Armadillo 3.820.1 - Changelog

What's new in Armadillo 3.810.1:

· Minor fix for linking with MKL on Ubuntu systems.

What's new in Armadillo 3.800.2:

· Minor fix for installation on Mac OS X systems.

What's new in Armadillo 3.6.0:

· Faster handling of compound expressions with submatrices and subcubes
· Added support for loading matrices as text files with NaN and Inf elements
· Added stable_sort_index(), which preserves the relative order of elements with equivalent values
· Added handling of sparse matrices by mean(), var(), norm(), abs(), square(), sqrt()

What's new in Armadillo 3.4.3:

· Fix for aliasing issue in diagmat()
· Fix for speye() signature
· Fixes for handling empty sparse matrices
· Minor fixes for handling sparse submatrix views
· Minor speedups for sparse matrices
· Workaround for a bug in the Mac OS X accelerate framework
· Added documentation for saving & loading matrices in HDF5 format

What's new in Armadillo 3.4.1:

· Fixes for handling empty sparse matrices.
· Added documentation for saving & loading matrices in HDF5 format.
· Faster dot() and cdot() for complex numbers.

What's new in Armadillo 3.4.0:

· Added .each_col() & .each_row() for vector operations repeated on each column or row
· Added preliminary support for sparse matrices
· Faster singular value decomposition via divide-and-conquer algorithm
· Faster .randn()

What's new in Armadillo 3.3.91:

· Faster singular value decomposition via "divide and conquer" algorithm.
· Added economical QR decomposition: qr_econ().
· Added .each_col() & .each_row() for vector operations repeated on each column or row.

What's new in Armadillo 3.2.3:

· Minor correction for declaration of fixed size vectors and matrices.

What's new in Armadillo 3.2.2:

· Minor fix for compiling without debugging enabled (aka release mode).
· Better detection of ATLAS during installation on Fedora and Red Hat systems.

What's new in Armadillo 3.2.1:

· Minor fix for compiling without debugging enabled (aka release mode).

What's new in Armadillo 3.0.0:

· Added non-contiguous submatrix views
· Added shorthand for inverse: .i()
· Added hist() and histc()
· Faster repmat()
· Faster handling of submatrix views with one row or column
· Faster generation of random numbers
· Faster element access in fixed size matrices
· Better detection of vector expressions by sum(), cumsum(), prod(), min(), max(), mean(), median(), stddev(), var().

What's new in Armadillo 2.99.2:

· Added .i().
· Much faster handling of .col() and .row().
· Expressions X=A.i()*B and X=inv(A)*B are automatically converted to X=solve(A,B).

What's new in Armadillo 2.4.4:

· Fixes for qr() and syl().
· More portable wall_clock class.
· Faster relational operators on submatrices.

What's new in Armadillo 2.4.2:

· Clarified documentation for .reshape().
· Fix for handling of empty matrices by .resize().

What's new in Armadillo 2.4.0:

· Added shorter forms of transposes: .t() and .st()
· Added optional use of 64 bit indices, allowing matrices to have more than 4 billion elements
· Added experimental support for C++11 initialiser lists
· Faster pinv()
· Faster inplace transpose
· Faster handling of expressions with diagonal views
· Fixes for handling expressions with aliasing and submatrices
· Fixes for linking on Ubuntu and Debian systems
· Fixes for inconsistencies in interactions between matrices and cubes
· Refactored code to eliminate warnings when using the Clang C++ compiler
· .print_trans() and .raw_print_trans() are deprecated

What's new in Armadillo 2.3.91:

· Added shorter forms of transposes: .t() and .st()
· Added optional use of 64 bit indices, allowing matrices to have more than 4 billion elements
· Added experimental support for C++11 initialiser lists
· Faster pinv()
· Faster inplace transpose
· Bug fixes for handling expressions with aliasing and submatrices
· Refactored code to eliminate warnings when using the Clang C++ compiler
· .print_trans() and .raw_print_trans() are deprecated

What's new in Armadillo 2.2.1:

· Faster multiplication of small matrices.
· Faster trans().
· Faster handling of submatrices by norm().
· Added economical singular value decomposition: svd_thin().
· Added circ_toeplitz().
· Added .is_colvec() & .is_rowvec().
· Fixes for handling of complex numbers by cov(), cor(), running_stat_vec.

What's new in Armadillo 2.1.91:

· Faster multiplication of small matrices.
· Faster trans().
· Faster handling of submatrices by norm().
· Added economical singular value decomposition: svd_thin().
· Added circ_toeplitz().
· Added .is_colvec() & .is_rowvec().

What's new in Armadillo 2.0.2:

· Fix for handling of conjugate transpose by as_scalar().
· Fix for handling of aliasing by diagmat().
· Fix for handling of empty matrices by symmatu()/symmatl().

What's new in Armadillo 2.0.0:

· Faster compound expressions containing submatrices
· Faster inverse of symmetric positive definite matrices
· Faster element access for fixed size matrices
· Added handling of arbitrarily sized empty matrices (eg. 5x0)
· Added loading & saving of matrices as CSV text files
· Added .count() member function to running_stat and running_stat_vec
· Added syl(), strans(), symmatu()/symmatl()
· Added submatrices of submatrices
· det(), inv() and solve() can be forced to use more precise algorithms for tiny matrices (≤ 4x4)
· htrans() has been deprecated; use trans() instead
· API change: trans() now takes the complex conjugate when transposing a complex matrix
· API change: .is_vec() now outputs true for empty vectors (eg. 0x1)
· API change: forms of chol(), eig_sym(), eig_gen(), inv(), lu(), pinv(), princomp(), qr(), solve(), svd(), syl() that do not return a bool indicating success now throw std::runtime_error exceptions when failures are detected
· API change: princomp_cov() has been removed; princomp() in conjunction with cov() can be used instead
· API change: set_log_stream() & get_log_stream() have been replaced by set_stream_err1() & get_stream_err1()

What's new in Armadillo 1.99.5:

· Forms of chol(), eig_sym(), eig_gen(), inv(), lu(), pinv(), princomp(), qr(), solve(), svd(), syl() that do not return a bool indicating success now throw std::runtime_error exceptions when failures are detected.
· princomp_cov() has been removed; princomp() in conjunction with cov() can be used instead.
· set_log_stream() & get_log_stream() have been replaced by set_stream_err1() & get_stream_err1().
· det(), inv() and solve() can be forced to use more precise algorithms for tiny matrices (≤ 4x4).
· Added loading & saving of matrices as CSV text files.

What's new in Armadillo 1.99.4:

· Fixes for handling of tiny matrices.

What's new in Armadillo 1.99.2:

· Faster inverse of symmetric matrices.
· Faster element access for fixed size matrices.
· Faster multiplication of tiny matrices (eg. 4x4).
· Faster compund expressions containing submatrices.
· Added handling of arbitrarily sized empty matrices (eg. 5x0).
· Added syl(), strans(), symmatu()/symmatl().
· Added submatrices of submatrices.
· htrans() has been deprecated; use trans() instead.
· trans() now takes the complex conjugate when transposing a complex matrix.
· .is_vec() now outputs true for empty matrices.
· Most functions with matrix inputs no longer throw exceptions for empty matrices.

What's new in Armadillo 1.2.0:

· Added ability to use Blas & Lapack libraries with capitalized function names.
· Reduction of pedantic compiler warnings.

What's new in Armadillo 1.1.92:

· Bugfix in cor().
· Automatic installation now requires CMake >= 2.6.

What's new in Armadillo 1.1.90:

· Added .min() & .max(), which can provide the extremum's location.
· More robust mean(), var(), stddev().