JSO Compliance : LMTR #201

MaxenceGollier · 2025-08-08T09:59:37Z

We should merge after #199 and #200.

test/test_bounds.jl

src/LMModel.jl

src/LMTR_alg.jl

codecov · 2025-08-08T10:02:58Z

Codecov Report

❌ Patch coverage is 91.51515% with 14 lines in your changes missing coverage. Please review.
✅ Project coverage is 85.56%. Comparing base (e0f214d) to head (31569ea).
⚠️ Report is 184 commits behind head on master.

Files with missing lines	Patch %	Lines
src/LMTR_alg.jl	91.51%	14 Missing ⚠️

Additional details and impacted files

@@             Coverage Diff             @@
##           master     #201       +/-   ##
===========================================
+ Coverage   61.53%   85.56%   +24.02%     
===========================================
  Files          11       13        +2     
  Lines        1292     1579      +287     
===========================================
+ Hits          795     1351      +556     
+ Misses        497      228      -269

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

src/LMTR_alg.jl

dpo · 2025-09-21T15:38:27Z

Did you compare with the master branch here?

MaxenceGollier · 2025-09-23T16:57:50Z

@dpo i can not compare with the old version because LMTR uses $$\sqrt{\xi}$$ as an optimality measure instead of $$\sqrt{\xi/\nu}$$...
All tests pass at least..

dpo · 2025-09-23T19:31:19Z

@dpo i can not compare with the old version because LMTR uses ξ as an optimality measure instead of ξ / ν ... All tests pass at least..

In that case, please open a separate PR that introduces the correct stopping test in LMTR. Then we'll be able to compare.

MaxenceGollier · 2025-09-24T03:28:09Z

LMTR Tests

Without bounds

Subsolver : R2

On this branch,

using LinearAlgebra, NLPModels, NLPModelsModifiers, RegularizedProblems, ProximalOperators, Random, ShiftedProximalOperators
Random.seed!(0)
bpdn = bpdn_model(1)[2]
λ = norm(grad(bpdn, zeros(bpdn.meta.nvar)), Inf) / 10
h = NormL0(λ)
x0 = randn(bpdn.meta.nvar)

import Pkg; Pkg.add(url = "https://github.com/MaxenceGollier/RegularizedOptimization.jl.git", rev = "LMTR-JSO")
using RegularizedOptimization
options = ROSolverOptions(β = 1.0, α = 1.0, ϵa = 1e-3, ϵr = 1e-3, verbose = 1)

LMTR(bpdn, h, NormLinf(1.0), options, x0 = x0)
[ Info:  outer  inner     f(x)     h(x)  √(ξ1/ν)        ρ        Δ      ‖x‖      ‖s‖        ν LMTR 
[ Info:      0    238  9.6e+01  2.3e+01  1.4e+01  1.0e+00  5.7e-01  4.0e+00  5.7e-01  3.3e-01 ↗
[ Info:      1     21  1.2e+01  1.8e+01  5.3e+00  1.0e+00  4.3e-01  3.4e+00  4.3e-01  4.6e-01 ↗
[ Info:      2      6  5.3e-01  1.3e+01  1.7e+00  1.0e+00  2.1e-01  3.0e+00  2.1e-01  4.6e-01 ↗
[ Info:      3     17  9.5e-02  1.2e+01  7.3e-01  1.0e+00  2.1e-01  2.9e+00  2.1e-01  4.6e-01 ↗
[ Info:      4      6  9.2e-03  1.2e+01  9.0e-01  1.0e+00  2.1e-01  2.9e+00  2.1e-01  4.6e-01 ↗
[ Info:      5      1  8.2e-03  1.1e+01  3.1e-02  1.0e+00  2.5e-03  2.9e+00  2.5e-03  4.6e-01 ↗
[ Info:      6      1  7.4e-03  1.1e+01  2.9e-02  1.0e+00  2.6e-03  2.9e+00  2.6e-03  4.6e-01 ↗
[ Info:      7      1  6.8e-03  1.1e+01  2.7e-02  1.0e+00  2.6e-03  2.9e+00  2.6e-03  4.6e-01 ↗
[ Info:      8      1  6.2e-03  1.1e+01  2.5e-02  1.0e+00  2.5e-03  2.9e+00  2.5e-03  4.6e-01 ↗
[ Info:      9      1  5.7e-03  1.1e+01  2.4e-02  1.0e+00  2.4e-03  2.9e+00  2.4e-03  4.6e-01 ↗
[ Info:     10      1  5.3e-03  1.1e+01  2.2e-02  1.0e+00  2.2e-03  2.9e+00  2.2e-03  4.6e-01 ↗
[ Info:     11      1  4.9e-03  1.1e+01  2.1e-02  1.0e+00  2.1e-03  2.9e+00  2.1e-03  4.6e-01 ↗
[ Info:     12      1  4.6e-03  1.1e+01  2.0e-02  1.0e+00  2.0e-03  2.9e+00  2.0e-03  4.6e-01 ↗
[ Info:     13      1  4.3e-03  1.1e+01  1.9e-02  1.0e+00  1.9e-03  2.9e+00  1.9e-03  4.6e-01 ↗
[ Info:     14      0  4.0e-03  1.1e+01  1.8e-02  1.0e+00  1.8e-03  2.9e+00  1.8e-03  4.6e-01 ↗
[ Info:     15      0  3.8e-03  1.1e+01  1.8e-02  1.0e+00  1.6e-03  2.9e+00  1.6e-03  4.6e-01 ↗
[ Info:     16      0  3.7e-03  1.1e+01  1.7e-02  1.0e+00  1.4e-03  2.9e+00  1.4e-03  4.6e-01 ↗
[ Info:     17      0  3.6e-03  1.1e+01  1.7e-02  1.0e+00  1.3e-03  2.9e+00  1.3e-03  4.6e-01 ↗
[ Info:     18      0  3.4e-03  1.1e+01  1.6e-02  1.0e+00  1.2e-03  2.9e+00  1.2e-03  4.6e-01 ↗
[ Info:     19      0  3.3e-03  1.1e+01  1.6e-02  1.0e+00  1.2e-03  2.9e+00  1.2e-03  4.6e-01 ↗
[ Info:     20      0  3.2e-03  1.1e+01  1.5e-02  1.0e+00  1.1e-03  2.9e+00  1.1e-03  4.6e-01 ↗
[ Info:     21      0  3.1e-03  1.1e+01  1.5e-02  1.0e+00  1.7e+00  2.9e+00  1.1e-03  4.6e-01
[ Info: LMTR: terminating with √(ξ1/ν) = 0.015115537832344829
"Execution stats: first-order stationary"

