Merge pull request #224 from JuliaRobotics/master

release v0.6.2-rc1

.github/workflows/ci.yml

@@ -17,6 +17,7 @@ jobs:
         version:
           - '1.6'
           - '1.8'
+          - '~1.9.0-0'
           - 'nightly'
         os:
           - ubuntu-latest

Project.toml

@@ -1,7 +1,7 @@
 name = "ApproxManifoldProducts"
 uuid = "9bbbb610-88a1-53cd-9763-118ce10c1f89"
 keywords = ["MM-iSAMv2", "SLAM", "inference", "sum-product", "belief-propagation", "nonparametric", "manifolds", "functional"]
-version = "0.6.1"
+version = "0.6.2"
 
 [deps]
 Base64 = "2a0f44e3-6c83-55bd-87e4-b1978d98bd5f"

src/CommonUtils.jl

@@ -60,21 +60,49 @@ function updateProductSample( dest::BallTreeDensity,
   manikde!(pts, manifolds)
 end
 
-# Returns the covariance (square), not deviation
-function calcCovarianceBasic(M::AbstractManifold, ptsArr::Vector{P}) where P
-  #TODO double check the maths,. it looks like its working at least for groups
-  μ = mean(M, ptsArr)
-  Xcs = vee.(Ref(M), Ref(μ), log.(Ref(M), Ref(μ), ptsArr))
-  Σ = mean(Xcs .* transpose.(Xcs))
-  @debug "calcCovarianceBasic" μ
-  @debug "calcCovarianceBasic" Σ
-  # TODO don't know what to do here so keeping as before, #FIXME it will break
-  # a change between this and previous is that a full covariance matrix is returned
-  msst = Σ
-  msst_ = 0 < sum(1e-10 .< msst) ? maximum(msst) : 1.0
-  return msst_
+
+# TODO this should be a public method relating to getManifold
+function _getManifoldFullOrPart(mkd::ManifoldKernelDensity, aspartial::Bool=true)
+  if aspartial && isPartial(mkd)
+    getManifoldPartial(mkd.manifold, mkd._partial)
+  else
+    mkd.manifold
+  end
 end
 
+function Statistics.mean(mkd::ManifoldKernelDensity, aspartial::Bool=true; kwargs...)
+  return mean(_getManifoldFullOrPart(mkd,aspartial), getPoints(mkd, aspartial); kwargs...)
+end
+"""
+    $SIGNATURES
+
+Alias for overloaded `Statistics.mean`.
+"""
+calcMean(mkd::ManifoldKernelDensity, aspartial::Bool=true) = mean(mkd, aspartial)
+
+function Statistics.std(mkd::ManifoldKernelDensity, aspartial::Bool=true; kwargs...)
+  std(_getManifoldFullOrPart(mkd,aspartial), getPoints(mkd, aspartial); kwargs...)
+end
+function Statistics.var(mkd::ManifoldKernelDensity, aspartial::Bool=true; kwargs...)
+  var(_getManifoldFullOrPart(mkd,aspartial), getPoints(mkd, aspartial); kwargs...)
+end
+function Statistics.cov(mkd::ManifoldKernelDensity, aspartial::Bool=true; basis::Manifolds.AbstractBasis = Manifolds.DefaultOrthogonalBasis(), kwargs...)
+  return cov(_getManifoldFullOrPart(mkd,aspartial), getPoints(mkd, aspartial); basis, kwargs... )
+end
+# function Statistics.mean(mkd::ManifoldKernelDensity; kwargs...)
+#   return mean(mkd.manifold, getPoints(mkd); kwargs...)
+# end
+# function Statistics.cov(mkd::ManifoldKernelDensity; kwargs...) 
+#   cov(mkd.manifold, getPoints(mkd); kwargs...)
+# end
+# function Statistics.std(mkd::ManifoldKernelDensity; kwargs...)
+#   return std(mkd.manifold, getPoints(mkd); kwargs...)
+# end
+# function Statistics.var(mkd::ManifoldKernelDensity; kwargs...)
+#   return var(mkd.manifold, getPoints(mkd); kwargs...)
+# end
+
+
 """
     $SIGNATURES
 

src/Deprecated.jl

@@ -7,6 +7,17 @@
 @deprecate R(th::Real) _Rot.RotMatrix2(th).mat # = [[cos(th);-sin(th)]';[sin(th);cos(th)]'];
 @deprecate R(;x::Real=0.0,y::Real=0.0,z::Real=0.0) (M=SpecialOrthogonal(3);exp(M,identity_element(M),hat(M,Identity(M),[x,y,z]))) # convert(SO3, so3([x,y,z]))
 
