| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
This release is exclusively in order to support Python 3.7.
|
|
Algorithm
Rewrite of the core int32/avx2 implementation for (1) higher speed and
(2) reduced memory consumption. Stack allocation is now at most a few
kilobytes, even for gigantic arrays.
Internally, the sorting algorithm is now mostly bitonic to simplify
indexing, although odd-even speedups are still applied when
convenient. Lanes are complemented to take the down-up decision out of
the inner loops.
As in previous djbsort versions, data is sorted first in vector lanes
and then transposed for final merges, reducing the overall number of
vector permutations. Unlike previous versions, transposition is done
in-place. The transposition in this version is bit-reversal on the outer
6 bits (bottom 3 bits and the top 3 bits), but leaves intermediate bits
alone. Non-power-of-2 array sizes are handled by an extra, more
traditional, merge step.
Sizes 2, 3, 4, 5, 6, 7, 8, 16, 32 are now special-cased. Non-power-of-2
sizes below 256 are padded to the next power of 2.
Portable implementations: The out-of-place int32/portable1 and
int32/portable2 implementations are now gone; the in-place
int32/portable3 and int32/portable4 implementations remain.
C API
float32_sort is now supported. The arithmetic in the reduction from
float32 to int32 is int32 31-bit right shift, uint32 1-bit right shift,
xor; this is slightly more efficient than the reduction from float32 to
uint32 from 2001 Herf.
Compiling
Tests now have more variation (without much slowdown): the uint32 test
cases now deviate from int32 in more than the sign; float32 uses
floating-point numbers that aren't integers; int32 does more loops for
small cases, and some larger cases.
Internals
API for 2-input sorting is now MINMAX macro operating on two
inputs in place.
Better inline assembly from Jason Donenfeld for 2-input sorting: more
flexibility in compiler's register allocation.
The package version number is now automatically copied to version.c as
the implementation version number for implementations that don't provide
version.c.
Verification
minmax now supports more peephole optimizations for complemented bitonic
sorting and for padding: xor(s,xor(s,t)) ⇒ t; xor(-1,s) ⇒ invert(s);
Reverse(Reverse(s)) ⇒ s; signedmin(invert(s),invert(t)) ⇒
invert(signedmax(s,t)); signedmax(invert(s),invert(t)) ⇒
invert(signedmin(s,t)); invert(s)[high:low] ⇒ invert(s[high:low]);
s[bits-1:0] ⇒ s; s[high:low][high2:low2] ⇒ s[high2+low:low2+low];
Concat(...)[high:low] ⇒ ...[high-pos:low-pos] when possible;
Reverse(s)[high:low] ⇒ Reverse(s[...]) when possible; eliminate
signedmin/signedmax when one input is the minimum or maximum constant.
verifymany now includes the implementation version number on
verified lines.
|
|
Install with pax, not cp. On deinstall, remove whatever's left in
share/djbsort.
|
|
machine, not on the machine preparing a binary package." (Also: "The
issues are explained in Section 8 of
https://pqcrypto.eu.org/deliverables/d2.4.pdf.")
For this to work, we install the source tree (with built objects) to
${PREFIX}/share/djbsort. Then we run tests, install to ${PREFIX}/include
and ${PREFIX}/lib, and check the installed files against pseudo-PLIST.
This means pkg_add(1) will fail if no C compiler is present, which is
unusual behavior for pkg_add but perhaps not entirely unreasonable for a
C library.
Bump PKGREVISION.
|
|
|
|
djbsort is a software library for sorting arrays of integers. It
provides the following features:
- Speed: djbsort holds new speed records for in-memory sorting.
- Security: djbsort is designed to be safe for cryptographic contexts.
- Verification: djbsort includes tools to automatically verify correctness.
These features are not separate options: there is a single sorting
function that is simultaneously fast, secure, and verified.
