Changes between Initial Version and Version 6 of Ticket #28873


Ignore:
Timestamp:
Dec 26, 2019, 10:09:49 PM (3 years ago)
Author:
gh-kliem
Comment:

Legend:

Unmodified
Added
Removed
Modified
  • Ticket #28873

    • Property Status changed from new to needs_review
    • Property Commit changed from to add61a5dcf0194da376d34e2c3959c22d2b48bb4
    • Property Branch changed from to public/28873-reb
  • Ticket #28873 – Description

    initial v6  
    4444}}}
    4545
    46 Follow up:
     46This also speeds up calculation in case the polyhedron is not full-dimensional:
    4747
    48 Once `affine_hull` completely respects the backend, this method will be used to calculate the induced volume by default (automatically).
     48{{{
     49sage: P = polytopes.permutahedron(6, backend='normaliz')
     50sage: %time P.volume(measure='induced')
     51CPU times: user 37.7 s, sys: 42.4 ms, total: 37.7 s
     52Wall time: 24.2 s
     531296*sqrt(6)
     54sage: %time P.volume(engine='internal', measure='induced')
     55CPU times: user 42.9 s, sys: 71.9 ms, total: 43 s
     56Wall time: 41.8 s
     571296*sqrt(6)
     58}}}
     59
     60However, note that calculation of the affine hull takes quite some time in this case. So we leave the inexact normaliz option for now:
     61{{{
     62sage: %time P.volume(engine='normaliz', measure='induced')
     63CPU times: user 710 ms, sys: 0 ns, total: 710 ms
     64Wall time: 92.3 ms
     653174.5387066469984
     66}}}
     67
     68Possible follow ups:
     69
     70There is probably a good and fast way to convert the lattice volume to the euclidean volume exactly. Also, once we can set up a normaliz polyhedron with both Vrep and Hrep, affine hull should be really quick.