On the branch with fixed optimality measure,

using LinearAlgebra, NLPModels, NLPModelsModifiers, RegularizedProblems, ProximalOperators, Random, ShiftedProximalOperators
Random.seed!(0)
bpdn = bpdn_model(1)[2]
λ = norm(grad(bpdn, zeros(bpdn.meta.nvar)), Inf) / 10
h = NormL0(λ)
x0 = randn(bpdn.meta.nvar)

import Pkg; Pkg.add(url = "https://github.com/MaxenceGollier/RegularizedOptimization.jl.git", rev = "LMTR_OptMeasure")
using RegularizedOptimization
options = ROSolverOptions(β = 1.0, α = 1.0, ϵa = 1e-3, ϵr = 1e-3, verbose = 10)

LMTR(bpdn, h, NormLinf(1.0), options, x0 = x0)
[ Info:  outer    inner     f(x)     h(x)     √ξ1      √ξ        ρ       Δ     ‖x‖     ‖s‖       ν TR
[ Info:      1      238  9.6e+01  2.3e+01 8.2e+00 9.5e+00  1.0e+00 5.7e-01 4.0e+00 5.7e-01 3.3e-01 ↗
[ Info:      2       21  1.2e+01  1.8e+01 3.6e+00 4.0e+00  1.0e+00 4.3e-01 3.4e+00 4.3e-01 4.6e-01 ↗
[ Info:      3        6  5.3e-01  1.3e+01 1.2e+00 1.3e+00  1.0e+00 2.1e-01 3.0e+00 2.1e-01 4.6e-01 ↗
[ Info:      4       17  9.5e-02  1.2e+01 4.9e-01 5.6e-01  1.0e+00 2.1e-01 2.9e+00 2.1e-01 4.6e-01 ↗
[ Info:      5        6  9.2e-03  1.2e+01 6.1e-01 6.1e-01  1.0e+00 2.1e-01 2.9e+00 2.1e-01 4.6e-01 ↗
[ Info:      6        1  8.2e-03  1.1e+01 2.1e-02 2.7e-02  1.0e+00 2.5e-03 2.9e+00 2.5e-03 4.6e-01 ↗
[ Info:      2       21  1.2e+01  1.8e+01 3.6e+00 4.0e+00  1.0e+00 4.3e-01 3.4e+00 4.3e-01 4.6e-01 ↗
[ Info:      3        6  5.3e-01  1.3e+01 1.2e+00 1.3e+00  1.0e+00 2.1e-01 3.0e+00 2.1e-01 4.6e-01 ↗
[ Info:      4       17  9.5e-02  1.2e+01 4.9e-01 5.6e-01  1.0e+00 2.1e-01 2.9e+00 2.1e-01 4.6e-01 ↗
[ Info:      5        6  9.2e-03  1.2e+01 6.1e-01 6.1e-01  1.0e+00 2.1e-01 2.9e+00 2.1e-01 4.6e-01 ↗
[ Info:      6        1  8.2e-03  1.1e+01 2.1e-02 2.7e-02  1.0e+00 2.5e-03 2.9e+00 2.5e-03 4.6e-01 ↗
[ Info:      6        1  8.2e-03  1.1e+01 2.1e-02 2.7e-02  1.0e+00 2.5e-03 2.9e+00 2.5e-03 4.6e-01 ↗
[ Info:      7        1  7.4e-03  1.1e+01 2.0e-02 2.5e-02  1.0e+00 2.6e-03 2.9e+00 2.6e-03 4.6e-01 ↗
[ Info:      8        1  6.8e-03  1.1e+01 1.8e-02 2.4e-02  1.0e+00 2.6e-03 2.9e+00 2.6e-03 4.6e-01 ↗
[ Info:      9        1  6.2e-03  1.1e+01 1.7e-02 2.2e-02  1.0e+00 2.5e-03 2.9e+00 2.5e-03 4.6e-01 ↗
[ Info:     10        1  5.7e-03  1.1e+01 1.6e-02 2.1e-02  1.0e+00 2.4e-03 2.9e+00 2.4e-03 4.6e-01 ↗
[ Info:     11        1  5.3e-03  1.1e+01 1.5e-02 2.0e-02  1.0e+00 2.2e-03 2.9e+00 2.2e-03 4.6e-01 ↗
[ Info:     12        1  4.9e-03  1.1e+01 1.4e-02 1.9e-02  1.0e+00 2.1e-03 2.9e+00 2.1e-03 4.6e-01 ↗
[ Info:     13        1  4.6e-03  1.1e+01 1.4e-02 1.8e-02  1.0e+00 2.0e-03 2.9e+00 2.0e-03 4.6e-01 ↗
[ Info:     14        1  4.3e-03  1.1e+01 1.3e-02 1.7e-02  1.0e+00 1.9e-03 2.9e+00 1.9e-03 4.6e-01 ↗
[ Info:     15        0  4.0e-03  1.1e+01 1.2e-02 1.2e-02  1.0e+00 1.8e-03 2.9e+00 1.8e-03 4.6e-01 ↗
[ Info:     16        0  3.8e-03  1.1e+01 1.2e-02 1.2e-02  1.0e+00 1.6e-03 2.9e+00 1.6e-03 4.6e-01 ↗
[ Info:     17        0  3.7e-03  1.1e+01 1.2e-02 1.1e-02  1.0e+00 1.4e-03 2.9e+00 1.4e-03 4.6e-01 ↗
[ Info:     18        0  3.6e-03  1.1e+01 1.1e-02 1.1e-02  1.0e+00 1.3e-03 2.9e+00 1.3e-03 4.6e-01 ↗
[ Info:     19        0  3.4e-03  1.1e+01 1.1e-02 1.1e-02  1.0e+00 1.2e-03 2.9e+00 1.2e-03 4.6e-01 ↗
[ Info:     20        0  3.3e-03  1.1e+01 1.1e-02 1.1e-02  1.0e+00 1.2e-03 2.9e+00 1.2e-03 4.6e-01 ↗
[ Info:     21        0  3.2e-03  1.1e+01 1.0e-02 1.0e-02  1.0e+00 1.1e-03 2.9e+00 1.1e-03 4.6e-01 ↗
[ Info:     22        1  3.1e-03  1.1e+01 1.0e-02 1.0e-02          1.7e+00 2.9e+00 1.1e-03 4.6e-01
[ Info: LMTR: terminating with √(ξ1/ν) = 0.01511553783234532
"Execution stats: first-order stationary"

