I think you should use qrfact(A,value{true}):
http://docs.julialang.org/en/release-0.4/stdlib/linalg/?highlight=qrfact#Base.qrfact
On Wed, Sep 7, 2016 at 3:37 AM, Stuart Brorson <s...@cloud9.net> wrote:
> Hello Julia users,
>
> Matlab has a variant of the QR decomposition invoked like this:
>
> [Q,R,P] = qr(A)
>
> This variant of qr() returns matrix R where the diagonal elements are
> sorted from largest to smallest magnitude as you go from upper left to
> lower right. The matrix P is the permutation matrix which permutes
> the rows/cols of A to give this ordering of the diagonal elements of
> R. That is, Q*R = A*P.
>
> I tried doing the naive, analogous thing in Julia, but get an error:
>
> julia> A = rand(3,3)
> 3x3 Array{Float64,2}:
> 0.243071 0.454947 0.89657
> 0.112843 0.802457 0.375417
> 0.154241 0.0182734 0.992542
>
> julia> Q,R,P = qr(A)
> ERROR: BoundsError: attempt to access (
> 3x3 Array{Float64,2}:
> -0.786117 0.0985642 -0.610168
> -0.364946 -0.870763 0.329523
> -0.498833 0.481723 0.720492,
>
> 3x3 Array{Float64,2}:
> -0.309204 -0.659611 -1.33693
> 0.0 -0.645106 0.2396
> 0.0 0.0 0.29177)
> at index [3]
> in indexed_next at tuple.jl:21
>
> My question: What's the best way to get the equivalent of Matlab's
> [Q,R,P] = qr(A) in Julia? Should I write my own qr() (not too
> difficult)? Or just do some row/col permutation of the output of
> regular qr()?
>
> Thanks for any advice,
>
> Stuart
>
> p.s. I am using Version 0.4.3-pre+6 (2015-12-11 00:38 UTC)
>
