v0.7.7

DFTK v0.7.7

Diff since v0.7.6

Merged pull requests:

• CompatHelper: bump compat for GPUArraysCore to 0.2, (keep existing compat) (#1008) (@github-actions[bot])
• Start deprecating DFTK-bundled pseudopotential repository (#1047) (@mfherbst)

v0.7.6

DFTK v0.7.6

Diff since v0.7.5

Merged pull requests:

• Move from IterativeSolvers to KrylovKit (#1043) (@mfherbst)
• Automatically use rcut from PseudoFamily when possible (#1044) (@Technici4n)
• Calculator calls interpolate for near-matching density (#1045) (@mfherbst)
• Add phonon example (#1046) (@mfherbst)

Closed issues:

• Phonon example (#1040)

v0.7.5

DFTK v0.7.5

Diff since v0.7.4

Merged pull requests:

• Be more conservative when selecting initial number of bands (#1038) (@mfherbst)
• Expose supersampling in PlaneWaveBasis interface (#1041) (@mfherbst)
• Small updates to harmonise with new PseudoLibrary (#1042) (@mfherbst)

v0.7.4

DFTK v0.7.4

Diff since v0.7.3

Merged pull requests:

• Minor optimization of nonlocal term (#1036) (@abussy)
• Mention response and DFPT more prominently (#1039) (@mfherbst)

v0.7.3

DFTK v0.7.3

Diff since v0.7.2

Merged pull requests:

• Update to AtomsBase 0.5 (#1033) (@mfherbst)

Closed issues:

• Cannot import Materials Project CIF files (#1035)

v0.7.2

DFTK v0.7.2

Diff since v0.7.1

Merged pull requests:

• Speed-up symmetrize_ρ (#1025) (@abussy)
• Change DFTK thread count to a runtime setting (#1028) (@Technici4n)
• Enable force calculations on GPU runs (#1031) (@abussy)

Closed issues:

• Setting a >1 threading preference will trigger a recompilation and make it fail (#1012)

v0.7.1

DFTK v0.7.1

Diff since v0.7.0

Merged pull requests:

• No longer allow high-level model construction without pseudos (#1030) (@mfherbst)

v0.7.0

DFTK v0.7.0

Diff since v0.6.20

Note that this release increases the minor version and contains a number of breaking changes:

• SCF solver implementations have changed their API from ρout = F(ρin) to ρout, info_out = F(ρin, info_in). This means that many of DFTK's learning resources (which use this feature to introduce the concepts of basic SCF algorithms) need to be adapted for DFTK 0.7. (#811)

• The model_LDA, model_PBE and model_SCAN functions are now deprecated and will be removed in a later DFTK version. Use model_DFT(system; functionals=LDA(), kwargs...) or similar (functionals=PBE(), functionals=SCAN()) (#994).

• The model_DFT syntax with an array of functionals in the last argument is deprecated and will be removed in a later DFTK version. Instead of model_DFT(system, [:lda_x]; kwargs...) use model_DFT(system; functionals=[:lda_x], kwargs...) (#994)

• The model constructors in combination with AtomsBase systems no longer respect the :pseudopotential atom keys. To specify the pseudopotentials in combination with the AtomsBase interface now use the pseudopotentials keyword argument of the model constructors. E.g. instead of running system = attach_psp(system; Si="hgh/lda/si-q4"); model = model_DFT(system; kwargs...) use model = model_DFT(system; pseudopotentials=Dict(:Si => "hgh/lda/si-q4")). Note that the new PseudoPotentialData.PseudoFamily objects can also be passed as values to the pseudopotentials keywords, which is often simpler. (#1020)

• The attach_psp function has been removed due to the above change. Use the new model_DFT(system; pseudopotential=Dict(:Si => "hgh/lda/si-q4"), kwargs...) syntax instead. (#1020)

• The ElementPsp(key; psp) constructor is deprecated and will be removed in a later DFTK version, use instead Element(key, psp) which is now more powerful (psp can be a PspHgh, PspUpf or PseudoPotentialData.PseudoFamily). (#1020)

• The deprecated kgrid_from_minimal_spacing function has been removed in favour of kgrid_from_maximal_spacing (#994)

• The deprecated ScfDiagtol function has been removed. Use AdaptiveDiagtol now (#994)

Merged pull requests:

