Code

PBC fix bundle#563

Code

Draft

evaleev wants to merge 112 commits intomasterm-a-d-n-e-s-s/madness:masterfrom jm/cleanup/nuclear-density-functorm-a-d-n-e-s-s/madness:jm/cleanup/nuclear-density-functorCopy head branch name to clipboard

+5,297-1,635

Commits

Commits on Nov 7, 2024

• Preliminary changes to support periodic MRA-SCF

  

  JonathonMisiewicz

  authored and

  evaleev

  committed
  42319c1

  Copy full SHA for 42319c1

• Fix issues in potential.

  

  JonathonMisiewicz

  authored and

  evaleev

  committed
  15b87dc

  Copy full SHA for 15b87dc

• Change periodic handling in NuclearDensityFunctor

  

  JonathonMisiewicz

  authored and

  evaleev

  committed
  70ab9df

  Copy full SHA for 70ab9df

• removed unused Molecule::mol_nuclear_charge_density()

  evaleev

  committed
  c48e4a7

  Copy full SHA for c48e4a7

• Molecule::nuclear_charge_density() can (uniformly) adjust size of the nuclei programmatically

  evaleev

  committed
  18d9ccf

  Copy full SHA for 18d9ccf

• document "default" nuclear model that produces smoothed coulomb potential

  evaleev

  committed
  7fbd9e3

  Copy full SHA for 7fbd9e3

• introduce BoundaryConditions::is_periodic_any

  evaleev

  committed
  f2f0b6d

  Copy full SHA for f2f0b6d

• dox++

  evaleev

  committed
  b115e80

  Copy full SHA for b115e80

• clean up NuclearDensityFunctor

  evaleev

  committed
  122a3b4

  Copy full SHA for 122a3b4

• Hotfixes

  

  JonathonMisiewicz

  authored and

  evaleev

  committed
  5d162a4

  Copy full SHA for 5d162a4

• testfilter

  

  JonathonMisiewicz

  authored and

  evaleev

  committed
  52e28cd

  Copy full SHA for 52e28cd

• fixed erfcr example

  evaleev

  committed
  47d59ba

  Copy full SHA for 47d59ba

• testfilter: can only filter out moments in compressed representation

  evaleev

  committed
  afa1122

  Copy full SHA for afa1122

Commits on Nov 22, 2024

• Truncation of Gaussian expansion of integral kernel in SeparatedConvolution can be controlled by the user, if needed; the default is unchanged (truncate the expansion if periodic, not otherwise)

  evaleev

  committed
  084edc6

  Copy full SHA for 084edc6

• bump actions/{ccache,checkout} to v4

  evaleev

  committed
  b4c1c54

  Copy full SHA for b4c1c54

• one more ctor for XCOperator to be able to construct outside SCF/Nemo

  evaleev

  committed
  7084146

  Copy full SHA for 7084146

Commits on Nov 28, 2024

• typos

  evaleev

  committed
  5d487c4

  Copy full SHA for 5d487c4

• factored out BoundaryConditions from funcdefaults.h

  evaleev

  committed
  bf94970

  Copy full SHA for bf94970

• introduced Key::distsq_{bc,periodic} to compute |l|^2 for periodic and mixed BC

  evaleev

  committed
  c5811cb

  Copy full SHA for c5811cb

• Displacements ctor uses Key::distsq_periodic for PBC

  evaleev

  committed
  aba1178

  Copy full SHA for aba1178

• FunctionImpl::do_apply : shell-based screening in PBC case computes distances using distsq_periodic

  evaleev

  committed
  602f4d3

  Copy full SHA for 602f4d3

Commits on Nov 29, 2024

• FunctionImpl::{do_apply,do_apply_directed_screening} : no need to consider periodic displacements since operator kernel includes lattice summation

  evaleev

  committed
  4ae06b0

  Copy full SHA for 4ae06b0

Commits on Dec 3, 2024

• [stubmpi] implement Gather, Gatherv, and Type_get_extent

  evaleev

  committed
  9db4790

  Copy full SHA for 9db4790

Commits on Dec 4, 2024

• typos

  evaleev

  committed
  6ecab21

  Copy full SHA for 6ecab21

• SeparatedConvolution can deal with mixed boundary conditions (periodic in some directions, nonperiodic in others) ... ?

