Error using linear indexing

[ngphuoc]

julia> a = rand(3,4)
3×4 Array{Float64,2}:
 0.748611  0.985436  0.376666   0.912038
 0.262307  0.514791  0.860165   0.262203
 0.685377  0.273166  0.0209705  0.561497

julia> k = AFArray(a)
AFArray: 3×4 Array{Float64,2}:
 0.748611  0.985436  0.376666   0.912038
 0.262307  0.514791  0.860165   0.262203
 0.685377  0.273166  0.0209705  0.561497

julia> a[:]==k[:]
ERROR: AssertionError: length(idx1) == length(size(a)) - 1
Stacktrace:
 [1] getindex(::ArrayFire.AFArray{Float64,2}, ::Colon) at /home/phuoc/.julia/v0.6/ArrayFire/src/indexing.jl:73

[ngphuoc]

This one works

julia> k[:].=a[:]
12-element SubArray{Float64,1,ArrayFire.AFArray{Float64,1},Tuple{Base.Slice{Base.OneTo{Int64}}},false}:
 0.748611
 0.262307
 0.685377
 0.985436
 0.514791
 0.273166
 0.376666
 0.860165
 0.0209705
 0.912038
 0.262203
 0.561497

[ghost]

In general mixing Array with AFArray in the same expression is not supported.

[ngphuoc]

Error indexing by array:

julia> k.>0.5
AFArray: 3×4 Array{Bool,2}:
  true   true  false   true
 false  false   true  false
 false  false  false  false

julia> k[k.>0.5]
ERROR: AssertionError: length(idx1) == length(size(a)) - 1
Stacktrace:
 [1] getindex(::ArrayFire.AFArray{Float64,2}, ::ArrayFire.AFArray{Bool,2}) at /home/phuoc/.julia/v0.6/ArrayFire/src/indexing.jl:73

[ghost]

duplicate: indexing is broken - #181

[ngphuoc]

But the following gives error

julia> k[:]
ERROR: AssertionError: length(idx1) == length(size(a)) - 1
Stacktrace:
 [1] getindex(::ArrayFire.AFArray{Float64,2}, ::Colon) at /home/phuoc/.julia/v0.6/ArrayFire/src/indexing.jl:73

[ngphuoc]

It's strange, assignment works but print doesn't:

julia> k[find(k.>0.5)]
ERROR: AssertionError: length(idx1) == length(size(a)) - 1
Stacktrace:
 [1] getindex(::ArrayFire.AFArray{Float64,2}, ::ArrayFire.AFArray{Int64,2}) at /home/phuoc/.julia/v0.6/ArrayFire/src/indexing.jl:73

julia> k[find(k.>0.5)] = 0
0

[ngphuoc]

No, it works for 1D array in #181 but not 2D:

julia> k[find(k.>0.5)] = 0
0

julia> k
AFArray: 3×4 Array{Float64,2}:
 0.790436  0.522505  0.333179  0.632241
 0.751422  0.841604  0.715114  0.928544
 0.884082  0.39492   0.654672  0.21895

[ngphuoc]

Some more tests:

julia> k[]
ERROR: ArrayFire Error (202) : Invalid input argument
In function af_err af_index_gen(void**, af_array, dim_t, const af_index_t*)
In file src/api/c/index.cpp:160
Invalid argument at index 2
Expected: (ndims>0)

Stacktrace:
 [1] _error(::UInt32) at /home/phuoc/.julia/v0.6/ArrayFire/src/util.jl:85
 [2] index_gen(::ArrayFire.AFArray{Float64,2}, ::Int64, ::Ptr{ArrayFire.af_index_t}) at /home/phuoc/.julia/v0.6/ArrayFire/src/wrap.jl:721
 [3] getindex(::ArrayFire.AFArray{Float64,2}) at /home/phuoc/.julia/v0.6/ArrayFire/src/indexing.jl:110

julia> k[1,:].>0.5
AFArray: 1×4 Array{Bool,2}:
 true  true  false  true

julia> k[:,k[1,:].>0.5]
AFArray: 3×3 Array{Float64,2}:
 0.574376  0.984427  0.514344
 0.460361  0.514912  0.416142
 0.386582  0.525051  0.862911

julia> k[:,k[1,:].>0.5]=0
ERROR: ArrayFire Error (203) : Invalid input size
In function af_err af_assign_gen(void**, af_array, dim_t, const af_index_t*, af_array)
In file src/api/c/assign.cpp:311
Size mismatch between input and output