+export calcCovarianceBasic
+# Returns the covariance (square), not deviation
+function calcCovarianceBasic(M::AbstractManifold, ptsArr::Vector{P}) where P
+  @warn "`calcCovarianceBasic` is deprecated. Replace with IIF.calcSTDBasicSpread from IIF or `cov` or `var` from Manifolds. See issue AMP#150."
+  μ = mean(M, ptsArr)
+  Xcs = vee.(Ref(M), Ref(μ), log.(Ref(M), Ref(μ), ptsArr))
+  Σ = mean(Xcs .* transpose.(Xcs))
+  msst = Σ
+  msst_ = 0 < sum(1e-10 .< msst) ? maximum(msst) : 1.0
+  return msst_
+end
 
 ## ======================================================================================================
 ## Remove below before v0.7

src/ExportAPI.jl

@@ -19,7 +19,6 @@ export
   getKDEManifoldBandwidths,
   manifoldProduct,
   manikde!,
-  calcCovarianceBasic,
   isPartial,
   calcProductGaussians
 

src/services/ManifoldKernelDensity.jl

@@ -77,39 +77,6 @@ manikde!( M::MB.AbstractManifold,
 ## a few utilities
 ## ==========================================================================================
 
-# TODO this should be a public method relating to getManifold
-function _getManifoldFullOrPart(mkd::ManifoldKernelDensity, aspartial::Bool=true)
-  if aspartial && isPartial(mkd)
-    getManifoldPartial(mkd.manifold, mkd._partial)
-  else
-    mkd.manifold
-  end
-end
-
-function Statistics.mean(mkd::ManifoldKernelDensity, aspartial::Bool=true)
-  M = _getManifoldFullOrPart(mkd, aspartial)
-  mean(M, getPoints(mkd, aspartial))
-end
-
-"""
-    $SIGNATURES
-
-Alias for overloaded `Statistics.mean`.
-"""
-calcMean(mkd::ManifoldKernelDensity, aspartial::Bool=true) = mean(mkd, aspartial)
-
-function Statistics.std(mkd::ManifoldKernelDensity, aspartial::Bool=true; kwargs...)
-  std(_getManifoldFullOrPart(mkd,aspartial), getPoints(mkd, aspartial); kwargs...)
-end
-function Statistics.var(mkd::ManifoldKernelDensity, aspartial::Bool=true; kwargs...)
-  var(_getManifoldFullOrPart(mkd,aspartial), getPoints(mkd, aspartial); kwargs...)
-end
-
-function Statistics.cov(mkd::ManifoldKernelDensity, aspartial::Bool=true; basis::Manifolds.AbstractBasis = Manifolds.DefaultOrthogonalBasis(), kwargs...)
-  return cov(_getManifoldFullOrPart(mkd,aspartial), getPoints(mkd, aspartial); basis, kwargs... )
-end
-
-
 
 _getFieldPartials(mkd::ManifoldKernelDensity{M,B,Nothing}, field::Function, aspartial::Bool=true) where {M,B} = field(mkd)
 

test/runtests.jl

@@ -16,5 +16,6 @@ include("testPartialProductSE2.jl")
 include("basic_se3.jl")
 include("testSymmetry.jl")
 include("testUpdateMKD.jl")
+include("testMKDStats.jl")
 
 #

test/testMKDStats.jl

@@ -0,0 +1,26 @@
+using Test
+using ApproxManifoldProducts
+using Manifolds
+
+##
+
+@testset "Test basic MKD statistics" begin
+##
+
+M = SpecialEuclidean(2)
+u0 = ArrayPartition(zeros(2),[1 0; 0 1.0])
+ϵ = identity_element(M, u0)
+
+pts = [exp(M, ϵ, hat(M, ϵ, [0.05*randn(2);0.75*randn()])) for i in 1:100]
+
+P = manikde!(M, pts)
+
+@test isapprox(M, mean(P), mean(M, pts))
+@test isapprox(var(P), var(M, pts))
+@test isapprox(std(P), std(M, pts))
+@test_broken isapprox(cov(P), cov(M, pts))
+@test isapprox(cov(P), cov(M, pts; basis=DefaultOrthogonalBasis()))
+@test isapprox(cov(P; basis=DefaultOrthonormalBasis()), cov(M, pts))
+
+##
+end