Deprecate sumabs, sumabs2, minabs, maxabs

Author: ararslan

NEWS.md

@@ -98,6 +98,10 @@ Deprecated or removed
 
   * `cummin` and `cummax` have been deprecated in favor of `accumulate`.
 
+  * `sumabs` and `sumabs2` have been deprecated in favor of `sum(abs, x)` and `sum(abs2, x)`, respectively.
+    `maxabs` and `minabs` have similarly been deprecated in favor of `maximum(abs, x)` and `minimum(abs, x)`.
+    Likewise for the in-place counterparts of these functions ([#19598]).
+
 Julia v0.5.0 Release Notes
 ==========================
 

base/deprecated.jl

@@ -1134,4 +1134,22 @@ end)
 @deprecate cummin(A, dim=1) accumulate(min, A, dim=1)
 @deprecate cummax(A, dim=1) accumulate(max, A, dim=1)
 
+# #19598
+@deprecate sumabs(x)                    sum(abs, x)
+@deprecate sumabs(A, region)            sum(abs, A, region)
+@deprecate sumabs2(x)                   sum(abs2, x)
+@deprecate sumabs2(A, region)           sum(abs2, A, region)
+@deprecate minabs(x)                    minimum(abs, x)
+@deprecate minabs(A, region)            minimum(abs, A, region)
+@deprecate maxabs(x)                    maximum(abs, x)
+@deprecate maxabs(A, region)            maximum(abs, A, region)
+@deprecate sumabs!(r, A)                sum!(abs, r, A)
+@deprecate sumabs!(r, A; init=false)    sum!(abs, r, A; init=false)
+@deprecate sumabs2!(r, A)               sum!(abs2, r, A)
+@deprecate sumabs2!(r, A; init=false)   sum!(abs2, r, A; init=false)
+@deprecate minabs!(r, A)                minimum!(abs, r, A)
+@deprecate minabs!(r, A; init=false)    minimum!(abs, r, A; init=false)
+@deprecate maxabs!(r, A)                maximum!(abs, r, A)
+@deprecate maxabs!(r, A; init=false)    maximum!(abs, r, A; init=false)
+
 # End deprecations scheduled for 0.6

base/docs/helpdb/Base.jl

@@ -112,13 +112,6 @@ Cumulative minimum along a dimension. The dimension defaults to 1.
 """
 cummin
 
-"""
-    minabs!(r, A)
-
-Compute the minimum absolute values over the singleton dimensions of `r`, and write values to `r`.
-"""
-minabs!
-
 """
     eigfact!(A, [B])
 
@@ -530,13 +523,6 @@ Convert `y` to the type of `x` (`convert(typeof(x), y)`).
 """
 oftype
 
-"""
-    maxabs!(r, A)
-
-Compute the maximum absolute values over the singleton dimensions of `r`, and write values to `r`.
-"""
-maxabs!
-
 """
     isfinite(f) -> Bool
 
@@ -727,13 +713,6 @@ Show every part of the representation of a value.
 """
 dump
 
-"""
-    sumabs(A, dims)
-
-Sum absolute values of elements of an array over the given dimensions.
-"""
-sumabs(A, dims)
-
 """
     consume(task, values...)
 
@@ -1308,13 +1287,6 @@ Compute the inverse error complementary function of a real `x`, defined by
 """
 erfcinv
 
-"""
-    minabs(A, dims)
-
-Compute the minimum absolute values over given dimensions.
-"""
-minabs(A, dims)
-
 """
     popdisplay()
     popdisplay(d::Display)
@@ -1531,14 +1503,6 @@ Byte-swap an integer.
 """
 bswap
 
-"""
-    sumabs2!(r, A)
-
-Sum squared absolute values of elements of `A` over the singleton dimensions of `r`, and
-write results to `r`.
-"""
-sumabs2!
-
 """
     tanh(x)
 
@@ -1850,14 +1814,6 @@ false
 """
 isempty
 
-"""
-    sumabs!(r, A)
-
-Sum absolute values of elements of `A` over the singleton dimensions of `r`, and write
-results to `r`.
-"""
-sumabs!
-
 """
     hex2num(str)
 
@@ -2589,13 +2545,6 @@ it for new types as appropriate.
 """
 promote_rule
 
-"""
-    sumabs2(A, dims)
-
-Sum squared absolute values of elements of an array over the given dimensions.
-"""
-sumabs2(A,dims)
-
 """
     showall(x)
 
@@ -2656,13 +2605,6 @@ Compute cosine of `x`, where `x` is in radians.
 """
 cos
 
-"""
-    maxabs(A, dims)
-
-Compute the maximum absolute values over given dimensions.
-"""
-maxabs(A,dims)
-
 """
     done(iter, state) -> Bool
 

base/exports.jl

@@ -541,13 +541,9 @@ export
     logspace,
     mapslices,
     max,
-    maxabs,
-    maxabs!,
     maximum!,
     maximum,
     min,
-    minabs,
-    minabs!,
     minimum!,
     minimum,
     minmax,
@@ -601,10 +597,6 @@ export
     sub2ind,
     sum!,
     sum,
-    sumabs!,
-    sumabs,
-    sumabs2!,
-    sumabs2,
     sum_kbn,
     vcat,
     vec,

base/reduce.jl

@@ -350,20 +350,6 @@ Returns the sum of all elements in a collection.
 sum(a) = mapreduce(identity, +, a)
 sum(a::AbstractArray{Bool}) = countnz(a)
 
