phase out outdict

src/LMTR_alg.jl

@@ -35,9 +35,6 @@ where F(x) and J(x) are the residual and its Jacobian at x, respectively, ψ(s;
 ### Return values
 
 * `xk`: the final iterate
-* `Fobj_hist`: an array with the history of values of the smooth objective
-* `Hobj_hist`: an array with the history of values of the nonsmooth objective
-* `Complex_hist`: an array with the history of number of inner iterations.
 """
 function LMTR(
   nls::AbstractNLSModel,
@@ -107,12 +104,6 @@ function LMTR(
     treats_bounds ? shifted(h, xk, max.(-Δk, l_bound - xk), min.(Δk, u_bound - xk), selected) :
     shifted(h, xk, Δk, χ)
 
-  Fobj_hist = zeros(maxIter)
-  Hobj_hist = zeros(maxIter)
-  Complex_hist = zeros(Int, maxIter)
-  Grad_hist = zeros(Int, maxIter)
-  Resid_hist = zeros(Int, maxIter)
-
   if verbose > 0
     #! format: off
     @info @sprintf "%6s %8s %8s %8s %7s %7s %8s %7s %7s %7s %7s %1s" "outer" "inner" "f(x)" "h(x)" "√ξ1" "√ξ" "ρ" "Δ" "‖x‖" "‖s‖" "1/ν" "TR"
@@ -141,10 +132,6 @@ function LMTR(
   while !(optimal || tired)
     k = k + 1
     elapsed_time = time() - start_time
-    Fobj_hist[k] = fk
-    Hobj_hist[k] = hk
-    Grad_hist[k] = nls.counters.neval_jtprod_residual + nls.counters.neval_jprod_residual
-    Resid_hist[k] = nls.counters.neval_residual
 
     # model for first prox-gradient iteration
     φ1(d) = begin
@@ -206,8 +193,6 @@ function LMTR(
     subsolver_options.ϵa = ϵa_subsolver
     subsolver_options.Δk = Δk_subsolver
 
-    Complex_hist[k] = iter
-
     sNorm = χ(s)
     xkn .= xk .+ s
     residual!(nls, xkn, Fkn)
@@ -293,11 +278,5 @@ function LMTR(
   set_residuals!(stats, zero(eltype(xk)), ξ1 ≥ 0 ? sqrt(ξ1) : ξ1)
   set_iter!(stats, k)
   set_time!(stats, elapsed_time)
-  set_solver_specific!(stats, :Fhist, Fobj_hist[1:k])
-  set_solver_specific!(stats, :Hhist, Hobj_hist[1:k])
-  set_solver_specific!(stats, :NonSmooth, h)
-  set_solver_specific!(stats, :SubsolverCounter, Complex_hist[1:k])
-  set_solver_specific!(stats, :NLSGradHist, Grad_hist[1:k])
-  set_solver_specific!(stats, :ResidHist, Resid_hist[1:k])
   return stats
 end

src/LM_alg.jl

@@ -34,9 +34,6 @@ and σ > 0 is a regularization parameter.
 ### Return values
 
 * `xk`: the final iterate
-* `Fobj_hist`: an array with the history of values of the smooth objective
-* `Hobj_hist`: an array with the history of values of the nonsmooth objective
-* `Complex_hist`: an array with the history of number of inner iterations.
 """
 function LM(
   nls::AbstractNLSModel,
@@ -102,11 +99,6 @@ function LM(
 
   local ξ1
   k = 0
-  Fobj_hist = zeros(maxIter)
-  Hobj_hist = zeros(maxIter)
-  Complex_hist = zeros(Int, maxIter)
-  Grad_hist = zeros(Int, maxIter)
-  Resid_hist = zeros(Int, maxIter)
 
   if verbose > 0
     #! format: off
@@ -134,10 +126,6 @@ function LM(
   while !(optimal || tired)
     k = k + 1
     elapsed_time = time() - start_time
-    Fobj_hist[k] = fk
-    Hobj_hist[k] = hk
-    Grad_hist[k] = nls.counters.neval_jtprod_residual + nls.counters.neval_jprod_residual
-    Resid_hist[k] = nls.counters.neval_residual
 
