test suite for the complex math functions
|
|
Packit |
67cb25 |
have been removed, making the distribution size smaller.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Installed programs gsl-randist and gsl-histogram now use shared
|
|
Packit |
67cb25 |
libraries.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
* What was new in gsl-0.9.1:
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The single precision ffts now uses float throughout, rather than
|
|
Packit |
67cb25 |
mixing float and double.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The random number distributions now include the Landau distribution.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The fft function interface has been reorganized, with workspaces
|
|
Packit |
67cb25 |
separate from wavetables to eliminate unnecessary recomputation of
|
|
Packit |
67cb25 |
trigonometric factors.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The gsl_interval type has been eliminated and replaced by two double
|
|
Packit |
67cb25 |
arguments for simplicity.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The order of the arguments to the minimization routines is no more
|
|
Packit |
67cb25 |
logical, with function values assocatied with x-values.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Modified initialization of vector and matrix views to work with the
|
|
Packit |
67cb25 |
SunPro compiler.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Renamed gsl_Efunc_t to gsl_siman_Efunc_t, in accordance with
|
|
Packit |
67cb25 |
namespace conventions.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Improved accuracy and fixed bugs in gsl_sf_hyperg_1F1,
|
|
Packit |
67cb25 |
gsl_sf_bessel_I0_scaled, gsl_sf_erfc, gsl_sf_log_erfc,
|
|
Packit |
67cb25 |
gsl_sf_legendre_Q0 and gsl_sf_legendre_Q1, and gsl_sf_zeta.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Improved IEEE compliance of special functions, overflows now return
|
|
Packit |
67cb25 |
Inf and domain errors return NaN.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Improved checking for underflows in special functions when using
|
|
Packit |
67cb25 |
extended precision registers
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
* What was new in gsl-0.9:
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** There is a new system of vector and matrix views. Any code using
|
|
Packit |
67cb25 |
vector and matrix views will need to be updated.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The order of arguments of the view functions involving strides have
|
|
Packit |
67cb25 |
been changed to be consistent with the rest of the library.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The ode solvers have been reorganized.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** There are new eigensystem routines for real symmetric and complex
|
|
Packit |
67cb25 |
hermitian matrices.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The linear algebra directory now includes functions for computing
|
|
Packit |
67cb25 |
symmetric tridiagonal decompositions and bidiagonal decompositions.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The svd routines now include the Golub-Reinsch and Modified
|
|
Packit |
67cb25 |
Golub-Reinsch algorithms in addition to the Jacobi algorithm.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The interpolation directory has been reorganized and a higher-level
|
|
Packit |
67cb25 |
"spline" interface has been added which simplifies the handling of
|
|
Packit |
67cb25 |
interpolation arguments.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** IEEE support is now available on OpenBSD.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
* What was new in gsl-0.8:
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The build process now uses the latest libtool and automake.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The library should now compile with Microsoft Visual C++.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Portable versions of the isinf, isnan and finite functions are
|
|
Packit |
67cb25 |
available as gsl_isinf(x), gsl_isnan(x) and gsl_finite(x).
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The definitions of GSL_POSINF, GSL_NEGINF and GSL_NAN no longer
|
|
Packit |
67cb25 |
cause divisions by zero during compilation.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The gsl_interp_obj has been renamed to gsl_interp.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The poly_eval and pow_int functions have been moved from the
|
|
Packit |
67cb25 |
specfunc directory to the poly and sys directories.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The Chebyshev functions are now available as an independent module
|
|
Packit |
67cb25 |
in their own directory.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The error handling conventions have been unified across the
|
|
Packit |
67cb25 |
library. This simplifies the use of the special functions.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** A full CBLAS implementation is now included for systems where ATLAS
|
|
Packit |
67cb25 |
has not been installed. The CBLAS library can also be used
|
|
Packit |
67cb25 |
independently of GSL. The organisation of the BLAS directories has been
|
|
Packit |
67cb25 |
simplified.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** IEEE support for HPUX-11, NetBSD, Apple Darwin and OS/2 are now
|
|
Packit |
67cb25 |
included.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The library now includes implementations of log1p, expm1, hypot,
|
|
Packit |
67cb25 |
acosh, asinh, atanh for platforms which do not provide them.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The convention for alloc and set functions has changed so that they
|
|
Packit |
67cb25 |
are orthogonal. After allocating an object it is now necessary to
|
|
Packit |
67cb25 |
initialize it.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** There is a new module for estimating numerical derivatives of functions
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** There is a new module for handling data with ntuples
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The histogram lookup functions are now optimized for the case of
|
|
Packit |
67cb25 |
uniform bins, and include an inline binary search for speed.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The Chebyschev coefficients for the QAWO algorithm are now
|
|
Packit |
67cb25 |
precomputed in a table for efficiency, rather than being computed on
|
|
Packit |
67cb25 |
the fly.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** There are several new sorting functions for selecting the k-th
|
|
Packit |
67cb25 |
smallest or largest elements of a dataset.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Iterator functions are now available for permutations,
|
|
Packit |
67cb25 |
