Self-contained code here:
--8<---------------cut here---------------start------------->8---
function aggregator_matrix{T <: Integer}(a::Vector{T})
n = length(a)
(min_, k) = extrema(a)
@assert 1 <= min_ "Aggregator index smaller than 1."
sparse(a, 1:n, 1.0, k, n)
end
A = aggregator_matrix([3, 3, 2, 1, 1])
p = [0.0, 0.138946, 0.0203463, 0.648933, 0.191775]
full(A * spdiagm(p))
full(A) * spdiagm(p)
--8<---------------cut here---------------end--------------->8---
bug occurs in debian's 0.3.5-1, not in Ubuntu's 0.3.5~utopic4, or the
nighly 0.4.0-1820~ubuntu14.10.1. So now I am using the ubuntu version.
If this is not known to be fixed, maybe worth adding to the test suite.
Best,
Tamas
On Wed, Feb 18 2015, Andreas Noack <[email protected]> wrote:
> I'm unable to reproduce this on 0.3.5, 0.3.7pre or master. How did you
> construct A?
>
> 2015-02-18 9:17 GMT-05:00 Tamas Papp <[email protected]>:
>
>> Is this a bug, or am I missing something obvious?
>>
>> julia> versioninfo()
>> Julia Version 0.3.5
>> Platform Info:
>> System: Linux (x86_64-linux-gnu)
>> CPU: Intel(R) Core(TM) i7-3520M CPU @ 2.90GHz
>> WORD_SIZE: 64
>> BLAS: libblas
>> LAPACK: liblapack
>> LIBM: libopenlibm
>> LLVM: libLLVM-3.5
>>
>> julia> A
>> 3x5 sparse matrix with 5 Float64 entries:
>> [3, 1] = 1.0
>> [3, 2] = 1.0
>> [2, 3] = 1.0
>> [1, 4] = 1.0
>> [1, 5] = 1.0
>>
>> julia> p
>> 5-element Array{Float64,1}:
>> 0.0
>> 0.138946
>> 0.0203463
>> 0.648933
>> 0.191775
>>
>> julia> full(A * spdiagm(p))
>> 3x5 Array{Float64,2}:
>> 0.0 0.0 0.0 0.648933 0.0
>> 0.0 0.0 0.0203463 0.0 0.191775
>> 0.0 0.138946 0.0 0.0 0.0
>>
>> julia> full(A) * spdiagm(p)
>> 3x5 Array{Float64,2}:
>> 0.0 0.0 0.0 0.648933 0.191775
>> 0.0 0.0 0.0203463 0.0 0.0
>> 0.0 0.138946 0.0 0.0 0.0
>>
>> Best,
>>
>> Tamas
>>
