Menu

#131 Python interface for the conversion between numpy and tensor lib

0.2.3
done
nobody
None
nobody
2018-09-14
2018-08-17
Ulf Lorenz
No

Within ticket [#117], some conversion functions for the conversion between CTensor and numpy.array have been implemented. These were done in a rather simple manner, only for 2D tensors and such.

Within this ticket, the interface should be finalized. You should be able to transform arbitrary tensors between the C++ and numpy world.

  • Conversion function between CTensor and numpy.array
    • for arbitrary rank and dimensions
    • for several different data types (float, double, int, cdouble) on the numpy side
  • Conversion function between RTensor and numpy.array
    • for several different data types
    • for arbitrary rank and dimensions

Did not implement integer numpy arrays, since these did not work as easily as expected. Should not really matter, since floating-point arrays are what you usually want. Otherwise everything has tests and templated code assigned.

Related

Tickets: #117

Discussion

  • Ulf Lorenz

    Ulf Lorenz - 2018-08-18
    • Milestone: 0.3 goal --> 0.2.3
     
  • Ulf Lorenz

    Ulf Lorenz - 2018-09-06
    • status: open --> assigned
    • assigned_to: Ulf Lorenz
     
  • Ulf Lorenz

    Ulf Lorenz - 2018-09-14
    • Description has changed:

    Diff:

    --- old
    +++ new
    @@ -8,3 +8,7 @@
     * Conversion function between RTensor and numpy.array
         * for several different data types
         * for arbitrary rank and dimensions
    +
    +----
    +
    +Did not implement integer numpy arrays, since these did not work as easily as expected. Should not really matter, since floating-point arrays are what you usually want. Otherwise everything has tests and templated code assigned.
    
    • status: assigned --> done
    • assigned_to: Ulf Lorenz --> nobody
     

Log in to post a comment.

Want the latest updates on software, tech news, and AI?
Get latest updates about software, tech news, and AI from SourceForge directly in your inbox once a month.