     # model for first prox-gradient iteration
     φ1(d) = begin
@@ -199,8 +187,6 @@ function LM(
     subsolver_options.ν = ν_subsolver
     subsolver_options.ϵa = ϵa_subsolver
 
-    Complex_hist[k] = iter
-
     xkn .= xk .+ s
     residual!(nls, xkn, Fkn)
     fkn = dot(Fkn, Fkn) / 2
@@ -245,7 +231,7 @@ function LM(
       σmax = opnorm(Jk)
       νInv = (1 + θ) * (σmax^2 + σk)  # ‖J'J + σₖ I‖ = ‖J‖² + σₖ
 
-      Complex_hist[k] += 1
+      # Complex_hist[k] += 1
     end
 
     if ρk < η1 || ρk == Inf
@@ -282,11 +268,5 @@ function LM(
   set_residuals!(stats, zero(eltype(xk)), ξ1 ≥ 0 ? sqrt(ξ1) : ξ1)
   set_iter!(stats, k)
   set_time!(stats, elapsed_time)
-  set_solver_specific!(stats, :Fhist, Fobj_hist[1:k])
-  set_solver_specific!(stats, :Hhist, Hobj_hist[1:k])
-  set_solver_specific!(stats, :NonSmooth, h)
-  set_solver_specific!(stats, :SubsolverCounter, Complex_hist[1:k])
-  set_solver_specific!(stats, :NLSGradHist, Grad_hist[1:k])
-  set_solver_specific!(stats, :ResidHist, Resid_hist[1:k])
   return stats
 end

src/R2_alg.jl

@@ -11,9 +11,6 @@ mutable struct R2Solver{R, S <: AbstractVector{R}} <: AbstractOptimizationSolver
   u_bound::S
   l_bound_m_x::S
   u_bound_m_x::S
-  Fobj_hist::Vector{R}
-  Hobj_hist::Vector{R}
-  Complex_hist::Vector{Int}
 end
 
 function R2Solver(
@@ -36,24 +33,7 @@ function R2Solver(
     l_bound_m_x = similar(xk, 0)
     u_bound_m_x = similar(xk, 0)
   end
-  Fobj_hist = zeros(R, maxIter)
-  Hobj_hist = zeros(R, maxIter)
-  Complex_hist = zeros(Int, maxIter)
-  return R2Solver(
-    xk,
-    ∇fk,
-    mν∇fk,
-    xkn,
-    s,
-    has_bnds,
-    l_bound,
-    u_bound,
-    l_bound_m_x,
-    u_bound_m_x,
-    Fobj_hist,
-    Hobj_hist,
-    Complex_hist,
-  )
+  return R2Solver(xk, ∇fk, mν∇fk, xkn, s, has_bnds, l_bound, u_bound, l_bound_m_x, u_bound_m_x)
 end
 
 """
@@ -96,9 +76,6 @@ In the second form, instead of `nlp`, the user may pass in
 ### Return values
 
 * `xk`: the final iterate
-* `Fobj_hist`: an array with the history of values of the smooth objective
-* `Hobj_hist`: an array with the history of values of the nonsmooth objective
-* `Complex_hist`: an array with the history of number of inner iterations.
 """
 function R2(nlp::AbstractNLPModel, args...; kwargs...)
   kwargs_dict = Dict(kwargs...)
@@ -120,10 +97,6 @@ function R2(nlp::AbstractNLPModel, args...; kwargs...)
   set_residuals!(stats, zero(eltype(xk)), ξ)
   set_iter!(stats, k)
   set_time!(stats, outdict[:elapsed_time])
-  set_solver_specific!(stats, :Fhist, outdict[:Fhist])
-  set_solver_specific!(stats, :Hhist, outdict[:Hhist])
-  set_solver_specific!(stats, :NonSmooth, outdict[:NonSmooth])
-  set_solver_specific!(stats, :SubsolverCounter, outdict[:Chist])
   return stats
 end
 
