Polymake.jl 0.3.0 released

This is a major release streamlining and simplifying the architecture of the project. Among the largest changes are: the usage of polymake-4.0 and CxxWrap-0.9.0 which rises our requirement to c++-17 compatible compiler as well. More importantly we took the direct perspecive of integrating Polymake.jl into Oscar in this cycle.

New features

polymake now comes with flint and Singular dependencies and it is possible to perform some polynomial algebra in Polymake.jl. However we provide only the very basics, the more advanced task will be delegated to Oscar level.

Notable changes

Among many changes (see release notes for a detailed list), here are the most notable ones:

We hope the two decisions above will make developing in Oscar a much cleaner experience and hopefully we will see other packages following suit. Moreover,

To summarize: the only names exported from Polymake at the moment are: visual (for producing visualisation of polymake objects), application names and macros: @pm and @convert_to.

Constructors and macros

All BigObjects (in polymake nomenclature) gained corresponding structs in Polymake.jl. Of course none of it is written by hand, but auto-generated from polymake itself.

These constructors allow us to deprecate @pm macro in multiple cases. Just a quick reminder: the @pm macro can be used to issue more complicated calls to polymake from julia. If You need to pass templates to BigObjects, some limited support is provided in costructors. For example one can construct polytope.Polytope{Float64}(...). However for this to work, the templates need to be valid julia types/object, hence it is not possible to construct a Polytope<QuadraticExtension> through such call. For this (and in general: for passing more complicated templates) one still needs the @pm macro:

$obj = new BigObject<Template,Parameters>(args)


obj = @pm appname.BigObject{Template, Parameters}(args)

@convert_to macro is provided to facilitate conversion between different polymake types:

julia> c = polytope.cube(3);

julia> f = c.FACETS;

julia> f[1,1] # f is an opaque pm::perl::PropertyValue to julia
ERROR: MethodError: no method matching getindex(::Polymake.PropertyValueAllocated, ::Int64, ::Int64)

julia> m = @convert_to Matrix{Integer} f # the template must consist of C++ names
1 1 0 0
1 -1 0 0
1 0 1 0
1 0 -1 0
1 0 0 1
1 0 0 -1

Known Bugs

Edit this page Contact Imprint Privacy policy © 2018-2024 The OSCAR Team