Issues

• R factors of tall task blocks in MultitaskLassoPath, gated on the expected work

  mlakolar/HighDimensionalModels.jl#194 · Filter by author mlakolar opened on Oct 29, 2026

• Docs still link unscrubbed study trees in the retired-benchmarks snapshot

  mlakolar/HighDimensionalModels.jl#192 · Filter by author mlakolar opened on Oct 29, 2026

• Study archives: a builder, a pre-upload check and a sharper rule

  mlakolar/HighDimensionalModels.jl#190 · Filter by author mlakolar opened on Oct 29, 2026

• Benchmark harness: a generalized-lasso certificate with an unconverged dual reads as a missed target

  mlakolar/HighDimensionalModels.jl#189 · Filter by author mlakolar opened on Oct 29, 2026

• Trend-filtering ADMM: measure rho and iteration policies against certified references

  mlakolar/HighDimensionalModels.jl#188 · Filter by author mlakolar opened on Oct 29, 2026

• Benchmark harness: adaptive, FISTA, Float32 and ill-conditioned sparse reduced-rank cases

  mlakolar/HighDimensionalModels.jl#187 · Filter by author mlakolar opened on Oct 29, 2026

• Float32 sparse reduced-rank fits on designs beyond 1/eps(Float32): solve the block without forming X'X

  mlakolar/HighDimensionalModels.jl#186 · Filter by author mlakolar opened on Oct 29, 2026

• Float32 nuclear_regression: the default tolerance is unattainable

  mlakolar/HighDimensionalModels.jl#185 · Filter by author mlakolar opened on Oct 29, 2026

• Faster transfer paths and CV (CVTransferMultitaskPath)

  mlakolar/HighDimensionalModels.jl#182 · Filter by author mlakolar opened on Oct 29, 2026

• Joint-sparse transfer: target-certified projection, inference and missing responses

  mlakolar/HighDimensionalModels.jl#181 · Filter by author mlakolar opened on Oct 29, 2026

• Coordinate descent reports an infinite relative gap after a few passes in some baseline fold fits

  mlakolar/HighDimensionalModels.jl#180 · Filter by author mlakolar opened on Oct 29, 2026

• Shared-design multitask CV copies the design once per task in every fold

  mlakolar/HighDimensionalModels.jl#179 · Filter by author mlakolar opened on Oct 29, 2026

• Faster sparse_nuclear_regression: restarts with a certificate at the returned point

  mlakolar/HighDimensionalModels.jl#178 · Filter by author mlakolar opened on Oct 29, 2026

• PenaltyGrid: optional zero levels on the automatic axes

  mlakolar/HighDimensionalModels.jl#177 · Filter by author mlakolar opened on Oct 29, 2026

• Task-specific designs for sparse_nuclear_regression and lowrank_plus_sparse

  mlakolar/HighDimensionalModels.jl#176 · Filter by author mlakolar opened on Oct 29, 2026

• Precompile workload: the sqrt_lasso call warns in about 38% of builds

  mlakolar/HighDimensionalModels.jl#175 · Filter by author mlakolar opened on Oct 29, 2026

• ProximalSolvers: row steps for any group penalty and a certified GroupNormLinf, then move the package's row solver onto them

  mlakolar/HighDimensionalModels.jl#174 · Filter by author mlakolar opened on Oct 29, 2026

• Faster ℓ∞-type row solves for row ℓ1/ℓ∞ and the dirty model with many tasks

  mlakolar/HighDimensionalModels.jl#173 · Filter by author mlakolar opened on Oct 29, 2026

• Quantile regression: fits over a grid of quantile levels with warm starts

  mlakolar/HighDimensionalModels.jl#159 · Filter by author mlakolar opened on Oct 28, 2026

• Quantile regression: observation weights in fits, pivotal calibration and cross-validation

  mlakolar/HighDimensionalModels.jl#157 · Filter by author mlakolar opened on Oct 28, 2026

• Quantile regression: exact solution paths in λ and τ (breakpoint enumeration)

  mlakolar/HighDimensionalModels.jl#156 · Filter by author mlakolar opened on Oct 28, 2026

• Quantile regression: committed R reference fixtures (quantreg, rqPen, conquer)

  mlakolar/HighDimensionalModels.jl#154 · Filter by author mlakolar opened on Oct 28, 2026

• Signal cross-validation after #113: quadratic and million-point scale, Float32 quadratic stopping

  mlakolar/HighDimensionalModels.jl#151 · Filter by author mlakolar opened on Oct 28, 2026

• Re-measure Gaussian SLOPE calibration, sparse reduced-rank setup reuse and knockoff strategies on the current stack

  mlakolar/HighDimensionalModels.jl#150 · Filter by author mlakolar opened on Oct 28, 2026

• Knockoff covariance validation: scale-aware positive definiteness and a symmetry tolerance

  mlakolar/HighDimensionalModels.jl#148 · Filter by author mlakolar opened on Oct 28, 2026