Masahiro TANAKA

http://rubyforge.org/tracker/?func=detail&atid=4625&aid=29408&group_id=1177

making v0.9

Method name corresponding to numpy.zeros_like

4

Numo::NArray needs to have a method corresponding to: ``` numpy.zeros_like(a) ``` In Ruby, the standard way is ``` a.zeros_like ``` This is wrong order if we regard this as an...

Currently `a == b` is implemented as `(a.eq b).count_false == 0`, but this is not efficient.

Define ! operator as an alias of Numo::NArray#inplace

5

``` a.inplace + 1 ``` is equivalent to ``` !a + 1 ``` Any problem? * In Ruby, ! is used to be added to the name of some destructive...

How to implement?

CUDA support

1

It is inevitable to use GPU for applications like deep neural network. It will be nice if there were a CUDA interface like CuPy: https://cupy.chainer.org/

Method? - rb_thread_call_without_gvl() ? Issues? - block access from another thread during operation.

Convention for abbreviation

3

``` a = Numo::DFloat.new(20,30).seq ``` This style is right as a Ruby code, but I prefer less typing. First idea: ``` include Numo a = DFloat.new(20,30).seq ``` In this case,...

The name of a method corresponding to numpy.maximum

6

Numpy has confusing two functions: numpy.max and numpy.maximum. Numo::NArray#max() method is same as numpy.max. Now I need to define the specification of a method corresponding to [numpy.maximum](https://docs.scipy.org/doc/numpy/reference/generated/numpy.maximum.html). Current ideas: ```...