TRY

README.md

@@ -92,6 +92,7 @@ Confidence intervals:
 ```
 
 ```julia
+using CairoMakie
 fig = plot_profiles(prof,
     true_vals=[σ, β...],
     axis_kwargs=(width=200, height=200),

ext/ProfileLikelihoodMakieExt.jl

@@ -104,6 +104,9 @@ function latexify(str)
 end
 
 function default_latex_names(prof, vars)
+    if vars isa Symbol || vars isa String
+        vars = (vars,)
+    end
     variable_symbols = ProfileLikelihood.variable_symbols(prof)
     names = Dict{eltype(vars), Makie.LaTeXString}()
     for var in vars

src/abstract_type_definitions.jl

@@ -59,8 +59,9 @@ abstract type AbstractLikelihoodSolution{N, P} end
 get_mle(sol::AbstractLikelihoodSolution) = sol.mle
 get_mle(sol::AbstractLikelihoodSolution, i) = sol.mle[i]
 get_problem(sol::AbstractLikelihoodSolution) = sol.problem
-function get_optimiser(sol::AbstractLikelihoodSolution)
-    opts = sol.optimiser
+get_optimiser(sol::AbstractLikelihoodSolution) = sol.optimiser
+function _get_optimiser(sol::AbstractLikelihoodSolution)
+    opts = get_optimiser(sol)
     if opts isa Tuple 
         return opts[end]
     else

src/bivariate.jl

@@ -42,7 +42,7 @@ For plotting, see the `plot_profiles` function (requires that you have loaded a
 Returns a [`BivariateProfileLikelihoodSolution`](@ref).
 """
 function bivariate_profile(prob::LikelihoodProblem, sol::LikelihoodSolution, n::NTuple{M,NTuple{2,Int}};
-    alg=get_optimiser(sol),
+    alg=_get_optimiser(sol),
     conf_level::F=0.95,
     confidence_region_method=Val(:contour),
     threshold=get_chisq_threshold(conf_level, 2),

src/profile_likelihood.jl

@@ -46,7 +46,7 @@ the results. For plotting, see the `plot_profiles` function (requires that you h
 Returns a [`ProfileLikelihoodSolution`](@ref).
 """
 function profile(prob::LikelihoodProblem, sol::LikelihoodSolution, n=1:number_of_parameters(prob);
-    alg=get_optimiser(sol),
+    alg=_get_optimiser(sol),
     conf_level::F=0.95,
     confidence_interval_method=:spline,
     threshold=get_chisq_threshold(conf_level),

test/custom_hessian.jl

@@ -1,5 +1,5 @@
 using Optimization
-using ProfileLikelihood
+using ..ProfileLikelihood
 using OptimizationOptimJL
 using ForwardDiff
 using Test

test/mle.jl

@@ -38,15 +38,15 @@ end
     sol = mle(prob, (alg1,))
     @test ProfileLikelihood.get_mle(sol) == sol.mle ≈ [1.0, 1.0]
     @test ProfileLikelihood.get_problem(sol) == sol.problem == prob
-    @test ProfileLikelihood.get_optimiser(sol) == sol.optimiser[end] == NLopt.LN_NELDERMEAD
+    @test ProfileLikelihood._get_optimiser(sol) == sol.optimiser[end] == NLopt.LN_NELDERMEAD
     @test ProfileLikelihood.get_maximum(sol) == sol.maximum ≈ 3.0
     @test ProfileLikelihood.get_retcode(sol) == sol.retcode
     @test ProfileLikelihood.number_of_parameters(prob) == 2
 
     prob = LikelihoodProblem(loglik, θ₀; data=dat, syms=[:α, :β],
         f_kwargs=(adtype=Optimization.AutoFiniteDiff(),))
     sol = mle(prob, (alg1, alg2))
-    @test ProfileLikelihood.get_optimiser(sol) == sol.optimiser[end] == NLopt.LN_BOBYQA
+    @test ProfileLikelihood._get_optimiser(sol) == sol.optimiser[end] == NLopt.LN_BOBYQA
 end
 
 @testset "Check that we can index correctly" begin

test/readme.jl

@@ -49,6 +49,7 @@ resolutions = [1200, 200, 200, 200, 200, 200, 200] # use many points for σ
 param_ranges = construct_profile_ranges(sol, prof_lb, prof_ub, resolutions)
 prof = profile(prob, sol; param_ranges, parallel=true)
 
+using CairoMakie
 fig = plot_profiles(prof,
     true_vals=[σ, β...],
     axis_kwargs=(width=200, height=200),