You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Ran into a strange bug in Yota when I tried to broadcast a multiplication of a ComplexF64 to a Matrix of ComplexF64:
ERROR: UndefVarError: to_same_device not defined
Stacktrace:
[1] unbroadcast_prod_x(x::ComplexF64, y::Matrix{ComplexF64}, Δ::Matrix{ComplexF64})
@ Yota ~/.julia/packages/Yota/VCIzN/src/helpers.jl:78
I had a look in the source code and I can't see the function to_same_device anywhere inside the package.
P.S. I hit the error in the first place because I was lazy and didn't want to write a rrule for ChainRulesCore.rrule(::typeof(*), ::ComplexF64, ::Float64, ::Matrix{ComplexF64}) = ... and so I tried broadcasting the multiplication instead. I'm not sure if its a code path that anyone else would usually reach.
The text was updated successfully, but these errors were encountered:
Oh, I thought I got rid of them! to_same_device() used to move one array argument to the same device (e.g. CPU or GPU) as the other one. As a quick and dirty fix, you can try to redefine these methods without the conversion:
Ran into a strange bug in Yota when I tried to broadcast a multiplication of a ComplexF64 to a Matrix of ComplexF64:
I had a look in the source code and I can't see the function to_same_device anywhere inside the package.
P.S. I hit the error in the first place because I was lazy and didn't want to write a rrule for
ChainRulesCore.rrule(::typeof(*), ::ComplexF64, ::Float64, ::Matrix{ComplexF64}) = ...
and so I tried broadcasting the multiplication instead. I'm not sure if its a code path that anyone else would usually reach.The text was updated successfully, but these errors were encountered: