Commits

Breadcrumbs

History for

onmaster

User selector

All users

Datepicker

All time

Commit History

Commits on Oct 24, 2024

• redetermine nearest neighbors upon convergence for correlation-based approach in Vecchia approximation

  fabsig

  committed
  3622316

  Copy full SHA for 3622316

Commits on Sep 16, 2024

• fix bug in 'lbfgs' when nugget is profiled-out and the maximal number of line search steps is reached

  fabsig

  committed
  7ce27e2

  Copy full SHA for 7ce27e2

Commits on Jun 22, 2024

• enable option 'reuse_learning_rates_gp_model ' for 'lbfgs' for GPBoost algorithm

  fabsig

  committed
  9ba90d1

  Copy full SHA for 9ba90d1

Commits on Apr 10, 2024

• small code refactoring to avoid gcc warnings

  fabsig

  committed
  c9dfa6a

  Copy full SHA for c9dfa6a

• expose 'lr_cov' parameter to R/Python API for 'lbfgs'

  fabsig

  committed
  72a60ab

  Copy full SHA for 72a60ab

Commits on Apr 3, 2024

• avoid double calculation of intial likelihood for 'lbfgs'

  fabsig

  committed
  7764850

  Copy full SHA for 7764850

• change in 'lbfgs': reduce factor to 1/16 in backtracking if objective has large change

  fabsig

  committed
  103985d

  Copy full SHA for 103985d

• change in 'lbfgs': make step length much smaller in backtracking if objective has large change

  fabsig

  committed
  058040a

  Copy full SHA for 058040a

• change in Laplace approximation: reset mode to zero in every call to log-likelihood

  fabsig

  committed
  49963a1

  Copy full SHA for 49963a1

Commits on Apr 2, 2024

• changes for 'lbfgs': avoid unnecessary calculation of gradients, set learning rate to 0 after maximal number of line search steps

  fabsig

  committed
  75669a3

  Copy full SHA for 75669a3

Commits on Mar 26, 2024

• small correction in convergence criterion for 'lbfgs'

  fabsig

  committed
  a71c658

  Copy full SHA for a71c658

Commits on Feb 27, 2024

• rename .bazelversion to avoid CRAN warning

  fabsig

  committed
  a8548e4

  Copy full SHA for a8548e4

Commits on Feb 19, 2024

• update to fmt 10.2.0 (fmtlib/fmt@8e42eef)

  fabsig

  committed
  9edcc8a

  Copy full SHA for 9edcc8a

Commits on Feb 13, 2024

• convert error to debug-warning in bfgs_v2 backtracking, add 'lbfgs_linesearch_nocedal_wright'

  fabsig

  committed
  bf9eab7

  Copy full SHA for bf9eab7

Commits on Feb 12, 2024

• small change

  fabsig

  committed
  c19501f

  Copy full SHA for c19501f

• small change

  fabsig

  committed
  26a80b7

  Copy full SHA for 26a80b7

• changes in bfgs_v2: change linesearch = LBFGS_LINESEARCH_BACKTRACKING_ARMIJO, small change in covergence criterion, correct correlation-based determination of neighbors for Vecchia approximation

  fabsig

  committed
  026c2da

  Copy full SHA for 026c2da

Commits on Feb 7, 2024

• add safeguard against too large initial step sizes for linear regression coefficients

  fabsig

  committed
  5d43b05

  Copy full SHA for 5d43b05

Commits on Feb 6, 2024

• add temporary logging information for bfgs_v2

  fabsig

  committed
  b56d69b

  Copy full SHA for b56d69b

Commits on Feb 5, 2024

• small change

  fabsig

  committed
  04bf68d

  Copy full SHA for 04bf68d

Commits on Feb 1, 2024

• change handling of correlation-based neighbor search and changes for trace printing for external optimizers

  fabsig

  committed
  40e48b2

  Copy full SHA for 40e48b2

Commits on Jan 31, 2024

• changes in Amjiro condition for optimizer

  fabsig

  committed
  ad3e73b

  Copy full SHA for ad3e73b

Commits on Jan 30, 2024

• add alternative implementation of BFGS

  fabsig

  committed
  f63a9c0

  Copy full SHA for f63a9c0

Commits on Jan 21, 2024

• update to fmt 10.1.1, fast_double_parser 0.7.0

  fabsig

  committed
  189c14b

  Copy full SHA for 189c14b

Commits on Jan 18, 2024

• add vecchia approximation for cov_function = ‘matern_space_time’ with correlation-based neighbor selection

  fabsig

  committed
  ce4b5ed

  Copy full SHA for ce4b5ed

Commits on Jun 19, 2023

• eigen: unbreak macOS ppc build

  barracuda156

  committed
  d2dfb2b

  Copy full SHA for d2dfb2b

Commits on Jan 18, 2023

• faster gradient calculation for sparse matrices (tapering, Vecchia for non-Gaussian), code restructuring

  fabsig

  committed
  60c1ed6

  Copy full SHA for 60c1ed6

Commits on Aug 16, 2022

• [demo] add H-statistic to R demo and SHAP interaction values to Python demo

  fabsig

  committed
  37fde83

  Copy full SHA for 37fde83

Commits on Jun 9, 2022

• allow for grouped random coefficients without random intercept effects

  fabsig

  committed
  68c7b92

  Copy full SHA for 68c7b92

Commits on May 16, 2022

• Add ‘adam’ as optimizer option

  fabsig

  committed
  57f79aa

  Copy full SHA for 57f79aa

Commits on May 12, 2022

• small changes in code, formating, and examples

  fabsig

  committed
  1f78ac5

  Copy full SHA for 1f78ac5

Commits on Jan 12, 2022

• set 'nelder_mead' as automatic fallback option if problems in optimization occur

  fabsig

  committed
  5fc860b

  Copy full SHA for 5fc860b

Commits on Jan 11, 2022

• add 'bfgs' as optimizer option

  fabsig

  committed
  9209b2b

  Copy full SHA for 9209b2b

Commits on Aug 16, 2021

• add Eigen/Core and Eigen/src/Core

  fabsig

  committed
  9981e7c

  Copy full SHA for 9981e7c

• small change in Eigen gitignore

  fabsig

  committed
  ce7aa13

  Copy full SHA for ce7aa13

Pagination

Previous

Next