Stacktrace:
 [1] _error(::UInt32) at /home/phuoc/.julia/v0.6/ArrayFire/src/util.jl:85
 [2] assign_gen(::ArrayFire.AFArray{Float64,2}, ::Int64, ::Ptr{ArrayFire.af_index_t}, ::ArrayFire.AFArray{Float64,2}) at /home/phuoc/.julia/v0.6/ArrayFire/src/indexing.jl:51
 [3] setindex!(::ArrayFire.AFArray{Float64,2}, ::ArrayFire.AFArray{Float64,2}, ::Colon, ::ArrayFire.AFArray{Bool,2}) at /home/phuoc/.julia/v0.6/ArrayFire/src/indexing.jl:120
 [4] setindex!(::ArrayFire.AFArray{Float64,2}, ::Int64, ::Colon, ::ArrayFire.AFArray{Bool,2}) at /home/phuoc/.julia/v0.6/ArrayFire/src/indexing.jl:132

julia> k[:,k[1,:].>0.5].=0
ERROR: BoundsError: attempt to access 3×4 ArrayFire.AFArray{Float64,2} at index [Base.Slice(Base.OneTo(3)), Base.LogicalIndex(AFArray: Bool[true true false true])]
Stacktrace:
 [1] throw_boundserror(::ArrayFire.AFArray{Float64,2}, ::Tuple{Base.Slice{Base.OneTo{Int64}},Base.LogicalIndex{CartesianIndex{2},ArrayFire.AFArray{Bool,2}}}) at ./abstractarray.jl:433
 [2] checkbounds(::ArrayFire.AFArray{Float64,2}, ::Base.Slice{Base.OneTo{Int64}}, ::Base.LogicalIndex{CartesianIndex{2},ArrayFire.AFArray{Bool,2}}) at ./abstractarray.jl:362
 [3] view(::ArrayFire.AFArray{Float64,2}, ::Colon, ::Vararg{Any,N} where N) at ./subarray.jl:113
 [4] dotview(::ArrayFire.AFArray{Float64,2}, ::Colon, ::Vararg{Any,N} where N) at ./broadcast.jl:537

julia> pointer(k)
ERROR: conversion to pointer not defined for ArrayFire.AFArray{Float64,2}
Stacktrace:
 [1] pointer(::ArrayFire.AFArray{Float64,2}) at ./abstractarray.jl:863

[ghost]

fixed in 92de2d9

[hayatoikoma]

Is it really fixed? I just tried it on the master.
It worked on 1d array but didn't work for 2d array.

julia> a1d = AFArray(rand(5))
AFArray: 5-element Array{Float64,1}:
 0.983842
 0.398841
 0.217953
 0.770181
 0.478076

julia> a1d[find(a1d.>0.5)] = 1
1

julia> a1d
AFArray: 5-element Array{Float64,1}:
 1.0
 0.398841
 0.217953
 1.0
 0.478076

julia> a2d = AFArray(rand(5,5))
AFArray: 5×5 Array{Float64,2}:
 0.414244  0.703552   0.463749  0.223488  0.722257
 0.509738  0.216766   0.789492  0.684924  0.137261
 0.462514  0.0957764  0.663805  0.796688  0.887247
 0.753675  0.593526   0.169375  0.549912  0.120235
 0.537148  0.307667   0.125833  0.158792  0.00143258

julia> a2d[find(a2d.>0.5)] = 1
1

julia> a2d
AFArray: 5×5 Array{Float64,2}:
 0.414244  0.703552   0.463749  0.223488  0.722257
 0.509738  0.216766   0.789492  0.684924  0.137261
 0.462514  0.0957764  0.663805  0.796688  0.887247
 0.753675  0.593526   0.169375  0.549912  0.120235

[ghost]

Make sure you got the latest master from https://github.com/JuliaComputing/ArrayFire.jl , should work just fine even without find

julia> a2d = AFArray(rand(5,5))
AFArray: 5×5 Array{Float64,2}:
 0.786437  0.837583   0.845785   0.257565   0.417475 
 0.429835  0.912153   0.805885   0.0414585  0.0215296
 0.373297  0.322926   0.0918821  0.715103   0.956392 
 0.81288   0.0600418  0.312362   0.284699   0.703298 
 0.39498   0.542362   0.752585   0.038305   0.290987 

julia> a2d[find(a2d.>0.5)] = 1
1

julia> a2d
AFArray: 5×5 Array{Float64,2}:
 1.0       1.0        1.0        0.257565   0.417475 
 0.429835  1.0        1.0        0.0414585  0.0215296
 0.373297  0.322926   0.0918821  1.0        1.0      
 1.0       0.0600418  0.312362   0.284699   1.0      
 0.39498   1.0        1.0        0.038305   0.290987 

julia> a2d[a2d.>0.5] = 2
2

julia> a2d
AFArray: 5×5 Array{Float64,2}:
 2.0       2.0        2.0        0.257565   0.417475 
 0.429835  2.0        2.0        0.0414585  0.0215296
 0.373297  0.322926   0.0918821  2.0        2.0      
 2.0       0.0600418  0.312362   0.284699   2.0      
 0.39498   2.0        2.0        0.038305   0.290987 

[hayatoikoma]

Oh, sorry. I was accidentally using your fork on the machine I used today.
It worked fine. Thank you for checking it.