nanpctile() gives an error: UndefVarError: T not defined for multidimension array

[Rapsodia86]

Hello!
I wanted to apply some statistic functions over a big array: 1675×1265×20 Array{Float32, 3} that has NaNs.
While it works well to get e.g. median across third dimension and reduce to matrix:
median_ar = nanmedian(rast_fin, dim=3)
it gives: 1675×1265 Matrix{Float32}

however, when try:
iqr_25_ar =nanpctile(rast_fin, 25, dim=3)
it gives error: UndefVarError: T not defined

What am I missing?
Thanks!

Stacktrace:
[1] _nanquantile!(A::SubArray{Float32, 1, Array{Float32, 3}, Tuple{Int64, Int64, Base.Slice{Base.OneTo{Int64}}}, true}, q::Float64, #unused#::Colon)
@ NaNStatistics C:\Users\monikat.julia\packages\NaNStatistics\ryatP\src\Sorting\nanpctile.jl:152
[2] macro expansion
@ C:\Users\monikat.julia\packages\NaNStatistics\ryatP\src\Sorting\nanpctile.jl:212 [inlined]
[3] _nanquantile!(B::Array{Float32, 3}, A::Array{Float32, 3}, q::Float64, dims::Tuple{Static.StaticInt{3}})
@ NaNStatistics C:\Users\monikat.julia\packages\NaNStatistics\ryatP\src\Sorting\nanpctile.jl:268
[4] macro expansion
@ C:\Users\monikat.julia\packages\NaNStatistics\ryatP\src\Sorting\nanpctile.jl:268 [inlined]
[5] _nanquantile!(B::Array{Float32, 3}, A::Array{Float32, 3}, q::Float64, dims::Tuple{Int64})
@ NaNStatistics C:\Users\monikat.julia\packages\NaNStatistics\ryatP\src\Sorting\nanpctile.jl:268
[6] _nanquantile!
@ C:\Users\monikat.julia\packages\NaNStatistics\ryatP\src\Sorting\nanpctile.jl:141 [inlined]
[7] _nanquantile!
@ C:\Users\monikat.julia\packages\NaNStatistics\ryatP\src\Sorting\nanpctile.jl:131 [inlined]
[8] __nanpctile!(A::Array{Float32, 3}, p::Int64, #unused#::Colon, region::Int64)
@ NaNStatistics C:\Users\monikat.julia\packages\NaNStatistics\ryatP\src\Sorting\nanpctile.jl:74
[9] __nanpctile(A::Array{Float32, 3}, p::Int64, dims::Function, dim::Int64)
@ NaNStatistics C:\Users\monikat.julia\packages\NaNStatistics\ryatP\src\Sorting\nanpctile.jl:17
[10] #nanpctile#49
@ C:\Users\monikat.julia\packages\NaNStatistics\ryatP\src\Sorting\nanpctile.jl:14 [inlined]

[brenhinkeller]

Hmm, good question. Which versions of Julia and NaNStatistics are you on? I can't seem to reproduce locally:

julia> a = rand(Float32, 1675, 1265, 20);

