Code

Bug Fixes in MLE fits#73

Code

Merged

theHenks merged 48 commits intodevlegend-exp/LegendSpecFits.jl:devfrom fittinglegend-exp/LegendSpecFits.jl:fittingCopy head branch name to clipboardOct 6, 2024

+483-261

Commits

Commits on Sep 10, 2024

• Make yscale configurable

  theHenks

  committed
  4905e55

  Copy full SHA for 4905e55

• Add LinSearches to have better control over optimizers

  theHenks

  committed
  df82fb4

  Copy full SHA for df82fb4

• using LineSearches

  theHenks

  committed
  1a9d86e

  Copy full SHA for 1a9d86e

• Use MoreThuente LineSearch, decrease time_limit

  theHenks

  committed
  ea95a13

  Copy full SHA for ea95a13

• Harder standard prior for peak position

  theHenks

  committed
  77e5817

  Copy full SHA for 77e5817

• STASH

  theHenks

  committed
  2f3056b

  Copy full SHA for 2f3056b

Commits on Sep 25, 2024

• LBFGS with MoreThuente and advanced time control + debugging control for all MLE fits

  theHenks

  committed
  d8e7f60

  Copy full SHA for d8e7f60

• Advanced time and memomry control for CTC optimization

  theHenks

  committed
  4b7cf37

  Copy full SHA for 4b7cf37

• Bug fix for ex_gauss PDF in case both theta and sigma are zero

  theHenks

  committed
  b6af41b

  Copy full SHA for b6af41b

• Merge branch 'dev' into fitting

  theHenks

  authored
  f44b293

  Copy full SHA for f44b293

• Convergence warning in case fit did not converge

  theHenks

  committed
  d5c97ba

  Copy full SHA for d5c97ba

• Use LBFGS, memory and time control, convergence measures

  theHenks

  committed
  7e9c80b

  Copy full SHA for 7e9c80b

• Bug fix

  theHenks

  committed
  0ed6403

  Copy full SHA for 0ed6403

• Bug fix v_init target dist

  theHenks

  committed
  1713d55

  Copy full SHA for 1713d55

Commits on Oct 3, 2024

• Add Optimization.jl framework --> more Optimizer control

  theHenks

  committed
  94721c7

  Copy full SHA for 94721c7

• Fix fit_fwhm recipe for legend

  theHenks

  committed
  7ea9081

  Copy full SHA for 7ea9081

• Include lower bounds

  theHenks

  committed
  d9407d6

  Copy full SHA for d9407d6

• Use of LBFGS from Optimization.jl wiht ForwardDiff

  theHenks

  committed
  512961e

  Copy full SHA for 512961e

• Iterative peakstats algorhtihm by shrinking bin_width in case no FWHM can be determined

  theHenks

  committed
  075c2af

  Copy full SHA for 075c2af

• New peak stats extractor for single fits

  theHenks

  committed
  cf6a56a

  Copy full SHA for cf6a56a

• Include lower and upper bounds, use LBFGS from Optimization.jl

  theHenks

  committed
  373ed59

  Copy full SHA for 373ed59

• Better starting values for calibration fitting

  theHenks

  committed
  5a2fbaf

  Copy full SHA for 5a2fbaf

Commits on Oct 5, 2024

• xerrscaling and yerrscaling bug fixes for chi2fits

  theHenks

  committed
  bca4eee

  Copy full SHA for bca4eee

• Enable pseudo prior with better start values, func_err for error propagation

  theHenks

  committed
  cab6caa

  Copy full SHA for cab6caa

• FWHM extraction with pseudo prior, func_err for error propagation

  theHenks

  committed
  11d71fd

  Copy full SHA for 11d71fd

• BGMLE fitting with Optimization LBFGS solver

  theHenks

  committed
  2396681

  Copy full SHA for 2396681

• Bug fix

  theHenks

  committed
  1d9924c

  Copy full SHA for 1d9924c

• Prevent priors from running to 0

  theHenks

  committed
  f78cb04

  Copy full SHA for f78cb04

• Catch case if sigma is 0.0

  theHenks

  committed
  564bd7a

  Copy full SHA for 564bd7a

Commits on Oct 6, 2024

• Stricter prior on chi2fits

  theHenks

  committed
  396b8b3

  Copy full SHA for 396b8b3

• Change debug message

  theHenks

  committed
  c1ae07c

  Copy full SHA for c1ae07c

• Use Optimization.jl, possibility to use higher order corrections

  theHenks

  committed
  af7ddb9

  Copy full SHA for af7ddb9

• Switch to BOBYQA gradient free optimizer for more stable results

  theHenks

  committed
  9cae349

  Copy full SHA for 9cae349

• Stricter standard prior on intercept

  theHenks

  committed
  7fdb8b8

  Copy full SHA for 7fdb8b8

• Looser prior on ENC

  theHenks

  committed
  3833e2a

  Copy full SHA for 3833e2a

• Better p start routine with improved initial ENC guess

  theHenks

  committed
  a88df80

  Copy full SHA for a88df80

• Relax fano noise prior for pol_order 1 fit

  theHenks

  committed
  61c83ea

  Copy full SHA for 61c83ea

• Increase default time limit to 20s

  theHenks

  committed
  a393c5e

  Copy full SHA for a393c5e

• Better start values + bug fixes + remove legacy code

  theHenks

  committed
  cc39963

  Copy full SHA for cc39963

• Use LBFGS from Optimization.jl

  theHenks

  committed
  cfd198e

  Copy full SHA for cfd198e

• Fix psd peakstats

  theHenks

  committed
  efb4c6a

  Copy full SHA for efb4c6a

• Include Optimization.jl in fit_subpeaks

  theHenks

  committed
  84f25e8

  Copy full SHA for 84f25e8

• Use LBFGS from Optimization.jl

  theHenks

  committed
  9ea4c31

  Copy full SHA for 9ea4c31

• Change to LBFGS from Optimization.jl

  theHenks

  committed
  b736af9

  Copy full SHA for b736af9

• Merge branch 'dev' into fitting

  theHenks

  authored
  a845a7b

  Copy full SHA for a845a7b

• Remove Optim.jl dependencies

  theHenks

  committed
  2f3ff9b

  Copy full SHA for 2f3ff9b

• Merge branch 'fitting' of https://github.com/legend-exp/LegendSpecFits.jl into fitting

  theHenks

  committed
  155228a

  Copy full SHA for 155228a

• Fix imports

  theHenks

  committed
  74bcf77

  Copy full SHA for 74bcf77