mapml is a experimental maple interface to the libraries flint and pml, developed jointly with pml. See [Hyun, Neiger, Schost 2019].
Problem and bug reports may be sent to gilles.villard@ens-lyon.fr
Currently developed and tested with macOS >= 26.3.1 or linux, typically, and the compilers
# clang version 21.0.0 (clang-2100.1.1.101)
# Target: M2 Pro arm64-apple-darwin25.3.0
# gcc version 12.2.0 (Debian 12.2.0-14)
# Target: x86_64-linux-gnu
Should operate more generally.
(currently tested with:)
- GNU MP (6.3.0): http://gmplib.org
- For FLINT: MPFR (4.2.2): http://www.mpfr.org
- A recent FLINT development version is mandatory (>= Wed. 20 may 2026 12:51:14 CEST): https://github.com/flintlib. Currently, pml+mapml requires building FLINT with support for AVX2 or NEON instructions.
Download from pml GitHub repository.
Go to the pml*/flint-extras directory:
./bootstrap.sh
./configure
make
make install
If gmp or flint are installed under non-standard directories you can specify their locations by using configure with the options --with-gmp, --with-flint. By default, make install installs files under /usr/local, you can specify another installation directory by giving ./configure the option --prefix=dirname. Run ./configure --help for further details.
Another example:
./bootstrap.sh
./configure --with-gmp=/opt/homebrew --with-flint=/usr/local
make
sudo make install
make check
Works with Maple 2025 and 2026, should be checked for previous versions.
If maple is installed in a non standard directory (e.g. not in /Library/Frameworks using macos), for reasons related to maple libraries, the environment variables DYLD_LIBRARY_PATH or LD_LIBRARY_PATH must indicate the maple library directory. That is the directory where libmaplec.dylib or libmaplec.so can be found (kernelopts(bindir); command in maple). For example:
(csh and linux) setenv LD_LIBRARY_PATH /opt/maple/bin.X86_64_LINUX (do not forget ${LD_LIBRARY_PATH}: if already set)
(bash and macos) export DYLD_LIBRARY_PATH=/mymaplelib/bin.APPLE_UNIVERSAL_OSX (do not forget $DYLD_LIBRARY_PATH: if already set)
Then, do the same as for pml only, specifying at the same time the maple directories in order to find maple include directory (cat(kernelopts(mapledir),"/extern"); command in maple) and maple libraries (kernelopts(bindir); command in maple):
./bootstrap.sh
./configure --with-gmp=/opt/homebrew --with-flint=/usr/local \
--with-maplec=/Library/Frameworks/Maple.framework/Versions/2026/extern \
--with-maplelib=/Library/Frameworks/Maple.framework/Versions/2026/bin.APPLE_UNIVERSAL_OSX \
--prefix=some_directory_where_to_install
make
make install
Another example:
./bootstrap.sh
./configure --with-gmp=/opt/homebrew \
--with-maplec=/Library/Frameworks/Maple.framework/Versions/2026/extern \
--with-maplelib=/Library/Frameworks/Maple.framework/Versions/2026/bin.APPLE_ARM64_MACOS
make
sudo make install
Note that the compilation flag -Wstrict-prototypes in PML may produce four warnings related to the Maple library. They seem to be inconsequential.
mapml routines are then available by reading from maple the file mapml.mpl found in the pml include directory determined by pml installation (see --prefix). The default is:
read("/usr/local/include/pml/mapml.mpl");
For example, see the maple worksheet maple_check.mw from the pml examples directory. By specifying an appropriate directory for the mapml.mpl file, it should execute correctly.
Routines to manipulate matrices with univariate polynomial entries in Maple, in line with the LinearAlgebra package. Currently, these routines are global (not private to the mapml module).
MDeterminant(A, x) mod q;
WeakPopov(A, x, shift) mod q;
unimodular column transformation, shift is a list of integers, of length the row dimension of A
RowWeakPopov(A, x, shift) mod q;
unimodular row transformation, shift is a list of integers, of length the column dimension of A
PMBasis(A, x, shift, sigma) mod q;
approximant basis B and resulting defects such that A.B has order at lest sigma
shift is a list of integers, of length the column dimension of A
RowPMBasis(A, x, shift, sigma) mod q;
approximant basis B and resulting defects such that B.A has order at lest sigma
shift is a list of integers, of length the row dimension of A
MBasis(A, x, shift, sigma) mod q;
approximant basis B and resulting defects such that A.B has order at lest sigma
shift is a list of integers, of length the column dimension of A
RowMBasis(A, x, shift, sigma) mod q;
approximant basis B and resulting defects such that B.A has order at lest sigma
shift is a list of integers, of length the row dimension of A
RightDescription(M, x, d) mod q;
M is a matrix of rational fractions in x, with power series expansion at x=0
tries computes a matrix fraction description P,Q of degree at most d (reduced) such that M=PQ^(-1)
LeftDescription(M, x, d) mod q;
M is a matrix of rational fractions in x, with power series expansion at x=0
tries to compute a matrix fraction description Q,P of degree at most d (reduced) such that M=Q^(-1)P
Kernel(A, x, shift) mod q;
Normalized minimal kernel basis N such that A.N is zero
shift is a list of integers, of length the column dimension of A
RowKernel(A, x, shift) mod q;
Normalized minimal kernel basis N such that N.A is zero
shift is a list of integers, of length the row dimension of A
PMNullspace(A x) mod q;
Direct interface to flint nmod_poly_mat_nullspace
right (rational) nullspace polynomial matrix N such that A.N is zero