  Show description for 24db345
  evaleev

  committed
  24db345

  Copy full SHA for 24db345

• introduced array_of_bools, an ergonomic sequence of bools

  evaleev

  committed
  992195a

  Copy full SHA for 992195a

Commits on Dec 5, 2024

• Displacements: periodic displacements are generated according to the boundary conditions active

  evaleev

  committed
  0502f6e

  Copy full SHA for 0502f6e

• introduced array_of_bools::{and,or}_{front,back}

  evaleev

  committed
  e7c5c1f

  Copy full SHA for e7c5c1f

• enforce_bc: handle multiple displacements

  evaleev

  committed
  e4ce92a

  Copy full SHA for e4ce92a

• boundary condition handling switched to array_of_bools

  Show description for f2bae3e
  evaleev

  committed
  f2bae3e

  Copy full SHA for f2bae3e

Commits on Dec 6, 2024

• enforce_in_volume(n,l) takes l by const ref

  evaleev

  committed
  5db605c

  Copy full SHA for 5db605c

• FunctionDefaults: boundary conditions set before set_defaults or startup are not reset by those functions

  evaleev

  committed
  25dea91

  Copy full SHA for 25dea91

• Displacements: generate periodic displacements on the first call, even if the boundary conditions set at the time are not periodic, since that's the current default behavior. If periodic boundary c…

  Show description for 23ec4de
  evaleev

  committed
  23ec4de

  Copy full SHA for 23ec4de

• BoundaryConditions: array ctors

  evaleev

  committed
  4d1174b

  Copy full SHA for 4d1174b

Commits on Dec 10, 2024

• typo

  evaleev

  committed
  c89faf4

  Copy full SHA for c89faf4

• Convolution1D: arg really stands for wavenumber

  evaleev

  committed
  52535a5

  Copy full SHA for 52535a5

Commits on Dec 12, 2024

• copied Kahan accumulator from MPQC

  evaleev

  committed
  fbbceb3

  Copy full SHA for fbbceb3

• GaussianConvolution1D supports range restriction on kernel

  evaleev

  committed
  ccbb1e8

  Copy full SHA for ccbb1e8

• testing rang-restricted rnlp (aka trnlp) + cleanup of test_rnlp

  evaleev

  committed
  488a603

  Copy full SHA for 488a603

• typo

  evaleev

  committed
  4967add

  Copy full SHA for 4967add

Commits on Dec 13, 2024

• caching GaussianConvolution1D takes into account bloch-k and D

  evaleev

  committed
  18c7e19

  Copy full SHA for 18c7e19

• can construct range-restricted versions of integral operators

  evaleev

  committed
  befc1f8

  Copy full SHA for befc1f8

• fixed Convolution1D::get_issmall for range-restricted case

  Show description for 9e92446
  evaleev

  committed
  9e92446

  Copy full SHA for 9e92446

• can construct range-limited operators with range limits specified for each axis to handle mixed BC

  evaleev

  committed
  c149748

  Copy full SHA for c149748

Commits on Dec 16, 2024

• minor cleanup for issmall

  evaleev

  committed
  341ccfc

  Copy full SHA for 341ccfc

• small test for range-limited rnlij

  evaleev

  committed
  d0f0227

  Copy full SHA for d0f0227

• bigger test for range-limited rnlij

  evaleev

  committed
  96f3563

  Copy full SHA for 96f3563

Commits on Dec 17, 2024

• yet bigger test for range-limited rnlij

  evaleev

  committed
  8bd0652

  Copy full SHA for 8bd0652

Commits on Dec 18, 2024

• typo

  evaleev

  committed
  2dfa2f4

  Copy full SHA for 2dfa2f4

• introduced madness:is_trivially_copyable trait (can be specialized to override legally-required limitations of std::is_trivially_copyable)

  evaleev

  committed
  3f33d06

  Copy full SHA for 3f33d06

Commits on Dec 21, 2024

• typo

  evaleev

  committed
  2d8e16e

  Copy full SHA for 2d8e16e

Commits on Dec 27, 2024

• introduced SeparatedConvolution::get_{k,rank}

  evaleev

  committed
  d27d486

  Copy full SHA for d27d486

• Merge branch 'refs/heads/master' into jm/cleanup/nuclear-density-functor

  evaleev

  committed
  a69e9b3

  Copy full SHA for a69e9b3

Commits on Dec 30, 2024

• typo

  evaleev

  committed
  d8ae908

  Copy full SHA for d8ae908

Commits on Jan 1, 2025

• kernel range restriction of a convolution is encoded by a std::optional, not plain int

  Show description for e1686f4
  evaleev

  committed
  e1686f4

  Copy full SHA for e1686f4

• Q: why Convolution1D::nonstandard differs from MRAQC paper appendix?