Packaged for wip by schmonz@
|
|
Distfiles for older R packges are moved on CRAN from contrib/ to
contrib/Archive/${R_PKGNAME}/. As a result, they may be inaccessible to
'make fetch'. Add the additional directory to MASTER_SITES in
math/R/Makefile.extension so that older distfiles will be retrieved
directly from CRAN. In addition, remove the definition of MASTER_SITES
from math/R-prob/Makefile, which is the only package directly referring to
the distfile archive. At this point all R packages should rely on
math/R/Makefile.extension to define MASTER_SITES; they should not be
defining MASTER_SITES directly.
|
|
Each R package should include ../../math/R/Makefile.extension, which also
defines MASTER_SITES. Consequently, it is redundant for the individual
packages to do the same. Package-specific definitions also prevent
redefining MASTER_SITES in a single common place.
|
|
|
|
The Stan Math Library is a C++, reverse-mode automatic differentiation
library designed to be usable, extensive and extensible, efficient,
scalable, stable, portable, and redistributable in order to facilitate
the construction and utilization of algorithms that utilize
derivatives.
|
|
New features:
- Add documentation section on survival analysis and censored data
models
- Add check_test_point method to pm.Model
- Add Ordered Transformation and OrderedLogistic distribution
- Add Chain transformation
- Improve error message Mass matrix contains zeros on the
diagonal. Some derivatives might always be zero during tuning of
pm.sample
- Improve error message NaN occurred in optimization. during ADVI
- Save and load traces without pickle using pm.save_trace and
pm.load_trace
- Add Kumaraswamy distribution
- Add TruncatedNormal distribution
- Rewrite parallel sampling of multiple chains on py3. This resolves
long standing issues when transferring large traces to the main
process, avoids pickling issues on UNIX, and allows us to show a
progress bar for all chains. If parallel sampling is interrupted, we
now return partial results.
- Add sample_prior_predictive which allows for efficient sampling from
the unconditioned model.
- SMC: remove experimental warning, allow sampling using sample,
reduce autocorrelation from final trace.
- Add model_to_graphviz (which uses the optional dependency graphviz)
to plot a directed graph of a PyMC3 model using plate notation.
- Add beta-ELBO variational inference as in beta-VAE model
(Christopher P. Burgess et al. NIPS, 2017)
- Add __dir__ to SingleGroupApproximation to improve autocompletion in
interactive environments
|
|
|
|
Highlights
There are many changes in 1.2 (see below). Some of the highlights are
* Python 3.3 is no longer supported. If you require Python 3.3
support, use SymPy 1.1.1. See our [518]policy on dropping support
for major Python versions.
* Experimental LaTeX parsing with sympy.parsing.latex.parse_latex()
has been added, based on the latex2sympy project. This requires
antlr-python-runtime to be installed. [519]#13706
* The vector module has been improved to support orthogonal
curvilinear coordinate systems ([520]Szymon Mieszczak's GSoC
project)
* New module sympy.integrals.intpoly for integrating uni/bivariate
polynomials over 2-polytopes. ([521]Arif Ahmed's GSoC project)
* Improvements to the code generation module. ([522]Bj?rn Dahlgren's
GSoC project)
* Improvements to the group theory module. See below for more
information. ([523]Valeriia Gladkova's GSoC project)
* New module sympy.discrete for operating on discrete sequences.
polynomials over 2-polytopes. ([521]Arif Ahmed's GSoC project)
* Improvements to the code generation module. ([522]Bj?rn Dahlgren's
GSoC project)
* Improvements to the group theory module. See below for more
information. ([523]Valeriia Gladkova's GSoC project)
* New module sympy.discrete for operating on discrete sequences.
([524]Sidhant Nagpal's GSoC project)
Complete release notes at:
https://github.com/sympy/sympy/wiki/release-notes-for-1.2
|
|
|
|
Changes from 2.6.5 to 2.6.6:
Fix to the thread barrier that occassionally suffered from spurious wakeups on MacOSX.
|
|
|
|
C++14 default language.
|
|
|
|
build fix for newer qt.
bump PKGREVISION.
|
|
0.9.11:
make exception explicit
0.9.10:
add AMPGO and basin-hopping global optimization methods.
aborting a fit from the objective function now raises AbortFitException
fit statistics are more uniformly calculated.
the uncertainties package is now an external dependency, and an out-dated copy is no longer kept in lmfit.
more exceptions when import matplotlib are now tolerated.
many documentation fixes.
|
|
0.23.3:
This is a minor bug-fix release in the 0.23.x series and includes a fix for the source distribution on Python 3.7. We recommend that all users upgrade to this version.
|
|
Upstream changes:
Version 0.6-2
o dbGetQuery do not return NULL.
o use registration mechanism to call native routines.
Version 0.5-3
o Update the libpq library so that the package can be compiled in new versions of R and macOS.
