Numerical experiments of arXiv:1704.08072 *Perron-based algorithms for the multilinear pagerank*
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Failed to load latest commit information.
README.md
bertini_solve.m
bootstrap_derivativefree.m
bootstrap_first_derivative.m
bootstrap_second_derivative.m
bootstrap_second_derivative_plus_previous.m
bootstrap_variant.m
bootstrap_wrong_derivative.m
check_solution.m
commutation.m
experiment.m
fixed_point.m
fixed_point_from_zero.m
innout.m
load_tensor.m
newton.m
newton_from_zero.m
newton_on_y.m
optimistic.m
optimistic_fixed_point.m
optimistic_newton.m
optimistic_shifted.m
perfprof.m
perronvector.m
plot_minsol.m
shifted.m
solve_from_gleich.m
solve_symbolic.sage
tpr_method.m
try_all_methods.m
try_methods.m
vec.m

README.md

Contains the Matlab code used in the numerical experiments of https://arxiv.org/abs/1704.08072 (Meini, Poloni, * Perron-based algorithms for the multilinear pagerank*).

The main driver to run all experiments is try_methods. The single new methods are in bootstrap_* for the continuation algorithms, and optimistic and optimistic_newton for the Perron methods.

load_tensor loads tensors from a Gleich's mlpagerank repo, if it exists in a suitable subfolder, e.g., load_tensor('R6_3').

bertini_solve requires BertiniLab, and generates the last figure in the paper.