  Show description for 862a556
  evaleev

  committed
  862a556

  Copy full SHA for 862a556

• missing #includes and fwd decls

  evaleev

  committed
  0892d88

  Copy full SHA for 0892d88

• typos

  evaleev

  committed
  475fad8

  Copy full SHA for 475fad8

Commits on Jan 3, 2025

• introduced ConvolutionND::lattice_summed() and SeparatedConvolution::lattice_summed(), the latter assumes lattice summation applies to every term

  Show description for 9e3e08e
  evaleev

  committed
  9e3e08e

  Copy full SHA for 9e3e08e

• Convolution1D: support lattice summation for range-restricted kernels

  evaleev

  committed
  88602d9

  Copy full SHA for 88602d9

• cleanup FunctionImpl::do_apply to make processing multiple sets of displacements possible

  evaleev

  committed
  db30822

  Copy full SHA for db30822

Commits on Jan 5, 2025

• Displacements minor renames before adding range-related stuff

  evaleev

  committed
  42999a1

  Copy full SHA for 42999a1

• introduced KernelRange

  Show description for 0f1e502
  evaleev

  committed
  0f1e502

  Copy full SHA for 0f1e502

Commits on Jan 8, 2025

• replaced kernel range specification throughout by KernelRange

  evaleev

  committed
  d27b3b8

  Copy full SHA for d27b3b8

• test rnlp for erf-restricted Gaussian kernel

  evaleev

  committed
  b014f88

  Copy full SHA for b014f88

• introduced Key::operator[] to access translation's components

  evaleev

  committed
  73b121f

  Copy full SHA for 73b121f

• typo

  evaleev

  committed
  a1c4a35

  Copy full SHA for a1c4a35

• Key::neighbor variant that accepts Vector of Translations

  evaleev

  committed
  dc791f1

  Copy full SHA for dc791f1

• SeparatedConvolution::get_range accessor returns reference to range, rather than a copy

  evaleev

  committed
  fc281e8

  Copy full SHA for fc281e8

• FunctionImpl::do_apply: max distsq returned as std::optional<uint64> to support the case of zero displacements processed

  evaleev

  committed
  77ecb84

  Copy full SHA for 77ecb84

• introduce draft BoxSurfaceDisplacementRange that iterates over finite-thickness layer of displacements at the range boundary

  Show description for 12a6dd1
  evaleev

  committed
  12a6dd1

  Copy full SHA for 12a6dd1

• FunctionImpl::do_apply: for range restricted kernels include range boundary displacements

  Show description for 7caa96b
  evaleev

  committed
  7caa96b

  Copy full SHA for 7caa96b

• KernelRange: conversion to bool is explicit + introduced equality comparison

  evaleev

  committed
  ae47693

  Copy full SHA for ae47693

• BoundaryConditions::make_{range,range_vector} cleanup

  evaleev

  committed
  e613c4b

  Copy full SHA for e613c4b

• BoxSurfaceDisplacementRange: initial support for NDIM>1, not yes tested

  evaleev

  committed
  963a371

  Copy full SHA for 963a371

• FunctionImpl::do_apply supports NDIM>1 for range-restricted kernels; for NDIM>=3 limit the range surface contributions to n<=5 to avoid excessive costs

  evaleev

  committed
  07d6a3f

  Copy full SHA for 07d6a3f

• BoxSurfaceDisplacementRange filter uses not only displacement and destination point to skip unproductive displacements inside the range generator

  evaleev

  committed
  2236bd5

  Copy full SHA for 2236bd5

Commits on Jan 10, 2025

• fixed const-correctness issue in one SeparatedConvolution ctor

  evaleev

  committed
  bce231c

  Copy full SHA for bce231c

Commits on Jan 11, 2025

• constexpr constants in constant.h

  evaleev

  committed
  33ea2fb

  Copy full SHA for 33ea2fb

• constexpr Vector

  Show description for 38c9f91
  evaleev

  committed
  38c9f91

  Copy full SHA for 38c9f91

• KernelRange(N, sigma) uses hard range restrictor if sigma=0

  evaleev

  committed
  7daceb8

  Copy full SHA for 7daceb8

Commits on Jan 15, 2025

• muopxv_fast way inaccurate when used with operator blocks with medium-magnitude blocks, due to excessively-loose approximation of T

  Show description for 3d2555d
  evaleev

  committed
  3d2555d

  Copy full SHA for 3d2555d

Commits on Jan 16, 2025

• BoundaryConditions::make_range makes a hard range limit by default

  evaleev

  committed
  890c7ee

  Copy full SHA for 890c7ee

• sepop: added code used to test range-restricted diffuse Gaussian convolutions

  evaleev

  committed
  d5012a8

  Copy full SHA for d5012a8

• [ci] bump Linux jobs to use Ubuntu 24.04 and gcc 14

  Show description for b291c4e
  evaleev

  committed
  b291c4e

  Copy full SHA for b291c4e

• [ci] bump Linux jobs to use Ubuntu 24.04 and gcc 13

  Show description for 1e88600
  evaleev

  committed
  1e88600

  Copy full SHA for 1e88600

Commits on Jan 17, 2025

• introduced variant of Key<N>::distsq_bc that allows to mask dimensions

  evaleev

  committed
  bdc8b0a

  Copy full SHA for bdc8b0a

• introduced Displacements::get_disp() that allows to access standard (non-periodic) displacements without the need for superfluous arguments

