Code

Add L-BFGS reconstruction engine#485

Code

Open

jfowkes wants to merge 21 commits intoptycho:devptycho/ptypy:devfrom jfowkes:jari_lbfgsjfowkes/ptypy:jari_lbfgsCopy head branch name to clipboard

+2,032

Commits

Commits on Sep 10, 2024

• Add L-BFGS reconstruction engine

  

  jfowkes

  authored and

  ptim0626

  committed
  006eab4

• WIP L-BFGS serial, passes some tests

  ptim0626

  committed
  87be276

• Rework LBFGS serial using multiple inheritance

  ptim0626

  committed
  af10ed2

• Finish the GPU version of L-BFGS

  ptim0626

  committed
  5d45877

• Remove local path insert

  ptim0626

  committed
  1d233f7

• Pass num of iteration to engine_iteratre

  ptim0626

  committed
  8cc21b9

• Fix a bug in LBFGS serial about updating

  ptim0626

  committed
  504fd35

• Move smooth gradient to a function in LBFGS serial

  ptim0626

  committed
  d2da354

• Fix the LBFGS pycuda implementation

  ptim0626

  committed
  e211ea1

• Remove smooth gradient (accidentally left in)

  

  jfowkes

  authored and

  ptim0626

  committed
  cb2b2d7

• Fix inefficient copy operations

  

  jfowkes

  authored and

  ptim0626

  committed
  738ff5e

• Remove L-BFGS testing notebooks

  

  jfowkes

  authored and

  ptim0626

  committed
  d47228a

• Add L-BFGS example scripts and notebooks

  

  jfowkes

  authored and

  ptim0626

  committed
  fecdb53

• Add LBFGS cupy engine

  ptim0626

  committed
  1074eb6

• Add LBFGS cupy example

  ptim0626

  committed
  e338230

• Update LBFGS pycuda engine to use FFT Gaussian smoothing

  ptim0626

  committed
  8147dfe

• Add missing import in LBFGS cupy

  ptim0626

  committed
  8ecccf2

• Update LBFGS tests

  ptim0626

  committed
  abecab6

• Add LBFGS pycuda tests

  ptim0626

  committed
  9801f86

• Add LBFGS cupy tests

  ptim0626

  committed
  9eb9871

• Increase LBFGS accelerated tests atol (am I cheating?)

  ptim0626

  committed
  0a4faae