Subsolver : TRDH

On this branch,

using LinearAlgebra, NLPModels, NLPModelsModifiers, RegularizedProblems, ProximalOperators, Random, ShiftedProximalOperators
Random.seed!(0)
bpdn = bpdn_model(1)[2]
λ = norm(grad(bpdn, zeros(bpdn.meta.nvar)), Inf) / 10
h = NormL0(λ)
x0 = randn(bpdn.meta.nvar)

import Pkg; Pkg.add(url = "https://github.com/MaxenceGollier/RegularizedOptimization.jl.git", rev = "LMTR-JSO")
using RegularizedOptimization
options = ROSolverOptions(β = 1.0, α = 1.0, ϵa = 1e-3, ϵr = 1e-3, verbose = 1)

LMTR(bpdn, h, NormLinf(1.0), options, x0 = x0, subsolver = TRDHSolver)
[ Info:  outer  inner     f(x)     h(x)  √(ξ1/ν)        ρ        Δ      ‖x‖      ‖s‖        ν LMTR 
[ Info:      0      0  9.6e+01  2.3e+01  1.4e+01  1.0e+00  5.7e-01  4.0e+00  5.7e-01  3.3e-01 ↗
[ Info:      1      0  4.2e+01  2.0e+01  9.3e+00  1.0e+00  5.9e-01  3.5e+00  5.9e-01  4.6e-01 ↗
[ Info:      2      0  1.3e+01  1.9e+01  5.2e+00  1.0e+00  4.3e-01  3.0e+00  4.3e-01  4.7e-01 ↗
[ Info:      3      0  4.3e+00  1.8e+01  2.8e+00  1.0e+00  2.9e-01  2.8e+00  2.9e-01  4.7e-01 ↗
[ Info:      4      0  1.8e+00  1.7e+01  2.1e+00  1.0e+00  2.7e-01  2.6e+00  2.7e-01  4.7e-01 ↗
[ Info:      5      0  9.1e-01  1.6e+01  1.1e+00  1.0e+00  2.1e-01  2.5e+00  2.1e-01  4.7e-01 ↗
[ Info:      6      0  5.6e-01  1.6e+01  9.4e-01  1.0e+00  2.2e-01  2.5e+00  2.2e-01  4.7e-01 ↗
[ Info:      7      0  3.9e-01  1.6e+01  5.2e-01  1.0e+00  4.3e-02  2.5e+00  4.3e-02  4.7e-01 ↗
[ Info:      8      0  2.8e-01  1.6e+01  7.7e-01  1.0e+00  2.2e-01  2.5e+00  2.2e-01  4.7e-01 ↗
[ Info:      9      0  2.3e-01  1.6e+01  6.8e-01  1.0e+00  2.3e-01  2.5e+00  2.3e-01  4.7e-01 ↗
[ Info:     10      0  1.9e-01  1.6e+01  5.7e-01  1.0e+00  2.1e-01  2.5e+00  2.1e-01  4.7e-01 ↗
[ Info:     11      0  1.5e-01  1.6e+01  5.5e-01  1.0e+00  2.1e-01  2.5e+00  2.1e-01  4.7e-01 ↗
[ Info:     12      0  1.2e-01  1.6e+01  2.6e-01  1.0e+00  2.1e-02  2.5e+00  2.1e-02  4.7e-01 ↗
[ Info:     13      0  9.0e-02  1.6e+01  5.0e-01  1.0e+00  2.1e-01  2.5e+00  2.1e-01  4.7e-01 ↗
[ Info:     14      0  8.5e-02  1.5e+01  3.9e-01  1.0e+00  2.2e-01  2.5e+00  2.2e-01  4.7e-01 ↗
[ Info:     15      0  7.0e-02  1.5e+01  1.9e-01  1.0e+00  1.3e-02  2.5e+00  1.3e-02  4.7e-01 ↗
[ Info:     16      0  5.4e-02  1.5e+01  5.8e-01  1.0e+00  2.1e-01  2.5e+00  2.1e-01  4.7e-01 ↗
[ Info:     17      0  5.9e-02  1.5e+01  2.0e-01  1.0e+00  1.5e-02  2.5e+00  1.5e-02  4.7e-01 ↗
[ Info:     18      0  4.3e-02  1.5e+01  1.6e-01  1.0e+00  1.2e-02  2.5e+00  1.2e-02  4.7e-01 ↗
[ Info:     19      0  3.2e-02  1.5e+01  1.3e-01  1.0e+00  9.6e-03  2.5e+00  9.6e-03  4.7e-01 ↗
[ Info:     20      0  2.5e-02  1.5e+01  3.4e-01  1.0e+00  2.1e-01  2.5e+00  2.1e-01  4.7e-01 ↗
[ Info:     21      0  2.7e-02  1.5e+01  1.2e-01  1.0e+00  1.1e-02  2.5e+00  1.1e-02  4.7e-01 ↗
[ Info:     22      0  2.1e-02  1.5e+01  3.3e-01  1.0e+00  2.1e-01  2.5e+00  2.1e-01  4.7e-01 ↗
[ Info:     23      0  2.5e-02  1.5e+01  1.2e-01  1.0e+00  1.0e-02  2.5e+00  1.0e-02  4.7e-01 ↗
[ Info:     24      0  1.8e-02  1.5e+01  3.3e-01  1.0e+00  2.1e-01  2.5e+00  2.1e-01  4.7e-01 ↗
[ Info:     25      0  2.1e-02  1.5e+01  1.2e-01  1.0e+00  8.3e-03  2.5e+00  8.3e-03  4.7e-01 ↗
[ Info:     26      0  1.5e-02  1.5e+01  9.2e-02  1.0e+00  6.6e-03  2.5e+00  6.6e-03  4.7e-01 ↗
[ Info:     27      0  1.2e-02  1.5e+01  7.6e-02  1.0e+00  5.4e-03  2.5e+00  5.4e-03  4.7e-01 ↗
[ Info:     28      0  9.2e-03  1.5e+01  6.5e-02  1.0e+00  4.6e-03  2.5e+00  4.6e-03  4.7e-01 ↗
[ Info:     29      0  7.4e-03  1.5e+01  4.5e-01  1.0e+00  2.1e-01  2.5e+00  2.1e-01  4.7e-01 ↗
[ Info:     30      0  2.0e-02  1.5e+01  1.3e-01  1.0e+00  1.0e-02  2.5e+00  1.0e-02  4.7e-01 ↗
[ Info:     31      0  1.3e-02  1.5e+01  3.3e-01  1.0e+00  2.1e-01  2.5e+00  2.1e-01  4.7e-01 ↗
[ Info:     32      0  1.6e-02  1.5e+01  1.1e-01  1.0e+00  8.0e-03  2.5e+00  8.0e-03  4.7e-01 ↗
[ Info:     33      0  1.1e-02  1.5e+01  8.6e-02  1.0e+00  6.1e-03  2.5e+00  6.1e-03  4.7e-01 ↗
[ Info:     34      0  7.9e-03  1.5e+01  6.8e-02  1.0e+00  5.0e-03  2.5e+00  5.0e-03  4.7e-01 ↗
[ Info:     35      0  6.0e-03  1.5e+01  5.6e-02  1.0e+00  4.2e-03  2.5e+00  4.2e-03  4.7e-01 ↗
[ Info:     36      0  4.6e-03  1.5e+01  4.6e-02  1.0e+00  3.6e-03  2.5e+00  3.6e-03  4.7e-01 ↗
[ Info:     37      0  3.7e-03  1.5e+01  4.0e-02  1.0e+00  3.0e-03  2.5e+00  3.0e-03  4.7e-01 ↗
[ Info:     38      0  3.0e-03  1.5e+01  3.4e-02  1.0e+00  2.6e-03  2.5e+00  2.6e-03  4.7e-01 ↗
[ Info:     39      0  2.5e-03  1.5e+01  3.0e-02  1.0e+00  2.3e-03  2.4e+00  2.3e-03  4.7e-01 ↗
[ Info:     40      0  2.1e-03  1.5e+01  2.7e-02  1.0e+00  2.0e-03  2.4e+00  2.0e-03  4.7e-01 ↗
[ Info:     41      0  1.8e-03  1.5e+01  2.4e-02  1.0e+00  1.7e-03  2.4e+00  1.7e-03  4.7e-01 ↗
[ Info:     42      0  1.5e-03  1.5e+01  2.2e-02  1.0e+00  1.5e-03  2.4e+00  1.5e-03  4.7e-01 ↗
[ Info:     43      0  1.3e-03  1.5e+01  2.0e-02  1.0e+00  1.4e-03  2.4e+00  1.4e-03  4.7e-01 ↗
[ Info:     44      0  1.1e-03  1.5e+01  1.8e-02  1.0e+00  1.2e-03  2.4e+00  1.2e-03  4.7e-01 ↗
[ Info:     45      0  9.7e-04  1.5e+01  1.6e-02  1.0e+00  1.1e-03  2.4e+00  1.1e-03  4.7e-01 ↗
[ Info:     46      0  8.5e-04  1.5e+01  1.5e-02  1.0e+00  1.8e+00  2.4e+00  1.0e-03  4.7e-01
[ Info: LMTR: terminating with √(ξ1/ν) = 0.015080393658632408