gsl_permutation_next and gsl_permutation_prev.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The function gsl_complex_xy has been renamed gsl_complex_rect
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The API for simulated annealing has been changed to support search
|
|
Packit |
67cb25 |
spaces in which the points cannot be represented as contiguous-memory
|
|
Packit |
67cb25 |
data structures. gsl_siman_solve() now takes three extra arguments: a
|
|
Packit |
67cb25 |
copy constructor, a copy function and a destructor, allowing
|
|
Packit |
67cb25 |
gsl_siman_solve() to do its work with linked data structures. If all
|
|
Packit |
67cb25 |
three of these function pointers are NULL, then the traditioanl
|
|
Packit |
67cb25 |
approach of using malloc(), memcpy(), and free() with the element size
|
|
Packit |
67cb25 |
is used.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
* What was new in gsl-0.7:
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Linux/PowerPC should now be well supported.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Header files for common physical constants have been added.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Functions linear and nonlinear regression in one or more dimensions
|
|
Packit |
67cb25 |
are now available.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Vector and matrix views now have access to the address of the
|
|
Packit |
67cb25 |
underlying block for compatibility with VSIPL (www.vsipl.org).
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** There is a new library for generating low-discrepancy quasi-random
|
|
Packit |
67cb25 |
sequences.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The seeding procedure of the default random number generator
|
|
Packit |
67cb25 |
MT19937 has been updated to match the 10/99 release of the original
|
|
Packit |
67cb25 |
code. This fixes a weakness which occurred for seeds which were
|
|
Packit |
67cb25 |
powers of 2.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The blas library libgslblasnative has been renamed libgslblas to avoid
|
|
Packit |
67cb25 |
confusion with system blas library
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
* What was new in gsl-0.6:
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The library is now installed as a single shared or static libgsl
|
|
Packit |
67cb25 |
file using libtool.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The gsl-config script now works. There is also a gsl.m4 file which
|
|
Packit |
67cb25 |
people can use in their configure scripts.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** All header files are now in installed as pkginclude headers in a
|
|
Packit |
67cb25 |
gsl/ subdirectory.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The header files now use extern "C" to allow them to be included in
|
|
Packit |
67cb25 |
C++ programs
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** For consistency the following functions have been renamed,
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
gsl_vector_copy (dest, src) is now gsl_vector_memcpy (dest, src)
|
|
Packit |
67cb25 |
gsl_rng_cpy (dest, src) is now gsl_rng_memcpy (dest, src)
|
|
Packit |
67cb25 |
gsl_matrix_copy_row (v,m,i) is now gsl_matrix_get_row (v,m,i)
|
|
Packit |
67cb25 |
gsl_matrix_copy_col (v,m,j) is now gsl_matrix_get_col (v,m,j)
|
|
Packit |
67cb25 |
gsl_vector_swap is now gsl_vector_swap_elements
|
|
Packit |
67cb25 |
gsl_vector_swap_cols is now gsl_vector_swap_columns
|
|
Packit |
67cb25 |
gsl_vector_swap_row_col is now gsl_vector_swap_row_column
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
and the vector/matrix view allocation functions have been simplified.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** A new sort directory has been added for sorting objects and vectors.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** A permutation directory has been added for manipulating permutations
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Statistics functions now support a stride argument for generality, and
|
|
Packit |
67cb25 |
also support weighted samples and a covariance function.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The names of the statistics functions have been reorganized for
|
|
Packit |
67cb25 |
improved clarity. Consult manual for details.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The environment variable GSL_IEEE_MODE now uses "," as a separator
|
|
Packit |
67cb25 |
instead of ";"
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The autogen.sh script, mostly for use by developers who use the CVS
|
|
Packit |
67cb25 |
repository, now does not run configure.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The histogram directory now has additional functions for copying
|
|
Packit |
67cb25 |
and comparing histograms, performing arithmetic on histograms and
|
|
Packit |
67cb25 |
finding maximum and minimum values. Corresponding functions have been
|
|
Packit |
67cb25 |
added for vectors and matrices.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The linear algebra directory supports additional methods, including
|
|
Packit |
67cb25 |
rectangular QR, rectangular QRPT and Cholesky decomposition.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Complex arithmetic (+,-,*,/) and complex elementary functions
|
|
Packit |
67cb25 |
(sqrt, log, exp, sin, cos, tan, arcsin, arccos, arctan, sinh, cosh,
|
|
Packit |
67cb25 |
tanh, arcsinh, arccosh, arctanh) are now supported.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Multidimensional minimization methods are now available.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The special functions directory now includes a routine for
|
|
Packit |
67cb25 |
computing the value of the incomplete beta function.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
* Was new in gsl-0.5:
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** There is now a KNOWN-PROBLEMS file which lists compilation problems
|
|
Packit |
67cb25 |
and test failures which are known to the developers.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Many improvements have been made to the special functions directory.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The extrapolations from the Levin u-transform are now more reliable.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Linear algebra and Eigensystem routines are now available.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** ODE solvers are now available.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Multidimensional root finding algorithms are available.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Minimization now keeps track of function values.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Matrices and vectors now use a BLAS compatible format, and have a
|
|
Packit |
67cb25 |
separate memory handling layer (gsl_block).
