ERROR: MethodError: no method matching cholesky

[PharmCat]

Hello! Is it expected?

x = rand(4,4)
pdm = PDMat(x'*x)
isposdef(pdm)

Trace:

ERROR: MethodError: no method matching cholesky(::Hermitian{Float64, PDMat{Float64, Matrix{Float64}}}; check=false)
Closest candidates are:
  cholesky(::Number) at C:\buildbot\worker\package_win64\build\usr\share\julia\stdlib\v1.6\LinearAlgebra\src\cholesky.jl:406 got unsupported keyword argument "check"
  cholesky(::Number, ::Symbol) at C:\buildbot\worker\package_win64\build\usr\share\julia\stdlib\v1.6\LinearAlgebra\src\cholesky.jl:406 got unsupported keyword argument "check"
  cholesky(::Diagonal) at C:\buildbot\worker\package_win64\build\usr\share\julia\stdlib\v1.6\LinearAlgebra\src\diagonal.jl:725 got unsupported keyword argument "check"
  ...
Stacktrace:
 [1] isposdef(A::PDMat{Float64, Matrix{Float64}})
   @ LinearAlgebra C:\buildbot\worker\package_win64\build\usr\share\julia\stdlib\v1.6\LinearAlgebra\src\dense.jl:91
 [2] top-level scope
   @ none:1

[devmotion]

Your stacktrace indicates that you don't use the latest version of PDMats. PDMats 0.11.3 defines isposdef and hence you should not end up with the fallback definition in LinearAlgebra.

[devmotion]

I checked and I can reproduce the issue with PDMats 0.11.2. However, as expected, it is fixed in PDMats 0.11.3:

(@v1.6) pkg> activate --temp
  Activating new environment at `/tmp/jl_AJ3u6A/Project.toml`

(jl_AJ3u6A) pkg> add PDMats
    Updating registry at `~/.julia/registries/General`
   Resolving package versions...
    Updating `/tmp/jl_AJ3u6A/Project.toml`
  [90014a1f] + PDMats v0.11.3
    Updating `/tmp/jl_AJ3u6A/Manifest.toml`
  [90014a1f] + PDMats v0.11.3
  [8f399da3] + Libdl
  [37e2e46d] + LinearAlgebra
  [9a3f8284] + Random
  [9e88b42a] + Serialization
  [2f01184e] + SparseArrays
  [4607b0f0] + SuiteSparse

julia> using PDMats, LinearAlgebra

julia> x = rand(4,4)
4×4 Matrix{Float64}:
 0.78919   0.171766  0.839576   0.446731
 0.416406  0.96419   0.0670805  0.788688
 0.759789  0.67533   0.945455   0.73147
 0.923695  0.436137  0.516336   0.488594

julia> pdm = PDMat(x'*x)
4×4 PDMat{Float64, Matrix{Float64}}:
 2.22671  1.45302  1.8858   1.68805
 1.45302  1.60545  1.07258  1.54426
 1.8858   1.07258  1.86988  1.37182
 1.68805  1.54426  1.37182  1.59537

julia> isposdef(pdm)
true

Since it's fixed in the latest release I'll close the issue 🙂

[AnzhiZhang]

function cholesky is missing a fallback method accepts AbstractMatrix in Julia 1.6, since 1.6 is a LTS version, I reported an issue at JuliaLang/julia#55404