Better support for custom point types in conjugate_gradient_descent (#596)

Author: mateuszbaran

Changelog.md

@@ -6,14 +6,18 @@ The file was started with Version `0.4`.
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
-## [0.5.35] April 3, 2026
+## [0.5.35] April 16, 2026
 
 ### Changed
 
 * Improved formatting of the references in the Readme.md (#586)
 * Bump compat for RecursiveArrayTools.jl to include version 4
 * deactivate CompatHelper Action and solely use dependabot
 
+### Fixed
+
+* The default line search in `conjugate_gradient_descent` is now `ArmijoLinesearchStepsize` instead of `ArmijoLinesearch`, which makes it work well with custom point types.
+
 ## [0.5.34] March 3, 2026
 
 ### Fixed

Project.toml

@@ -1,6 +1,6 @@
 name = "Manopt"
 uuid = "0fc0a36d-df90-57f3-8f93-d78a9fc72bb5"
-version = "0.5.34"
+version = "0.5.35"
 authors = [{family-names = "Bergmann", given-names = "Ronny", alias = "kellertuer", city = "Trondheim", affiliation = "Norwegian University of Science and Technology", country = "NO", email = "manopt@ronnybergmann.net", orcid = "https://orcid.org/0000-0001-8342-7218", website = "https://ronnybergmann.net"}]
 
 [workspace]

src/solvers/conjugate_gradient_descent.jl

@@ -4,7 +4,7 @@ function default_stepsize(
         retraction_method = default_retraction_method(M),
     )
     # take a default with a slightly defensive initial step size.
-    return ArmijoLinesearchStepsize(
+    return ArmijoLinesearch(
         M; retraction_method = retraction_method, initial_stepsize = 1.0
     )
 end

test/solvers/test_conjugate_gradient.jl

@@ -1,5 +1,6 @@
 using Manopt, Manifolds, ManifoldsBase, Test, Random, LinearAlgebra
 using LinearAlgebra: Diagonal, dot, eigvals, eigvecs
+using ManifoldDiff: grad_distance
 
 @testset "Conjugate Gradient Descent" begin
     @testset "Conjugate Gradient coefficient rules" begin
@@ -31,7 +32,7 @@ using LinearAlgebra: Diagonal, dot, eigvals, eigvecs
             initial_gradient = zero_vector(M, x0),
         )
         @test s1.coefficient(dmp, s1, 1) == 0
-        @test default_stepsize(M, typeof(s1)) isa Manopt.ArmijoLinesearchStepsize
+        @test default_stepsize(M, typeof(s1)) isa Manopt.ManifoldDefaultsFactory{Manopt.ArmijoLinesearchStepsize}
         @test Manopt.get_message(s1) == ""
 
         dU = Manopt.ConjugateDescentCoefficient()
@@ -394,4 +395,13 @@ using LinearAlgebra: Diagonal, dot, eigvals, eigvecs
         )
         @test q2 ≈ [1, 0, 0] rtol = 1.0e-7
     end
+
+    @testset "Custom point types" begin
+        M = Hyperbolic(2)
+        data = PoincareBallPoint.([[0.1, 0.2], [0.3, 0.25], [0.35, 0.4]])
+        n = length(data)
+        f(M, p) = sum(1 / (2 * n) * distance.(Ref(M), Ref(p), data) .^ 2)
+        grad_f(M, p) = sum(1 / n * grad_distance.(Ref(M), data, Ref(p)))
+        @test conjugate_gradient_descent(M, f, grad_f, data[1]) isa PoincareBallPoint
+    end
 end