On the branch with fixed optimality measure,

using LinearAlgebra, NLPModels, NLPModelsModifiers, RegularizedProblems, ProximalOperators, Random, ShiftedProximalOperators
Random.seed!(0)
bpdn = bpdn_model(1)[2]
λ = norm(grad(bpdn, zeros(bpdn.meta.nvar)), Inf) / 10
h = NormL0(λ)
x0 = randn(bpdn.meta.nvar)

import Pkg; Pkg.add(url = "https://github.com/MaxenceGollier/RegularizedOptimization.jl.git", rev = "LMTR_OptMeasure")
using RegularizedOptimization
options = ROSolverOptions(β = 1.0, α = 1.0, ϵa = 1e-3, ϵr = 1e-3, verbose = 10)

LMTR(bpdn, h, NormLinf(1.0), options, x0 = x0, subsolver = TRDH)
[ Info:  outer    inner     f(x)     h(x)     √ξ1      √ξ        ρ       Δ     ‖x‖     ‖s‖       ν TR
[ Info:      1        0  9.6e+01  2.3e+01 8.2e+00 7.5e+00  1.0e+00 5.7e-01 4.0e+00 5.7e-01 3.3e-01 ↗
[ Info:      2        0  4.2e+01  2.0e+01 6.3e+00 5.5e+00  1.0e+00 5.9e-01 3.5e+00 5.9e-01 4.6e-01 ↗
[ Info:      3        0  1.3e+01  1.9e+01 3.6e+00 3.1e+00  1.0e+00 4.3e-01 3.0e+00 4.3e-01 4.7e-01 ↗
[ Info:      4        0  4.3e+00  1.8e+01 1.9e+00 1.7e+00  1.0e+00 2.9e-01 2.8e+00 2.9e-01 4.7e-01 ↗
[ Info:      5        0  1.8e+00  1.7e+01 1.4e+00 1.3e+00  1.0e+00 2.7e-01 2.6e+00 2.7e-01 4.7e-01 ↗
[ Info:      6        0  9.1e-01  1.6e+01 7.2e-01 6.6e-01  1.0e+00 2.1e-01 2.5e+00 2.1e-01 4.7e-01 ↗
[ Info:      7        0  5.6e-01  1.6e+01 6.5e-01 5.9e-01  1.0e+00 2.2e-01 2.5e+00 2.2e-01 4.7e-01 ↗
[ Info:      8        0  3.9e-01  1.6e+01 3.5e-01 3.3e-01  1.0e+00 4.3e-02 2.5e+00 4.3e-02 4.7e-01 ↗
[ Info:      9        0  2.8e-01  1.6e+01 5.3e-01 4.8e-01  1.0e+00 2.2e-01 2.5e+00 2.2e-01 4.7e-01 ↗
[ Info:     10        0  2.3e-01  1.6e+01 4.6e-01 4.3e-01  1.0e+00 2.3e-01 2.5e+00 2.3e-01 4.7e-01 ↗
[ Info:     11        0  1.9e-01  1.6e+01 3.9e-01 3.6e-01  1.0e+00 2.1e-01 2.5e+00 2.1e-01 4.7e-01 ↗
[ Info:     12        0  1.5e-01  1.6e+01 3.7e-01 3.5e-01  1.0e+00 2.1e-01 2.5e+00 2.1e-01 4.7e-01 ↗
[ Info:     13        0  1.2e-01  1.6e+01 1.8e-01 1.7e-01  1.0e+00 2.1e-02 2.5e+00 2.1e-02 4.7e-01 ↗
[ Info:     14        0  9.0e-02  1.6e+01 3.4e-01 3.1e-01  1.0e+00 2.1e-01 2.5e+00 2.1e-01 4.7e-01 ↗
[ Info:     15        0  8.5e-02  1.5e+01 2.7e-01 2.5e-01  1.0e+00 2.2e-01 2.5e+00 2.2e-01 4.7e-01 ↗
[ Info:     16        0  7.0e-02  1.5e+01 1.3e-01 1.3e-01  1.0e+00 1.3e-02 2.5e+00 1.3e-02 4.7e-01 ↗
[ Info:     17        0  5.4e-02  1.5e+01 4.0e-01 3.6e-01  1.0e+00 2.1e-01 2.5e+00 2.1e-01 4.7e-01 ↗
[ Info:     18        0  5.9e-02  1.5e+01 1.4e-01 1.3e-01  1.0e+00 1.5e-02 2.5e+00 1.5e-02 4.7e-01 ↗
[ Info:     19        0  4.3e-02  1.5e+01 1.1e-01 1.0e-01  1.0e+00 1.2e-02 2.5e+00 1.2e-02 4.7e-01 ↗
[ Info:     20        0  3.2e-02  1.5e+01 8.8e-02 8.4e-02  1.0e+00 9.6e-03 2.5e+00 9.6e-03 4.7e-01 ↗
