Data structures for threaded computing

General Usage Performance
21

Thanks, it fixed the speed issue right up:

@time threadedGram2(x);
#  0.728811 seconds (117 allocations: 947.766 KiB)
22

Slightly off topic but I think it would probably be useful to have this in Base:

import Base.zeros
"""
  x = rand(Float64, (4,4))
  zeros(x)
"""
function zeros(x::AbstractArray)
  T = eltype(x)
  dim = size(x)
  return zeros(T, dim)
end

import Base.ones
"""
  x = rand(Float64, (4,4))
  ones(x)
"""
function ones(x::AbstractArray)
  T = eltype(x)
  dim = size(x)
  return ones(T, dim)
end
23

you can simply do:

zero(x)

" Get the additive identity element for the type of `x`"

although, one(x) would give you identity matrix, not sure if you want that or not (because it’s the multiplicative identity, if not, you can always fill(one(eltype(x)), size(x)) or what you’re already doing.

1 Like
24

Thanks. I didn’t know about zero.