rank is currently not implemented for Symmetric, Hermitian, and SymTridiagonal special matrix types

[jakebolewski]

svdvals is not defined for these special Matrix types.

julia> SymA
5x5 Array{Float64,2}:
  1.93666   -0.47591    1.27561     1.49607    0.278443
 -0.47591    0.777672   0.70466    -1.02571   -0.936692
  1.27561    0.70466    5.6342      0.137856  -0.0847055
  1.49607   -1.02571    0.137856    2.13749    1.86326
  0.278443  -0.936692  -0.0847055   1.86326    3.86176

julia> rank(Symmetric(SymA))
ERROR: MethodError: `svdvals` has no method matching svdvals(::Base.LinAlg.Symmetric{Float64,Array{Float64,2}})
 in rank at linalg/generic.jl:216

julia> rank(Hermitian(SymA))
ERROR: MethodError: `svdvals` has no method matching svdvals(::Base.LinAlg.Hermitian{Float64,Array{Float64,2}})
 in rank at linalg/generic.jl:216

julia> rank(SymTridiagonal(SymA))
ERROR: MethodError: `svdvals` has no method matching svdvals(::Base.LinAlg.SymTridiagonal{Float64})
 in rank at linalg/generic.jl:216

cc @andreasnoack

[jakebolewski]

norm is also currently broken for the same reason.

[stevengj]

I wouldn't say it is "broken", I would just say it is missing. "Broken", to me, implies something that is implemented but buggy. See also #136