[ Info:     21        0  2.5e-02  1.5e+01 2.3e-01 2.1e-01  1.0e+00 2.1e-01 2.5e+00 2.1e-01 4.7e-01 ↗
[ Info:     22        0  2.7e-02  1.5e+01 8.4e-02 8.0e-02  1.0e+00 1.1e-02 2.5e+00 1.1e-02 4.7e-01 ↗
[ Info:     23        0  2.1e-02  1.5e+01 2.3e-01 2.0e-01  1.0e+00 2.1e-01 2.5e+00 2.1e-01 4.7e-01 ↗
[ Info:     24        0  2.5e-02  1.5e+01 8.3e-02 7.9e-02  1.0e+00 1.0e-02 2.5e+00 1.0e-02 4.7e-01 ↗
[ Info:     25        0  1.8e-02  1.5e+01 2.3e-01 2.1e-01  1.0e+00 2.1e-01 2.5e+00 2.1e-01 4.7e-01 ↗
[ Info:     26        0  2.1e-02  1.5e+01 7.9e-02 7.5e-02  1.0e+00 8.3e-03 2.5e+00 8.3e-03 4.7e-01 ↗
[ Info:     27        0  1.5e-02  1.5e+01 6.3e-02 6.0e-02  1.0e+00 6.6e-03 2.5e+00 6.6e-03 4.7e-01 ↗
[ Info:     28        0  1.2e-02  1.5e+01 5.2e-02 5.0e-02  1.0e+00 5.4e-03 2.5e+00 5.4e-03 4.7e-01 ↗
[ Info:     29        0  9.2e-03  1.5e+01 4.4e-02 4.3e-02  1.0e+00 4.6e-03 2.5e+00 4.6e-03 4.7e-01 ↗
[ Info:     30        0  7.4e-03  1.5e+01 3.1e-01 2.8e-01  1.0e+00 2.1e-01 2.5e+00 2.1e-01 4.7e-01 ↗
[ Info:     31        0  2.0e-02  1.5e+01 8.6e-02 8.0e-02  1.0e+00 1.0e-02 2.5e+00 1.0e-02 4.7e-01 ↗
[ Info:     32        0  1.3e-02  1.5e+01 2.3e-01 2.1e-01  1.0e+00 2.1e-01 2.5e+00 2.1e-01 4.7e-01 ↗
[ Info:     33        0  1.6e-02  1.5e+01 7.7e-02 7.2e-02  1.0e+00 8.0e-03 2.5e+00 8.0e-03 4.7e-01 ↗
[ Info:     34        0  1.1e-02  1.5e+01 5.9e-02 5.5e-02  1.0e+00 6.1e-03 2.5e+00 6.1e-03 4.7e-01 ↗
[ Info:     35        0  7.9e-03  1.5e+01 4.7e-02 4.4e-02  1.0e+00 5.0e-03 2.5e+00 5.0e-03 4.7e-01 ↗
[ Info:     36        0  6.0e-03  1.5e+01 3.8e-02 3.6e-02  1.0e+00 4.2e-03 2.5e+00 4.2e-03 4.7e-01 ↗
[ Info:     37        0  4.6e-03  1.5e+01 3.2e-02 3.1e-02  1.0e+00 3.6e-03 2.5e+00 3.6e-03 4.7e-01 ↗
[ Info:     38        0  3.7e-03  1.5e+01 2.7e-02 2.6e-02  1.0e+00 3.0e-03 2.5e+00 3.0e-03 4.7e-01 ↗
[ Info:     39        0  3.0e-03  1.5e+01 2.4e-02 2.3e-02  1.0e+00 2.6e-03 2.5e+00 2.6e-03 4.7e-01 ↗
[ Info:     40        0  2.5e-03  1.5e+01 2.1e-02 2.0e-02  1.0e+00 2.3e-03 2.4e+00 2.3e-03 4.7e-01 ↗
[ Info:     41        0  2.1e-03  1.5e+01 1.8e-02 1.8e-02  1.0e+00 2.0e-03 2.4e+00 2.0e-03 4.7e-01 ↗
[ Info:     42        0  1.8e-03  1.5e+01 1.6e-02 1.6e-02  1.0e+00 1.7e-03 2.4e+00 1.7e-03 4.7e-01 ↗
[ Info:     43        0  1.5e-03  1.5e+01 1.5e-02 1.5e-02  1.0e+00 1.5e-03 2.4e+00 1.5e-03 4.7e-01 ↗
[ Info:     44        0  1.3e-03  1.5e+01 1.3e-02 1.3e-02  1.0e+00 1.4e-03 2.4e+00 1.4e-03 4.7e-01 ↗
[ Info:     45        0  1.1e-03  1.5e+01 1.2e-02 1.2e-02  1.0e+00 1.2e-03 2.4e+00 1.2e-03 4.7e-01 ↗
[ Info:     46        0  9.7e-04  1.5e+01 1.1e-02 1.1e-02  1.0e+00 1.1e-03 2.4e+00 1.1e-03 4.7e-01 ↗
[ Info:     47        1  8.5e-04  1.5e+01 1.0e-02 1.0e-02          1.8e+00 2.4e+00 1.0e-03 4.7e-01
[ Info: LMTR: terminating with √(ξ1/ν) = 0.015080393658632542