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Roots of general polynomials can now be found using gsl_poly_complex_solve
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** IEEE modes support on Sparclinux, Tru64, AIX and IRIX
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** We have added the second generation RANLUX generators RANLXS and RANLXD
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Minimization algorithms are available (one-dimensional)
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Documentation now works out of the box with the standard Texinfo.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Full reimplementation of the QUADPACK integration library
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Introduced THANKS file.
|
|
Packit |
67cb25 |
We appreciate all patches from people on the net, even those which are
|
|
Packit |
67cb25 |
too small to warrant adding the author to the AUTHORS file. The
|
|
Packit |
67cb25 |
THANKS file should include everyone who sent in patches. They should
|
|
Packit |
67cb25 |
also be mentioned in the ChangeLog entry.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
* What was new in gsl-0.4.1:
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Two changes not making their way into the documentation
|
|
Packit |
67cb25 |
A couple of things are not getting into the docs, so here are the
|
|
Packit |
67cb25 |
errata:
|
|
Packit |
67cb25 |
*** The FFT routines now take a stride parameter. Passing 1 for the
|
|
Packit |
67cb25 |
stride will make them behave as documented.
|
|
Packit |
67cb25 |
*** The complex numbers are now an opaque type, and no assumptions can
|
|
Packit |
67cb25 |
be made about the format in which they are stored (they are not stored
|
|
Packit |
67cb25 |
as a simple structure anymore, since that is not portable). The type
|
|
Packit |
67cb25 |
is now gsl_complex (or gsl_complex_long_double or gsl_complex_float),
|
|
Packit |
67cb25 |
and the macros to access them are
|
|
Packit |
67cb25 |
GSL_REAL(z)
|
|
Packit |
67cb25 |
GSL_IMAG(z)
|
|
Packit |
67cb25 |
GSL_COMPLEX_P_REAL(zp)
|
|
Packit |
67cb25 |
GSL_COMPLEX_P_IMAG(zp)
|
|
Packit |
67cb25 |
GSL_COMPLEX_EQ(z1,z2)
|
|
Packit |
67cb25 |
GSL_SET_COMPLEX(zp,x,y)
|
|
Packit |
67cb25 |
GSL_SET_REAL(zp,x)
|
|
Packit |
67cb25 |
GSL_SET_IMAG(zp,y)
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
This change in the complex number API makes it important that you
|
|
Packit |
67cb25 |
start working with 0.4.1 or later.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** 0.4.1 is being released in occasion of the Red Hat 6.0 release.