@@ -142,17 +115,7 @@ function R2(
   solver = R2Solver(x0, options, similar(x0, 0), similar(x0, 0))
   k, status, fk, hk, ξ = R2!(solver, f, ∇f!, h, options, x0; selected = selected)
   elapsed_time = time() - start_time
-  outdict = Dict(
-    :Fhist => solver.Fobj_hist[1:k],
-    :Hhist => solver.Hobj_hist[1:k],
-    :Chist => solver.Complex_hist[1:k],
-    :NonSmooth => h,
-    :status => status,
-    :fk => fk,
-    :hk => hk,
-    :ξ => ξ,
-    :elapsed_time => elapsed_time,
-  )
+  outdict = Dict(:status => status, :fk => fk, :hk => hk, :ξ => ξ, :elapsed_time => elapsed_time)
   return solver.xk, k, outdict
 end
 
@@ -172,17 +135,7 @@ function R2(
   solver = R2Solver(x0, options, l_bound, u_bound)
   k, status, fk, hk, ξ = R2!(solver, f, ∇f!, h, options, x0; selected = selected)
   elapsed_time = time() - start_time
-  outdict = Dict(
-    :Fhist => solver.Fobj_hist[1:k],
-    :Hhist => solver.Hobj_hist[1:k],
-    :Chist => solver.Complex_hist[1:k],
-    :NonSmooth => h,
-    :status => status,
-    :fk => fk,
-    :hk => hk,
-    :ξ => ξ,
-    :elapsed_time => elapsed_time,
-  )
+  outdict = Dict(:status => status, :fk => fk, :hk => hk, :ξ => ξ, :elapsed_time => elapsed_time)
   return solver.xk, k, outdict
 end
 
@@ -223,9 +176,6 @@ function R2!(
     l_bound_m_x = solver.l_bound_m_x
     u_bound_m_x = solver.u_bound_m_x
   end
-  Fobj_hist = solver.Fobj_hist
-  Hobj_hist = solver.Hobj_hist
-  Complex_hist = solver.Complex_hist
 
   if verbose == 0
     ptf = Inf
@@ -278,23 +228,20 @@ function R2!(
   while !(optimal || tired)
     k = k + 1
     elapsed_time = time() - start_time
-    Fobj_hist[k] = fk
-    Hobj_hist[k] = hk
 
     # define model
     φk(d) = dot(∇fk, d)
     mk(d)::R = φk(d) + ψ(d)::R
 
     prox!(s, ψ, mν∇fk, ν)
-    Complex_hist[k] += 1
     mks = mk(s)
     ξ = hk - mks + max(1, abs(hk)) * 10 * eps()
     sqrt_ξ_νInv = ξ ≥ 0 ? sqrt(ξ / ν) : sqrt(-ξ / ν)
 
     if ξ ≥ 0 && k == 1
       ϵ += ϵr * sqrt_ξ_νInv # make stopping test absolute and relative
     end
-    
+
     if (ξ < 0 && sqrt_ξ_νInv ≤ neg_tol) || (ξ ≥ 0 && sqrt_ξ_νInv ≤ ϵ)
       optimal = true
       continue

src/TRDH_alg.jl

@@ -43,9 +43,6 @@ In the second form, instead of `nlp`, the user may pass in
 ### Return values
 
 * `xk`: the final iterate
-* `Fobj_hist`: an array with the history of values of the smooth objective
-* `Hobj_hist`: an array with the history of values of the nonsmooth objective
-* `Complex_hist`: an array with the history of number of inner iterations.
 """
 function TRDH(
   nlp::AbstractNLPModel{R},
@@ -75,10 +72,6 @@ function TRDH(
   set_residuals!(stats, zero(eltype(xk)), ξ)
   set_iter!(stats, k)
   set_time!(stats, outdict[:elapsed_time])
-  set_solver_specific!(stats, :Fhist, outdict[:Fhist])
-  set_solver_specific!(stats, :Hhist, outdict[:Hhist])
-  set_solver_specific!(stats, :NonSmooth, outdict[:NonSmooth])
-  set_solver_specific!(stats, :SubsolverCounter, outdict[:Chist])
   return stats
 end
 
@@ -182,9 +175,6 @@ function TRDH(
     end
   end
 
-  Fobj_hist = zeros(maxIter)
-  Hobj_hist = zeros(maxIter)
-  Complex_hist = zeros(Int, maxIter)
   if verbose > 0
     #! format: off
     if reduce_TR
@@ -217,16 +207,13 @@ function TRDH(
   while !(optimal || tired)
     k = k + 1
     elapsed_time = time() - start_time
-    Fobj_hist[k] = fk
-    Hobj_hist[k] = hk
 