Version 0.5-2
o Support dbBegin() API
version 0.5-1 --2014-02-08
o Remove segmentationfault when NULL is passed as host argument.
o Conversion routine for raw in R and bytea in PostgreSQL
o Use accompanied libpq when either of libpq and header is not found on darwin
Version 0.5 -- 2013-10-12
o postgresqlWriteTable: send column names at COPY
o System library header search path for OS X 10.9 (Mavericks)
o Memory leak was reduced
o Close the socket on failure
Version 0.4 -- 2013-03-27
o Initial implementation of prepared statement
o Use system libpq library when available on OS X (darwin)
o Force ISO datestyle for PostgreSQL-R communication by default
o Time zone aware Timestamp POSIXct conversion
Version 0.3-3 -- 2012-10-05
o Bugfix on dbColumnInfo and others reproducible by gctorture(TRUE)
o Do not implicitly make new connections, which interfere with transaction.
o Change the notation of the LICENSE
|
|
|
|
SUNDIALS is a SUite of Nonlinear and DIfferential/ALgebraic equation
Solvers. It consists of the following six solvers: CVODE, solves
initial value problems for ordinary differential equation (ODE)
systems; CVODES, solves ODE systems and includes sensitivity analysis
capabilities (forward and adjoint); ARKODE, solves initial value ODE
problems with additive Runge-Kutta methods, include support for IMEX
methods; IDA, solves initial value problems for differential-algebraic
equation (DAE) systems; IDAS, solves DAE systems and includes
sensitivity analysis capabilities (forward and adjoint); KINSOL,
solves nonlinear algebraic systems.
|
|
|
|
|
|
PyMC3 is a Python package for Bayesian statistical modeling and
Probabilistic Machine Learning focusing on advanced Markov chain Monte
Carlo (MCMC) and variational inference (VI) algorithms. Its
flexibility and extensibility make it applicable to a large suite of
problems.
|
|
|
|
Theano is a Python library that allows you to define, optimize, and evaluate
mathematical expressions involving multi-dimensional arrays efficiently
Packaged by Kamel Ibn Aziz Derouiche for pkgsrc-wip and updated by me.
|
|
0.9.0:
The Highlights
--------------
statespace refactoring, Markov Switching Kim smoother
3 Google summer of code (GSOC) projects merged - distributed estimation - VECM and enhancements to VAR (including cointegration test) - new count models: GeneralizedPoisson, zero inflated models
Bayesian mixed GLM
Gaussian Imputation
new multivariate methods: factor analysis, MANOVA, repeated measures within ANOVA
GLM var_weights in addition to freq_weights
Holt-Winters and Exponential Smoothing
|
|
Without the link option, install_name_tool may cause an error.
|
|
|
|
Performing substitutions during post-patch breaks tools such as mkpatches,
making it very difficult to regenerate correct patches after making changes,
and often leading to substituted string replacements being committed.
|
|
Changes from 2.6.4 to 2.6.5
- The maximum thread count can now be set at import-time by setting the
environment variable 'NUMEXPR_MAX_THREADS'. The default number of
max threads was lowered from 4096 (which was deemed excessive) to 64.
- A number of imports were removed (pkg_resources) or made lazy (cpuinfo) in
order to speed load-times for downstream packages (such as pandas, sympy,
and tables). Import time has dropped from about 330 ms to 90 ms. Thanks to
Jason Sachs for pointing out the source of the slow-down.
- Thanks to Alvaro Lopez Ortega for updates to benchmarks to be compatible with
Python 3.
- Travis and AppVeyor now fail if the test module fails or errors.
- Thanks to Mahdi Ben Jelloul for a patch that removed a bug where constants
in where calls would raise a ValueError.
- Fixed a bug whereby all-constant power operations would lead to infinite
recursion.
|
|
Upstream changes:
CHANGES IN R 3.5.1
BUG FIXES
file("stdin") is no longer considered seekable.
dput() and dump() are no longer truncating when options(deparse.max.lines = *) is set.
Calls with an S3 class are no longer evaluated when printed, fixing part of PR#17398, thanks to a patch from Lionel Henry.
Allow file argument of Rscript to include space even when it is first on the command line.
callNextMethod() uses the generic from the environment of the calling method. Reported by Hervé Pagès with well documented examples.