|
|
Packit |
67cb25 |
The specfunc module is still in an alpha state; if you run "make
|
|
Packit |
67cb25 |
check" in the specfunc directory you will see that some tests still
|
|
Packit |
67cb25 |
fail.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Most Alpha specific problems have been fixed. In particular the
|
|
Packit |
67cb25 |
random number generators rand48 and ranf now work on the Alpha
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Additional random number distributions:
|
|
Packit |
67cb25 |
Rayleigh distribution
|
|
Packit |
67cb25 |
n-dimensional spherical distribution
|
|
Packit |
67cb25 |
(ie, points at random on an n-dimensional sphere)
|
|
Packit |
67cb25 |
Gaussian tail distribution
|
|
Packit |
67cb25 |
(ie, choosing values from a gaussian distribution subject to a
|
|
Packit |
67cb25 |
constraint that they be larger than some fixed value, eg 5 sigmas)
|
|
Packit |
67cb25 |
Walker's algorithm for arbitrary discrete distributions
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
* What was new in gsl-0.4:
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** A single libgsl.a file is built in the top level directory and
|
|
Packit |
67cb25 |
installed, instead of separate .a files for each subdirectory.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The parts of the complex struct gsl_complex, .real and .imag, are
|
|
Packit |
67cb25 |
not supported anymore. The macros GSL_REAL(z) and GSL_IMAG(z) do the
|
|
Packit |
67cb25 |
same job. All complex numbers are considered as packed arrays of
|
|
Packit |
67cb25 |
floating point numbers, for portability since the layout of structs or
|
|
Packit |
67cb25 |
arrays of structs is not guaranteed.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The interface for matrices and vectors has changed. Vectors now
|
|
Packit |
67cb25 |
support strides, and can be used to access rows and columns of a
|
|
Packit |
67cb25 |
matrix. Many more types are available (float, double, long double,
|
|
Packit |
67cb25 |
int, long, short, char, signed and unsigned, plus complex floats,
|
|
Packit |
67cb25 |
doubles and long doubles) due to improvements in our preprocessor
|
|
Packit |
67cb25 |
template system.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The random number generators have a completely new thread-safe
|
|
Packit |
67cb25 |
interface and have moved from the random directory to the rng
|
|
Packit |
67cb25 |
directory. Any program using random numbers will have to be
|
|
Packit |
67cb25 |
updated. You can also choose generators and seeds using the
|
|
Packit |
67cb25 |
environment variables GSL_RNG_TYPE and GSL_RNG_SEED.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Some additional random number distributions have been added in the
|
|
Packit |
67cb25 |
randist directory. The available distributiosn are: bernoulli, beta,
|
|
Packit |
67cb25 |
binomial, cauchy, chisq, erlang, exponential, fdist, flat, gamma,
|
|
Packit |
67cb25 |
gauss, geometric, levy, logistic, lognormal, nbinomial, pareto,
|
|
Packit |
67cb25 |
poisson, sphere, tdist, twosidedexp, weibull.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The FFT interface has be extended to support strides, but the
|
|
Packit |
67cb25 |
implementation hasn't been finished for all the cases yet, The FFT
|
|
Packit |
67cb25 |
allocation functions now return a pointer to a newly allocated
|
|
Packit |
67cb25 |
wavetable struct, instead of taking the pointer to an existing struct
|
|
Packit |
67cb25 |
as an argument.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
e.g. status = gsl_fft_wavetable_alloc(n, w)
|
|
Packit |
67cb25 |
is now w = gsl_fft_wavetable_alloc(n) in accordance with usual practice
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The statistics directory now works with all the builtin
|
|
Packit |
67cb25 |
types. It has a new function for computing the lag1-autocorrelation and
|
|
Packit |
67cb25 |
an extra set of numerical accuracy tests from NIST as part of 'make
|
|
Packit |
67cb25 |
check'.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The simulated annealing routines no longer set the random number
|
|
Packit |
67cb25 |
seed with the time of day. You'll need to reseed the generator
|
|
Packit |
67cb25 |
yourself if you want subsequent runs to use different random numbers.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Work is in progress on a reimplementation of QUADPACK in the
|
|
Packit |
67cb25 |
`integration' subdirectory, but it is not finished yet.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Work is in progress on reimplementations of the VEGAS and
|
|
Packit |
67cb25 |
MISER Monte Carlo algorithms in the monte' subdirectory. They work
|
|
Packit |
67cb25 |
just fine, but the code is still evolving.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Work has started on a portable blas system in the `blas'
|
|
Packit |
67cb25 |
subdirectory.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** You can now set the IEEE arithmetic mode for your programs from the
|
|
Packit |
67cb25 |
environment variable GSL_IEEE_MODE by calling the function
|
|
Packit |
67cb25 |
gsl_ieee_env_setup(). Currently this only works with the Linux kernel,
|
|
Packit |
67cb25 |
HP-UX, SunOS4 and Solaris.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** There are some simple spline interpolation functions in the `interp'
|
|
Packit |
67cb25 |
subdir.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** The NEWS file now uses outline mode, like the Emacs NEWS file
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
* This covers changes made *after* the gsl-0.2 snapshot
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Added several new modules: histogram, integration, matrix, specfunc
|
|
Packit |
67cb25 |
and vectors.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Changed libgsl_statisctics.a to libgslstatistics.a and
|
|
Packit |
67cb25 |
libgsl_siman.a to libgslsiman.a, since most of the packages don't have
|
|
Packit |
67cb25 |
the underscore. Users will have to remove the old files, unless they
|
|
Packit |
67cb25 |
do a "make uninstall" on the previous release before they install this
|
|
Packit |
67cb25 |
new one.
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
** Changes to the random number suite
|
|
Packit |
67cb25 |
Incorporated the gauss, poisson and exponential distributions in
|
|
Packit |
67cb25 |
the standard libgslrandom.a
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
Local variables:
|
|
Packit |
67cb25 |
mode: outline
|
|
Packit |
67cb25 |
paragraph-separate: "[ ?]*$"
|
|
Packit |
67cb25 |
end:
|