• Change internal fixed-point API to (x, info) = f(x, info) to simplify SCF metadata tracking (#811) (@epolack)
• Adaptive-depth anderson (#939) (@mfherbst)
• Add DFT book to resources (#991) (@mfherbst)
• Solutions to discretisation exercises (#993) (@dianetambey)
• Drop deprecated features and a round of docs updating (#994) (@mfherbst)
• Minor refactoring of the FFTs (#996) (@abussy)
• Enhance user experience with MPI (#997) (@abussy)
• MPI workarounds for aarch64 (#999) (@abussy)
• CompatHelper: bump compat for JLD2 in [weakdeps] to 0.5, (keep existing compat) (#1000) (@github-actions[bot])
• Update literature (#1001) (@mfherbst)
• Reupload the projected density of states (#1002) (@xuequan818)
• Update README.md (#1004) (@antoine-levitt)
• Make solve_ΩplusK MPI-compatible (#1005) (@Technici4n)
• Optimize apply_K (#1006) (@Technici4n)
• Use latest pre-release version instead of nightly for CI against upcoming versions (#1010) (@Technici4n)
• ForwardDiff for response derivatives with PspUpf pseudopotentials (#1013) (@niklasschmitz)
• Fix typo in self_consistent_field documentation (#1015) (@Technici4n)
• Refinement of density and forces (#1016) (@Technici4n)
• Bump codecov/codecov-action from 4 to 5 (#1018) (@dependabot[bot])
• Remove dependency on DftFunctionals.change_parameters fallback behavior for functionals without parameters (#1019) (@niklasschmitz)
• Use PseudoPotentialData.jl instead of Artifacts (#1020) (@mfherbst)
• Fix Fermi-Dirac occupation overflow and remove custom ForwardDiff rule (#514) (#1022) (@niklasschmitz)
• Speed up AtomicLocal forces (#1024) (@abussy)
• Fix #1026: Allow GPU arrays in FFTGrid (#1029) (@Technici4n)

Closed issues:

• Fix AD for small occupations (#514)
• Typo in self_consistent_field documentation (#1003)
• DFTK broken on GPUs (#1026)

v0.6.20

DFTK v0.6.20

Diff since v0.6.19

Merged pull requests:

• Virial interface for AtomsCalculator (#959) (@CedricTravelletti)
• Calculator State (#961) (@CedricTravelletti)
• Add a preference option for DFTK threading (#972) (@antoine-levitt)
• Ignore optim_res when saving scfres (#973) (@antoine-levitt)
• Improve performance of stress calculations (#978) (@mfherbst)
• Use the Simpson rule for integration of local part of PSPs (#982) (@abussy)
• fix typo (#983) (@chillenb)
• Update chat url to zulip (#984) (@mfherbst)
• Updating TestItemRunner (#989) (@epolack)
• Update Hartree potential notation in introductory docs (#990) (@niklasschmitz)

Closed issues:

• Stresses need an unreasonable amount of memory and runtime (#893)
• Incompatible with TestItemRunner 1.0 (#988)

v0.6.19

DFTK v0.6.19

Diff since v0.6.18

Merged pull requests:

• Crude import of some material from DFTK school (#941) (@mfherbst)
• Nonlocal term for dynamical matrix (#944) (@epolack)
• Option to specify timeout for SCFs (#948) (@mfherbst)
• bump compat for Spglib to 0.9.4 (#949) (@github-actions[bot])
• Update GTH pseudopotential database (#950) (@mkrack)
• Phonons docs (#953) (@antoine-levitt)
• Bump codecov/codecov-action from 3 to 4 (#955) (@dependabot[bot])
• Fix mpi test (#956) (@epolack)
• Atoms calculators: empty state by default. (#958) (@CedricTravelletti)
• Fix normalization of wavefunctions (#960) (@antoine-levitt)
• Docfix: Comment on spin index on density ρ (#962) (@niklasschmitz)
• Sort eigenvalues after LOBPCG (#964) (@antoine-levitt)
• Minor Correction to Docs: Update collinear_magnetism.jl (#965) (@JaydevSR)
• fixing typos (#969) (@epolack)
• Update PspUpf.jl : the range of l for r2_pswfcs (#970) (@tyjerome)
• Bump julia-actions/cache from 1 to 2 (#971) (@dependabot[bot])

Closed issues:

• Change of dimension in last kpoint when using non-converging scf (#963)