     # model for prox-gradient step to update Δk if ||s|| is too small and ξ1
     φ1(d) = ∇fk' * d
     mk1(d) = φ1(d) + ψ(d)
 
     if reduce_TR
       prox!(s, ψ, mν∇fk, ν)
-      Complex_hist[k] += 1
       ξ1 = hk - mk1(s) + max(1, abs(hk)) * 10 * eps()
       sqrt_ξ_νInv = ξ1 ≥ 0 ? sqrt(ξ1 / ν) : sqrt(-ξ1 / ν)
 
@@ -252,12 +239,11 @@ function TRDH(
       set_radius!(ψ, Δ_effective)
     end
 
-    # model with diagonal hessian 
+    # model with diagonal hessian
     φ(d) = ∇fk' * d + (d' * (Dk.d .* d)) / 2
     mk(d) = φ(d) + ψ(d)
 
     iprox!(s, ψ, ∇fk, Dk)
-    Complex_hist[k] += 1
 
     sNorm = χ(s)
     xkn .= xk .+ s
@@ -361,17 +347,8 @@ function TRDH(
   else
     :exception
   end
-  outdict = Dict(
-    :Fhist => Fobj_hist[1:k],
-    :Hhist => Hobj_hist[1:k],
-    :Chist => Complex_hist[1:k],
-    :NonSmooth => h,
-    :status => status,
-    :fk => fk,
-    :hk => hk,
-    :ξ => sqrt_ξ_νInv,
-    :elapsed_time => elapsed_time,
-  )
+  outdict =
+    Dict(:status => status, :fk => fk, :hk => hk, :ξ => sqrt_ξ_νInv, :elapsed_time => elapsed_time)
 
   return xk, k, outdict
 end

src/TR_alg.jl

@@ -39,10 +39,7 @@ The Hessian is accessed as an abstract operator and need not be the exact Hessia
 
 ### Return values
 
-* `xk`: the final iterate
-* `Fobj_hist`: an array with the history of values of the smooth objective
-* `Hobj_hist`: an array with the history of values of the nonsmooth objective
-* `Complex_hist`: an array with the history of number of inner iterations.
+* `xk`: the final iterate.
 """
 function TR(
   f::AbstractNLPModel,
@@ -112,9 +109,6 @@ function TR(
     shifted(h, xk, max.(-Δk, l_bound - xk), min.(Δk, u_bound - xk), selected) :
     shifted(h, xk, Δk, χ)
 
-  Fobj_hist = zeros(maxIter)
-  Hobj_hist = zeros(maxIter)
-  Complex_hist = zeros(Int, maxIter)
   if verbose > 0
     #! format: off
     @info @sprintf "%6s %8s %8s %8s %7s %7s %8s %7s %7s %7s %7s %1s" "outer" "inner" "f(x)" "h(x)" "√(ξ1/ν)" "√ξ" "ρ" "Δ" "‖x‖" "‖s‖" "‖Bₖ‖" "TR"
@@ -142,8 +136,6 @@ function TR(
   while !(optimal || tired)
     k = k + 1
     elapsed_time = time() - start_time
-    Fobj_hist[k] = fk
-    Hobj_hist[k] = hk
 
     # model for first prox-gradient step and ξ1
     φ1(d) = ∇fk' * d
@@ -195,8 +187,6 @@ function TR(
     subsolver_options.ϵa = ϵa_subsolver
     subsolver_options.Δk = Δk_subsolver
 
-    Complex_hist[k] = sum(outdict[:Chist])
-
     sNorm = χ(s)
     xkn .= xk .+ s
     fkn = obj(f, xkn)
@@ -282,9 +272,5 @@ function TR(
   set_residuals!(stats, zero(eltype(xk)), sqrt_ξ1_νInv)
   set_iter!(stats, k)
   set_time!(stats, elapsed_time)
-  set_solver_specific!(stats, :Fhist, Fobj_hist[1:k])
-  set_solver_specific!(stats, :Hhist, Hobj_hist[1:k])
-  set_solver_specific!(stats, :NonSmooth, h)
-  set_solver_specific!(stats, :SubsolverCounter, Complex_hist[1:k])
   return stats
 end