On 11/4/20 5:47 PM, Charles R Harris wrote:
> Reminds me of the fast Hartley transform that was patented by Stanford in
> 1987. the patent was dropped in 1995.
There are cases where a algorithm is patented, but it is still possible
to have open-source implementations of it. One example is here:
https://www.aanda.org/articles/aa/full_html/2018/08/aa32858-18/aa32858-18.html
(ctrl-f "patent").
So maybe it's too early to give up hope.
Sure, it is legal to have open source implementations of patented algorithms. You are just pushing off the responsibility to acquire a patent license onto the user of your software. The copyright license to _redistribute_ your implementation is technically orthogonal to the patent on the _use_ of the algorithm itself. Note that in the case you point to, it is the patent holder that released the open source implementation, which would not be the case here.
As a matter of policy, we will not include implementations of patented algorithms in scipy because it makes the licensing story significantly more complicated. We want the rights to use the algorithms in scipy to be as free as the rights to redistribute. The proper place for code that is subject to patents is in a separate project.