-"""
-    sumabs(itr)
-
-Sum absolute values of all elements in a collection. This is equivalent to `sum(abs(itr))` but faster.
-"""
-sumabs(a) = mapreduce(abs, +, a)
-
-"""
-    sumabs2(itr)
-
-Sum squared absolute values of all elements in a collection.
-This is equivalent to `sum(abs2(itr))` but faster.
-"""
-sumabs2(a) = mapreduce(abs2, +, a)
 
 # Kahan (compensated) summation: O(1) error growth, at the expense
 # of a considerable increase in computational expense.
@@ -463,30 +449,6 @@ julia> minimum([1,2,3])
 """
 minimum(a) = mapreduce(identity, scalarmin, a)
 
-"""
-    maxabs(itr)
-
-Compute the maximum absolute value of a collection of values.
-
-```jldoctest
-julia> maxabs([-1, 3, 4*im])
-4.0
-```
-"""
-maxabs(a) = mapreduce(abs, scalarmax, a)
-
-"""
-    minabs(itr)
-
-Compute the minimum absolute value of a collection of values.
-
-```jldoctest
-julia> minabs([-1, 3, 4*im])
-1.0
-```
-"""
-minabs(a) = mapreduce(abs, scalarmin, a)
-
 ## extrema
 
 extrema(r::Range) = (minimum(r), maximum(r))

base/reducedim.jl

@@ -294,19 +294,6 @@ for (fname, op) in [(:sum, :+), (:prod, :*),
     end
 end
 
-for (fname, fbase, fun) in [(:sumabs, :sum, :abs),
-                            (:sumabs2, :sum, :abs2),
-                            (:maxabs, :maximum, :abs),
-                            (:minabs, :minimum, :abs)]
-    fname! = Symbol(fname, '!')
-    fbase! = Symbol(fbase, '!')
-    @eval begin
-        $(fname!)(r::AbstractArray, A::AbstractArray; init::Bool=true) =
-            $(fbase!)($(fun), r, A; init=init)
-        $(fname)(A::AbstractArray, region) = $(fbase)($(fun), A, region)
-    end
-end
-
 
 ##### findmin & findmax #####
 

base/sparse/sparse.jl

@@ -16,8 +16,8 @@ import Base: @get!, acos, acosd, acot, acotd, acsch, asech, asin, asind, asinh,
     atan, atand, atanh, broadcast!, chol, conj!, cos, cosc, cosd, cosh, cospi, cot,
     cotd, coth, countnz, csc, cscd, csch, ctranspose!, diag, diff, done, dot, eig,
     exp10, exp2, eye, findn, floor, hash, indmin, inv, issymmetric, istril, istriu,
-    log10, log2, lu, maxabs, minabs, next, sec, secd, sech, show, sin,
-    sinc, sind, sinh, sinpi, squeeze, start, sum, sumabs, sumabs2, summary, tan,
+    log10, log2, lu, next, sec, secd, sech, show, sin,
+    sinc, sind, sinh, sinpi, squeeze, start, sum, summary, tan,
     tand, tanh, trace, transpose!, tril!, triu!, trunc, vecnorm, abs, abs2,
     broadcast, ceil, complex, cond, conj, convert, copy, copy!, ctranspose, diagm,
     exp, expm1, factorize, find, findmax, findmin, findnz, float, full, getindex,

base/sparse/sparsevector.jl

@@ -1284,8 +1284,6 @@ end
 ### Reduction
 
 sum(x::AbstractSparseVector) = sum(nonzeros(x))
-sumabs(x::AbstractSparseVector) = sumabs(nonzeros(x))
-sumabs2(x::AbstractSparseVector) = sumabs2(nonzeros(x))
 
 function maximum{T<:Real}(x::AbstractSparseVector{T})
     n = length(x)
@@ -1305,8 +1303,14 @@ function minimum{T<:Real}(x::AbstractSparseVector{T})
      min(zero(T), minimum(nonzeros(x))))::T
 end
 
-maxabs{T<:Number}(x::AbstractSparseVector{T}) = maxabs(nonzeros(x))
-minabs{T<:Number}(x::AbstractSparseVector{T}) = nnz(x) < length(x) ? abs(zero(T)) : minabs(nonzeros(x))
+for f in [:sum, :maximum, :minimum], op in [:abs, :abs2]
+    SV = :AbstractSparseVector
+    if f == :minimum
+        @eval ($f){T<:Number}(::typeof($op), x::$SV{T}) = nnz(x) < length(x) ? ($op)(zero(T)) : ($f)($op, nonzeros(x))
+    else
+        @eval ($f)(::typeof($op), x::$SV) = ($f)($op, nonzeros(x))
+    end
+end
 
 vecnorm(x::AbstractSparseVector, p::Real=2) = vecnorm(nonzeros(x), p)
 