julia> using NaNStatistics

julia> nanpctile(a, 25, dim=3)
1675×1265 Matrix{Float32}:
 0.201905   0.213346  0.135244  0.312824   0.41867   …  0.0960994  0.154106   0.203496
 0.278074   0.236499  0.393272  0.21953    0.26126      0.174516   0.304279   0.307603
 0.239232   0.214348  0.262016  0.231451   0.276992     0.314799   0.370971   0.391675
 0.231702   0.49299   0.22068   0.20867    0.258393     0.199616   0.242694   0.131189
 0.0668727  0.336019  0.162326  0.218961   0.291756     0.249817   0.320886   0.234961
 0.158791   0.461463  0.284553  0.335379   0.287451  …  0.492462   0.317864   0.214808
 0.317202   0.251075  0.444729  0.355543   0.358705     0.195932   0.209493   0.306191
 0.381318   0.178761  0.212641  0.427774   0.141498     0.41383    0.139773   0.240451
 0.315014   0.173859  0.302293  0.23046    0.359963     0.359434   0.251495   0.229947
 0.216419   0.356514  0.440597  0.282261   0.172553     0.200759   0.179912   0.221218
 0.138713   0.388478  0.186638  0.233991   0.142028  …  0.433376   0.290844   0.216663
 0.158581   0.306969  0.276007  0.207912   0.314541     0.240828   0.278426   0.190032
 0.459776   0.188694  0.440165  0.181132   0.287091     0.168395   0.269304   0.375729
 0.310182   0.438171  0.391576  0.247264   0.271928     0.452179   0.161462   0.200436
 0.43586    0.154172  0.228099  0.201191   0.438236     0.286919   0.0969754  0.250468
 0.263719   0.26898   0.336066  0.326184   0.17316   …  0.363671   0.125518   0.323522
 0.33322    0.16957   0.107144  0.20838    0.285        0.286996   0.164111   0.284478
 0.257866   0.522421  0.306314  0.318473   0.284455     0.235368   0.248147   0.282748
 0.298578   0.32154   0.365487  0.189053   0.159999     0.265808   0.332364   0.365041
 0.271009   0.306294  0.171621  0.185273   0.377907     0.285846   0.0691201  0.292488
 0.277543   0.311748  0.191368  0.0777754  0.154933  …  0.193494   0.324027   0.271002
 0.229155   0.30209   0.184567  0.2792     0.301323     0.193195   0.256912   0.39149
 0.333308   0.24761   0.217494  0.406887   0.280803     0.183032   0.358676   0.207282
 0.357351   0.309931  0.191074  0.266505   0.199703     0.215182   0.345186   0.234963
 0.391713   0.283921  0.188846  0.344098   0.241541     0.268291   0.266011   0.197915
 0.314499   0.2841    0.30525   0.401528   0.152633  …  0.167575   0.377583   0.236827
 0.0793513  0.219738  0.139858  0.210716   0.14795      0.349948   0.312297   0.267818
 0.151663   0.155163  0.215905  0.190777   0.146133     0.286513   0.287612   0.250017
 0.176962   0.24979   0.359025  0.225649   0.318893     0.336431   0.339919   0.361817
 0.265544   0.318422  0.201094  0.228046   0.227932     0.28833    0.434047   0.319823
 0.283635   0.226708  0.219345  0.165486   0.274064  …  0.352266   0.19097    0.234822
 ⋮                                                   ⋱
 0.311681   0.323112  0.368197  0.267128   0.383434  …  0.143218   0.272182   0.186523
 0.251469   0.446104  0.151469  0.224057   0.436067     0.238204   0.266451   0.429897
 0.308806   0.301209  0.414878  0.337857   0.491233     0.231168   0.181843   0.356309
 0.237886   0.275072  0.261577  0.362267   0.303012     0.103261   0.311101   0.31108
 0.418175   0.383054  0.135382  0.30774    0.319802     0.211916   0.153432   0.315661
 0.394925   0.145993  0.240632  0.370278   0.372589  …  0.195052   0.404884   0.232299
 0.271849   0.281936  0.249793  0.299438   0.212459     0.124529   0.385071   0.297644
 0.154386   0.163948  0.279069  0.309223   0.346081     0.231901   0.395833   0.268941
 0.268111   0.497233  0.182525  0.335265   0.201526     0.233485   0.217399   0.389154
 0.373573   0.214063  0.277369  0.211133   0.515728     0.275293   0.335343   0.385843
 0.233291   0.30642   0.28293   0.177617   0.489046  …  0.224444   0.434184   0.275463
 0.173864   0.218865  0.109955  0.242567   0.152826     0.35337    0.217209   0.29524
 0.194855   0.493664  0.210693  0.155549   0.245459     0.219501   0.155598   0.178034
 0.133659   0.245331  0.323084  0.195487   0.333885     0.167532   0.208822   0.303439
 0.254429   0.202363  0.123782  0.371056   0.448275     0.316468   0.252061   0.372761
 0.343538   0.327914  0.288322  0.274093   0.298337  …  0.299169   0.080843   0.293657
 0.183983   0.452706  0.299203  0.376133   0.163436     0.124624   0.288508   0.412623
 0.191729   0.431828  0.310435  0.388841   0.276166     0.396557   0.361394   0.248067
 0.354493   0.168587  0.191503  0.276639   0.248208     0.119547   0.118179   0.222302
 0.390037   0.280628  0.388301  0.277035   0.297027     0.13783    0.177205   0.303176
 0.247677   0.147748  0.162335  0.280476   0.296622  …  0.36042    0.147374   0.416966
 0.220133   0.229865  0.270091  0.162126   0.316335     0.265098   0.464074   0.262431
 0.329481   0.342525  0.283391  0.28715    0.169664     0.374533   0.2507     0.283014
 0.476446   0.198537  0.255019  0.185916   0.198302     0.350275   0.149193   0.0631866
 0.195514   0.312179  0.316391  0.1724     0.37374      0.172825   0.205591   0.260561
 0.217447   0.260249  0.176607  0.101252   0.381988  …  0.289043   0.418625   0.281236
 0.113865   0.427509  0.197099  0.31183    0.297451     0.174565   0.359918   0.415651
 0.44377    0.204027  0.387042  0.282492   0.220362     0.300617   0.0903022  0.185137
 0.19224    0.415326  0.107594  0.41815    0.246445     0.415882   0.282922   0.285532
 0.211436   0.279145  0.263116  0.155819   0.265713     0.234683   0.169139   0.338886

[Rapsodia86]

Here is the info:
julia> VERSION
v"1.7.2"
julia> Pkg.status("NaNStatistics")
Status C:\Users\monikat\.julia\environments\v1.7\Project.toml
[b946abbf] NaNStatistics v0.6.7

Hm.. yeah, I see it works for you. I am using VSC julia extension.

[brenhinkeller]

Huh, odd. I don't have easy access to a Windows machine but I can try enabling CI on Windows and seeing if that catches anything.

[Rapsodia86]

One more, there are some cells that NaNs go across all arrays (20x NaN). But I would expect to get NaN for those cells as I get for median calculation right?

Huh, odd. I don't have easy access to a Windows machine but I can try enabling CI on Windows and seeing if that catches anything.

That would be great! Thank you for looking into it!

[brenhinkeller]

Ah, I think I found the problem

[brenhinkeller]

Ok, that should do it. Try updating to v0.6.9 once that gets registered and see if that works on your end!

[Rapsodia86]

Thank you! Will surely inform you if that works:)

[brenhinkeller]

JuliaRegistries/General#57788

[Rapsodia86]

Works! Perfect! Thank you very much!

[brenhinkeller]

Any time!