With bounds

Subsolver : R2

On this branch,

using LinearAlgebra, NLPModels, NLPModelsModifiers, RegularizedProblems, ProximalOperators, Random, ShiftedProximalOperators
Random.seed!(0)
bpdn = bpdn_model(1, bounds = true)[2]
λ = norm(grad(bpdn, zeros(bpdn.meta.nvar)), Inf) / 10
h = NormL1(λ)

#import Pkg; Pkg.add(url = "https://github.com/MaxenceGollier/RegularizedOptimization.jl.git", rev = "LMTR-JSO")
using RegularizedOptimization
options = ROSolverOptions(β = 1.0, α = 1.0, ϵa = 1e-3, ϵr = 1e-3, verbose = 1)

LMTR(bpdn, h, NormLinf(1.0), options)
[ Info:  outer  inner     f(x)     h(x)  √(ξ1/ν)        ρ        Δ      ‖x‖      ‖s‖        ν LMTR
[ Info:      0     34  2.2e+00  0.0e+00  1.3e+00  1.0e+00  1.8e-01  0.0e+00  1.8e-01  3.3e-01 ↗
[ Info:      1      4  1.1e+00  4.3e-01  7.8e-01  1.0e+00  1.2e-01  1.8e-01  1.2e-01  3.3e-01 ↗
[ Info:      2      3  8.5e-01  3.7e-01  6.9e-01  1.0e+00  1.0e-01  3.0e-01  1.0e-01  3.3e-01 ↗
[ Info:      3      3  6.9e-01  3.3e-01  6.0e-01  1.0e+00  9.2e-02  4.0e-01  9.2e-02  3.3e-01 ↗
[ Info:      4      3  5.3e-01  3.3e-01  5.1e-01  1.0e+00  7.9e-02  4.9e-01  7.9e-02  3.3e-01 ↗
[ Info:      5      3  4.0e-01  3.5e-01  4.2e-01  1.0e+00  6.5e-02  5.7e-01  6.5e-02  3.3e-01 ↗
[ Info:      6      3  3.0e-01  3.8e-01  3.3e-01  1.0e+00  5.3e-02  6.4e-01  5.3e-02  3.3e-01 ↗
[ Info:      7      4  2.2e-01  4.1e-01  2.6e-01  1.0e+00  4.3e-02  6.9e-01  4.3e-02  3.3e-01 ↗
[ Info:      8      5  1.6e-01  4.4e-01  2.0e-01  1.0e+00  3.4e-02  7.3e-01  3.4e-02  3.3e-01 ↗
[ Info:      9      3  1.3e-01  4.6e-01  1.5e-01  1.0e+00  2.7e-02  7.7e-01  2.7e-02  3.3e-01 ↗
[ Info:     10      3  1.0e-01  4.7e-01  1.2e-01  1.0e+00  2.2e-02  7.9e-01  2.2e-02  3.3e-01 ↗
[ Info:     11      2  8.5e-02  4.8e-01  8.8e-02  1.0e+00  1.8e-02  8.2e-01  1.8e-02  3.3e-01 ↗
[ Info:     12      3  7.3e-02  4.9e-01  6.6e-02  1.0e+00  1.4e-02  8.3e-01  1.4e-02  3.3e-01 ↗
[ Info:     13      3  6.5e-02  5.0e-01  5.0e-02  1.0e+00  1.1e-02  8.5e-01  1.1e-02  3.3e-01 ↗
[ Info:     14      2  6.0e-02  5.0e-01  4.0e-02  1.0e+00  9.3e-03  8.6e-01  9.3e-03  3.3e-01 ↗
[ Info:     15      1  5.7e-02  5.1e-01  3.3e-02  1.0e+00  7.8e-03  8.7e-01  7.8e-03  3.3e-01 ↗
[ Info:     16      1  5.6e-02  5.1e-01  2.8e-02  1.0e+00  6.6e-03  8.8e-01  6.6e-03  3.3e-01 ↗
[ Info:     17      1  5.5e-02  5.1e-01  2.3e-02  1.0e+00  5.5e-03  8.8e-01  5.5e-03  3.3e-01 ↗
[ Info:     18      1  5.4e-02  5.1e-01  1.9e-02  1.0e+00  4.6e-03  8.9e-01  4.6e-03  3.3e-01 ↗
[ Info:     19      1  5.4e-02  5.1e-01  1.6e-02  1.0e+00  3.9e-03  8.9e-01  3.9e-03  3.3e-01 ↗
[ Info:     20      1  5.3e-02  5.1e-01  1.4e-02  1.0e+00  3.3e-03  9.0e-01  3.3e-03  3.3e-01 ↗
[ Info:     21      1  5.3e-02  5.1e-01  1.2e-02  1.0e+00  2.8e-03  9.0e-01  2.8e-03  3.3e-01 ↗
[ Info:     22      1  5.2e-02  5.1e-01  9.8e-03  1.0e+00  2.4e-03  9.0e-01  2.4e-03  3.3e-01 ↗
[ Info:     23      0  5.2e-02  5.1e-01  8.3e-03  1.0e+00  2.0e-03  9.1e-01  2.0e-03  3.3e-01 ↗
[ Info:     24      0  5.2e-02  5.1e-01  7.0e-03  1.0e+00  1.7e-03  9.1e-01  1.7e-03  3.3e-01 ↗
[ Info:     25      0  5.2e-02  5.1e-01  6.0e-03  1.0e+00  1.4e-03  9.1e-01  1.4e-03  3.3e-01 ↗
[ Info:     26      0  5.2e-02  5.1e-01  5.1e-03  1.0e+00  1.2e-03  9.1e-01  1.2e-03  3.3e-01 ↗
[ Info:     27      0  5.2e-02  5.1e-01  4.4e-03  1.0e+00  1.0e-03  9.1e-01  1.0e-03  3.3e-01 ↗
[ Info:     28      0  5.2e-02  5.1e-01  3.8e-03  1.0e+00  8.6e-04  9.1e-01  8.6e-04  3.3e-01 ↗
[ Info:     29      0  5.2e-02  5.1e-01  3.3e-03  1.0e+00  7.3e-04  9.1e-01  7.3e-04  3.3e-01 ↗
[ Info:     30      0  5.1e-02  5.1e-01  2.8e-03  1.0e+00  6.2e-04  9.1e-01  6.2e-04  3.3e-01 ↗
[ Info:     31      0  5.1e-02  5.1e-01  2.5e-03  1.0e+00  5.3e-04  9.2e-01  5.3e-04  3.3e-01 ↗
[ Info:     32      0  5.1e-02  5.1e-01  2.2e-03  1.0e+00  1.0e+00  9.2e-01  4.5e-04  3.3e-01
[ Info: LMTR: terminating with √(ξ1/ν) = 0.002162985823872107
"Execution stats: first-order stationary"