  evaleev

  committed
  e4a10a6

  Copy full SHA for e4a10a6

• use sparse set of displacements to the range boundary for range-restricted kernels

  evaleev

  committed
  45ad7b5

  Copy full SHA for 45ad7b5

• testpercoul: test range-restricted periodic Coulomb operator

  evaleev

  committed
  b7b770b

  Copy full SHA for b7b770b

Commits on Jan 18, 2025

• testpercoul: cleanup + conversion to unit test

  evaleev

  committed
  8ba7aef

  Copy full SHA for 8ba7aef

Commits on Jan 20, 2025

• FunctionFactbory::special_points takes a vector as an lvalue ref

  evaleev

  committed
  07e633d

  Copy full SHA for 07e633d

• SeparatedConvolution ctors taking a vector of convolutions do not need to accept lattice_summed flag as it's recomputed from the constituent convolutions

  evaleev

  committed
  a54aeec

  Copy full SHA for a54aeec

• testgconv: test range-restricted kernels

  Show description for d2ae07e
  evaleev

  committed
  d2ae07e

  Copy full SHA for d2ae07e

• support surface dimensions for range-restricted kernels with mixed boundary conditions

  evaleev

  committed
  27118dd

  Copy full SHA for 27118dd

Commits on Jan 22, 2025

• dox cleanup

  evaleev

  committed
  a7964e4

  Copy full SHA for a7964e4

Commits on Jan 23, 2025

• Merge branch 'master' into jm/cleanup/nuclear-density-functor

  evaleev

  committed
  3204111

  Copy full SHA for 3204111

• inner(Function<R>,Function<T>) has optimizations that do not work for R!=T, constexpr avoid those paths

  evaleev

  committed
  def36ef

  Copy full SHA for def36ef

• restored BoxSurfaceDisplacementRange that was removed in 45ad7b5 , it provides probing_displacement method to enable screening out the entire surface contribution

  evaleev

  committed
  37ba7b9

  Copy full SHA for 37ba7b9

• FunctionImpl::do_apply: relax shell screening logic by counting displacement as "used" if it survived estimated AND actial norms of its contribution

  evaleev

  committed
  f1d3e5b

  Copy full SHA for f1d3e5b

• FunctionImpl::do_apply: range-restricted kernels need displacements to the entire surface of the boundary range, rather than to near the facet centers

  evaleev

  committed
  78f67fa

  Copy full SHA for 78f67fa

Commits on Jan 24, 2025

• FunctionFunctorInterface::special_level() is const

  evaleev

  committed
  97f156c

  Copy full SHA for 97f156c

• FunctionImpl: set special_level to the maximum of that provided by the factory and the functor

  evaleev

  committed
  f75d0a1

  Copy full SHA for f75d0a1

• fixes to stop compiler warnings about inconsistent use of override in FunctionFunctorInterface children

  evaleev

  committed
  11fb3dc

  Copy full SHA for 11fb3dc

• mw.h: evaluates scaling functions, useful for debugging

  evaleev

  committed
  02205a0

  Copy full SHA for 02205a0

• BoxSurfaceDisplacementRange: fixes for NDIM>1, testpercoul produces periodic potential for the tight (nuclear-sized) densities

  evaleev

  committed
  9daeb6e

  Copy full SHA for 9daeb6e

• SeparatedConvolution: take vector of shared_ptr to Convolution1D by const ref

  evaleev

  committed
  b059077

  Copy full SHA for b059077

• SeparatedConvolution::get_ops useful for accessing individual terms for testing/logging purposes

  evaleev

  committed
  55682b0

  Copy full SHA for 55682b0

• Vector: ctor/assignment taking Vector<Q> (Q!=T) are implemented using std::copy to invoke proper conversion Q->T

  evaleev

  committed
  feb38f4

  Copy full SHA for feb38f4

• displacements.h: removedCombinationGenerator and BoxSurfaceDisplacementRangeV2 which are no longer used

  evaleev

  committed
  1286e00

  Copy full SHA for 1286e00

• BoxSurfaceDisplacementRange: to detect that the entire surface layer is out of range filter function generalized to take Key pattern instead of Key; displacement is passed as an optional

  evaleev

  committed
  63a3c85

  Copy full SHA for 63a3c85

• dox++

  evaleev

  committed
  88e0c20

  Copy full SHA for 88e0c20