@@ -1421,7 +1425,7 @@ function _spdot(f::Function,
 end
 
 function dot{Tx<:Number,Ty<:Number}(x::AbstractSparseVector{Tx}, y::AbstractSparseVector{Ty})
-    x === y && return sumabs2(x)
+    x === y && return sum(abs2, x)
     n = length(x)
     length(y) == n || throw(DimensionMismatch())
 

base/statistics.jl

@@ -183,7 +183,7 @@ var{T}(A::AbstractArray{T}; corrected::Bool=true, mean=nothing) =
 Compute the sample variance of a vector or array `v`, optionally along dimensions in
 `region`. The algorithm will return an estimator of the generative distribution's variance
 under the assumption that each entry of `v` is an IID drawn from that generative
-distribution. This computation is equivalent to calculating `sumabs2(v - mean(v)) /
+distribution. This computation is equivalent to calculating `sum(abs2, v - mean(v)) /
 (length(v) - 1)`. If `corrected` is `true`, then the sum is scaled with `n-1`,
 whereas the sum is scaled with `n` if `corrected` is `false` where `n = length(x)`.
 The mean `mean` over the region may be provided.
@@ -296,7 +296,7 @@ _vmean(x::AbstractMatrix, vardim::Int) = mean(x, vardim)
 
 # core functions
 
-unscaled_covzm(x::AbstractVector) = sumabs2(x)
+unscaled_covzm(x::AbstractVector) = sum(abs2, x)
 unscaled_covzm(x::AbstractMatrix, vardim::Int) = (vardim == 1 ? _conj(x'x) : x * x')
 
 unscaled_covzm(x::AbstractVector, y::AbstractVector) = dot(x, y)
@@ -436,11 +436,11 @@ function corzm(x::AbstractMatrix, vardim::Int=1)
     return cov2cor!(c, sqrt!(diag(c)))
 end
 corzm(x::AbstractVector, y::AbstractMatrix, vardim::Int=1) =
-    cov2cor!(unscaled_covzm(x, y, vardim), sqrt(sumabs2(x)), sqrt!(sumabs2(y, vardim)))
+    cov2cor!(unscaled_covzm(x, y, vardim), sqrt(sum(abs2, x)), sqrt!(sum(abs2, y, vardim)))
 corzm(x::AbstractMatrix, y::AbstractVector, vardim::Int=1) =
-    cov2cor!(unscaled_covzm(x, y, vardim), sqrt!(sumabs2(x, vardim)), sqrt(sumabs2(y)))
+    cov2cor!(unscaled_covzm(x, y, vardim), sqrt!(sum(abs2, x, vardim)), sqrt(sum(abs2, y)))
 corzm(x::AbstractMatrix, y::AbstractMatrix, vardim::Int=1) =
-    cov2cor!(unscaled_covzm(x, y, vardim), sqrt!(sumabs2(x, vardim)), sqrt!(sumabs2(y, vardim)))
+    cov2cor!(unscaled_covzm(x, y, vardim), sqrt!(sum(abs2, x, vardim)), sqrt!(sum(abs2, y, vardim)))
 
 # corm
 

test/reduce.jl

@@ -61,19 +61,6 @@ z = [-4, -3, 2, 5]
 fz = float(z)
 a = randn(32) # need >16 elements to trigger BLAS code path
 b = complex(randn(32), randn(32))
-@test sumabs(Float64[]) === 0.0
-@test sumabs([Int8(-2)]) === Int32(2)
-@test sumabs(z) === 14
-@test sumabs(fz) === 14.0
-@test sumabs(a) ≈ sum(abs.(a))
-@test sumabs(b) ≈ sum(abs.(b))
-
-@test sumabs2(Float64[]) === 0.0
-@test sumabs2([Int8(-2)]) === Int32(4)
-@test sumabs2(z) === 54
-@test sumabs2(fz) === 54.0
-@test sumabs2(a) ≈ sum(abs2.(a))
-@test sumabs2(b) ≈ sum(abs2.(b))
 
 # check variants of summation for type-stability and other issues (#6069)
 sum2(itr) = invoke(sum, Tuple{Any}, itr)
@@ -152,14 +139,6 @@ prod2(itr) = invoke(prod, Tuple{Any}, itr)
 @test isnan(minimum([4., 3., NaN, 5., 2.]))
 @test isequal(extrema([4., 3., NaN, 5., 2.]), (NaN,NaN))
 
-@test maxabs(Int[]) == 0
-@test_throws ArgumentError Base.minabs(Int[])
-
-@test maxabs(-2) == 2
-@test minabs(-2) == 2
-@test maxabs([1, -2, 3, -4]) == 4
-@test minabs([-1, 2, -3, 4]) == 1
-
 @test maximum(abs2, 3:7) == 49
 @test minimum(abs2, 3:7) == 9