On the branch with fixed optimality measure,

using LinearAlgebra, NLPModels, NLPModelsModifiers, RegularizedProblems, ProximalOperators, Random, ShiftedProximalOperators
Random.seed!(0)
bpdn = bpdn_model(1, bounds = true)[2]
λ = norm(grad(bpdn, zeros(bpdn.meta.nvar)), Inf) / 10
h = NormL1(λ)

import Pkg; Pkg.add(url = "https://github.com/MaxenceGollier/RegularizedOptimization.jl.git", rev = "LMTR_OptMeasure")
using RegularizedOptimization
options = ROSolverOptions(β = 1.0, α = 1.0, ϵa = 1e-3, ϵr = 1e-3, verbose = 10)

LMTR(bpdn, h, NormLinf(1.0), options, x0 = x0)
[ Info:  outer    inner     f(x)     h(x)     √ξ1      √ξ        ρ       Δ     ‖x‖     ‖s‖       ν TR
[ Info:      1       34  2.2e+00  0.0e+00 7.6e-01 8.4e-01  1.0e+00 1.8e-01 0.0e+00 1.8e-01 3.3e-01 ↗
[ Info:      2        4  1.1e+00  4.3e-01 4.5e-01 5.1e-01  1.0e+00 1.2e-01 1.8e-01 1.2e-01 3.3e-01 ↗
[ Info:      3        3  8.5e-01  3.7e-01 4.0e-01 4.5e-01  1.0e+00 1.0e-01 3.0e-01 1.0e-01 3.3e-01 ↗
[ Info:      4        3  6.9e-01  3.3e-01 3.5e-01 3.9e-01  1.0e+00 9.2e-02 4.0e-01 9.2e-02 3.3e-01 ↗
[ Info:      5        3  5.3e-01  3.3e-01 3.0e-01 3.3e-01  1.0e+00 7.9e-02 4.9e-01 7.9e-02 3.3e-01 ↗
[ Info:      6        3  4.0e-01  3.5e-01 2.4e-01 2.7e-01  1.0e+00 6.5e-02 5.7e-01 6.5e-02 3.3e-01 ↗
[ Info:      7        3  3.0e-01  3.8e-01 1.9e-01 2.2e-01  1.0e+00 5.3e-02 6.4e-01 5.3e-02 3.3e-01 ↗
[ Info:      8        4  2.2e-01  4.1e-01 1.5e-01 1.7e-01  1.0e+00 4.3e-02 6.9e-01 4.3e-02 3.3e-01 ↗
[ Info:      9        5  1.6e-01  4.4e-01 1.2e-01 1.3e-01  1.0e+00 3.4e-02 7.3e-01 3.4e-02 3.3e-01 ↗
[ Info:     10        3  1.3e-01  4.6e-01 8.8e-02 1.0e-01  1.0e+00 2.7e-02 7.7e-01 2.7e-02 3.3e-01 ↗
[ Info:     11        3  1.0e-01  4.7e-01 6.7e-02 7.6e-02  1.0e+00 2.2e-02 7.9e-01 2.2e-02 3.3e-01 ↗
[ Info:     12        2  8.5e-02  4.8e-01 5.1e-02 5.7e-02  1.0e+00 1.8e-02 8.2e-01 1.8e-02 3.3e-01 ↗
[ Info:     13        3  7.3e-02  4.9e-01 3.8e-02 4.2e-02  1.0e+00 1.4e-02 8.3e-01 1.4e-02 3.3e-01 ↗
[ Info:     14        3  6.5e-02  5.0e-01 2.9e-02 3.0e-02  1.0e+00 1.1e-02 8.5e-01 1.1e-02 3.3e-01 ↗
[ Info:     15        2  6.0e-02  5.0e-01 2.3e-02 2.2e-02  1.0e+00 9.3e-03 8.6e-01 9.3e-03 3.3e-01 ↗
[ Info:     16        1  5.7e-02  5.1e-01 1.9e-02 1.8e-02  1.0e+00 7.8e-03 8.7e-01 7.8e-03 3.3e-01 ↗
[ Info:     17        1  5.6e-02  5.1e-01 1.6e-02 1.5e-02  1.0e+00 6.6e-03 8.8e-01 6.6e-03 3.3e-01 ↗
[ Info:     18        1  5.5e-02  5.1e-01 1.3e-02 1.3e-02  1.0e+00 5.5e-03 8.8e-01 5.5e-03 3.3e-01 ↗
[ Info:     19        1  5.4e-02  5.1e-01 1.1e-02 1.1e-02  1.0e+00 4.6e-03 8.9e-01 4.6e-03 3.3e-01 ↗
[ Info:     20        1  5.4e-02  5.1e-01 9.4e-03 9.2e-03  1.0e+00 3.9e-03 8.9e-01 3.9e-03 3.3e-01 ↗
[ Info:     21        1  5.3e-02  5.1e-01 7.9e-03 7.8e-03  1.0e+00 3.3e-03 9.0e-01 3.3e-03 3.3e-01 ↗
[ Info:     22        1  5.3e-02  5.1e-01 6.7e-03 6.6e-03  1.0e+00 2.8e-03 9.0e-01 2.8e-03 3.3e-01 ↗
[ Info:     23        1  5.2e-02  5.1e-01 5.6e-03 5.6e-03  1.0e+00 2.4e-03 9.0e-01 2.4e-03 3.3e-01 ↗
[ Info:     24        0  5.2e-02  5.1e-01 4.8e-03 4.6e-03  1.0e+00 2.0e-03 9.1e-01 2.0e-03 3.3e-01 ↗
[ Info:     25        0  5.2e-02  5.1e-01 4.0e-03 3.9e-03  1.0e+00 1.7e-03 9.1e-01 1.7e-03 3.3e-01 ↗
[ Info:     26        0  5.2e-02  5.1e-01 3.5e-03 3.3e-03  1.0e+00 1.4e-03 9.1e-01 1.4e-03 3.3e-01 ↗
[ Info:     27        0  5.2e-02  5.1e-01 3.0e-03 2.8e-03  1.0e+00 1.2e-03 9.1e-01 1.2e-03 3.3e-01 ↗
[ Info:     28        0  5.2e-02  5.1e-01 2.5e-03 2.4e-03  1.0e+00 1.0e-03 9.1e-01 1.0e-03 3.3e-01 ↗
[ Info:     29        0  5.2e-02  5.1e-01 2.2e-03 2.1e-03  1.0e+00 8.6e-04 9.1e-01 8.6e-04 3.3e-01 ↗
[ Info:     30        0  5.2e-02  5.1e-01 1.9e-03 1.8e-03  1.0e+00 7.3e-04 9.1e-01 7.3e-04 3.3e-01 ↗
[ Info:     31        0  5.1e-02  5.1e-01 1.6e-03 1.6e-03  1.0e+00 6.2e-04 9.1e-01 6.2e-04 3.3e-01 ↗
[ Info:     32        0  5.1e-02  5.1e-01 1.4e-03 1.4e-03  1.0e+00 5.3e-04 9.2e-01 5.3e-04 3.3e-01 ↗
[ Info:     33        1  5.1e-02  5.1e-01 1.2e-03 1.2e-03          1.0e+00 9.2e-01 4.5e-04 3.3e-01
[ Info: LMTR: terminating with √(ξ1/ν) = 0.002162985823872107
"Execution stats: first-order stationary"