Compressed file connections are marked as blocking.
optim(*, lower = c(-Inf, -Inf)) no longer warns (and switches the method), thanks to a suggestion by John Nash.
predict(fm, newdata) is now correct also for models where the formula has terms such as splines::ns(..) or stats::poly(..), fixing PR#17414, based on a patch from Duncan Murdoch.
simulate.lm(glm(*, gaussian(link = <non-default>))) has been corrected, fixing PR#17415 thanks to Alex Courtiol.
unlist(x) no longer fails in some cases of nested empty lists. Reported by Steven Nydick.
qr.coef(qr(<all 0, w/ colnames>)) now works. Reported by Kun Ren.
The radix sort is robust to vectors with >1 billion elements (but long vectors are still unsupported). Thanks to Matt Dowle for the fix.
Terminal connections (e.g., stdin) are no longer buffered. Fixes PR#17432.
deparse(x), dput(x) and dump() now respect c()'s argument names recursive and use.names, e.g., for x <- setNames(0, "recursive"), thanks to Suharto Anggono's PR#17427.
Unbuffered connections now work with encoding conversion. Reported by Stephen Berman.
‘.Renviron’ on Windows with Rgui is again by default searched for in user documents directory when invoked via the launcher icon. Reported by Jeroen Ooms.
printCoefmat() now also works with explicit right=TRUE.
print.noquote() now also works with explicit quote=FALSE.
The default method for pairs(.., horInd=*, verInd=*) now gets the correct order, thanks to reports by Chris Andrews and Gerrit Eichner. Additionally, when horInd or verInd contain only a subset of variables, all the axes are labeled correctly now.
agrep("..|..", .., fixed=FALSE) now matches when it should, thanks to a reminder by Andreas Kolter.
str(ch) now works for more invalid multibyte strings.
|
|
|
|
Changes from 3.4.3 to 3.4.4
Improvements
Environment variable to control the use of embedded libraries.
Include citation in repository. gh-690.
Bugs fixed
Fixed import error with numexpr 2.6.5.dev0 gh-685.
Fixed linter warnings.
Fixed for re.split() is version detection. gh-687.
Fixed test failures with Python 2.7 and NumPy 1.14.3 gh-688 & gh-689.
|
|
|
|
1.11 2018-06-07 14:28:04 -0400
- Move to Perl5-Alien org
|
|
Added math/p5-Math-Systems version 0.01 [mef 2018-06-20]
|
|
The following methods are available:
new
clone
size of matrix
concat
transpose
multiply
solve
multiply_scalar
add
substract
equal
slice
determinant
dot_product
absolute
normalizing
cross_product
print
|
|
Solves systems of equations using Krammer's rule. If you look at the
Solve method you'll understand Krammer's rule (I think?)
Basically a system of equations is more than one equation/variable your
solving. This module REQUIRES you have 1 equation per each coefficent.
Try not to use too many equations. 3 equations is probably the limit of
reason. But if you don't mind long compilation time, go for as many as
you want.
This module uses (and ovbiously requires you have) Math::Matrix to do
all the Matrix work. Please take a look at that module for details do
you can see exactly what is happening. There are no arguements in this
module except for the Matrix which will be brought to Math::Matrix.
Also, keep in mind, there are no error messenging. I will of course
change that someday, but for now I'm satisfied that I can solve systems
of equations. Don't allow error-prone things to be written in. But,
there are of course times when you won't be able to figure out the
stupid answer and this module will absolutly act like it did a great
job, and at the same time, fail. So, keep that in mind.
BETA means something. It's not good.
|
|
pandas 0.23.1
This is a minor release from 0.23.0 and includes a number of bug fixes and
performance improvements.
|
|
NumPy 1.14.5:
This is a bugfix release for bugs reported following the 1.14.4 release. The
most significant fixes are:
* fixes for compilation errors on alpine and NetBSD
|
|
|
|
go-units provides helpful unit multipliers and functions in Go. The
goal of this package is to have functionality similar to the time
package.
|
|
|
|
|
|
|
|
Upstream changes:
NumPy 1.14.4 Release Notes
This is a bugfix release for bugs reported following the 1.14.3 release. The most significant fixes are:
fixes for compiler instruction reordering that resulted in NaN's not being properly propagated in np.max and np.min,
fixes for bus faults on SPARC and older ARM due to incorrect alignment checks.
|