MaxenceGollier · 2025-09-24T03:29:40Z

@dpo, feel free to review when you have some time !

MaxenceGollier commented Aug 8, 2025

View reviewed changes

test/test_bounds.jl Outdated Show resolved Hide resolved

MaxenceGollier commented Aug 8, 2025

View reviewed changes

src/LMModel.jl Outdated Show resolved Hide resolved

MaxenceGollier commented Aug 8, 2025

View reviewed changes

src/LMTR_alg.jl Outdated Show resolved Hide resolved

dpo reviewed Sep 11, 2025

View reviewed changes

src/LMTR_alg.jl Outdated Show resolved Hide resolved

MaxenceGollier added 14 commits September 23, 2025 11:39

JSO compliant LMTR

1a4158b

update doc and function signatures

cbe1116

remove old doc

40247b2

add LMModel

6f12183

add JSO-LMTR version to tests

b51af50

implement LMTR regularizedNLP signatures

900baf0

update x0 to x kwarg

b9cada7

minor bug fix

d57fcb9

implicit jacobian update

128b338

remove temporary struct

d16d9ed

update callback docstring

4fd85b2

add test bound back

817ed93

fix bounds initialization

3a58dcc

remove FHist from tests

cb48329

MaxenceGollier force-pushed the LMTR-JSO branch from b2680ca to cb48329 Compare September 23, 2025 15:47

remove untracked files

b09d01c

MaxenceGollier requested a review from dpo September 23, 2025 18:28

Merge branch 'JuliaSmoothOptimizers:master' into LMTR-JSO

7c92060

MaxenceGollier mentioned this pull request Sep 24, 2025

Fix optimality measure in LMTR #223

Closed

MaxenceGollier added 3 commits September 23, 2025 21:47

LMTR :fix arrow display

64a092e

LMTR : remove Jv, Jtv

460143b

LMTR: use delta_k instead of delta_effective

31569ea

dpo approved these changes Sep 29, 2025

View reviewed changes

dpo merged commit 649bb81 into JuliaSmoothOptimizers:master Sep 29, 2025
13 of 14 checks passed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

JSO Compliance : LMTR #201

JSO Compliance : LMTR #201

Uh oh!

MaxenceGollier commented Aug 8, 2025

Uh oh!

Uh oh!

Uh oh!

Uh oh!

codecov bot commented Aug 8, 2025 •

edited

Loading

Uh oh!

Uh oh!

dpo commented Sep 21, 2025

Uh oh!

MaxenceGollier commented Sep 23, 2025

Uh oh!

dpo commented Sep 23, 2025

Uh oh!

MaxenceGollier commented Sep 24, 2025

Uh oh!

MaxenceGollier commented Sep 24, 2025

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

JSO Compliance : LMTR #201

JSO Compliance : LMTR #201

Uh oh!

Conversation

MaxenceGollier commented Aug 8, 2025

Uh oh!

Uh oh!

Uh oh!

Uh oh!

codecov bot commented Aug 8, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

Uh oh!

dpo commented Sep 21, 2025

Uh oh!

MaxenceGollier commented Sep 23, 2025

Uh oh!

dpo commented Sep 23, 2025

Uh oh!

MaxenceGollier commented Sep 24, 2025

LMTR Tests

Without bounds

Subsolver : R2

Subsolver : TRDH

With bounds

Subsolver : R2

Uh oh!

MaxenceGollier commented Sep 24, 2025

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

codecov bot commented Aug 8, 2025 •

edited

Loading