Herbst, Michael F.
Levitt, Antoine
Cancès, Eric
2022-12-20
DFTK v0.6.0
<p><a href="https://github.com/JuliaMolSim/DFTK.jl/compare/v0.5.15...v0.6.0">Diff since v0.5.15</a></p>
<p>Note that this release increases the minor version and contains a number of breaking changes:</p>
<ul>
<li><p>The <code>self_consistent_field</code> and <code>potential_mixing</code> SCF functions now check convergence in the density by default. This means that <code>self_consistent_field(basis; tol=1e-6)</code> now converges until the change in density between subsequent steps is below <code>1e-6</code>. Before it used to converge until the change in energy was less than <code>tol</code>. To recover the old behaviour use <code>self_consistent_field(basis; is_converged=DFTK.ScfConvergenceEnergy(tol))</code>. (#800)</p>
</li>
<li><p>The <code>load_lattice</code>, <code>load_atoms</code> and <code>load_positions</code> functions are now deprecated and will be removed in DFTK 0.6.1. Users are advised to use <code>AtomsIO</code> for the purpose of loading structures from external files. See <a href="https://docs.dftk.org/v0.6/examples/atomsbase/">https://docs.dftk.org/v0.6/examples/atomsbase/</a> and <a href="https://github.com/mfherbst/AtomsIO.jl">https://github.com/mfherbst/AtomsIO.jl</a> for examples.</p>
</li>
<li><p>Instead of using NLSolve's Anderson acceleration we have switched to a built-in implementation, which we found more stable in practice. (#791)
The NLSolve-been approach has been removed completely.</p>
</li>
<li><p>The deprecated <code>G_to_r</code> and <code>r_to_G</code> functions for FFTs have now been removed in favour of <code>ifft</code>, <code>fft</code> and similar. (#804)</p>
</li>
<li><p>The deprecated <code>n_bands</code> and <code>n_ep_extra</code> arguments from <code>self_consistent_field</code> have now been removed. (#804)</p>
</li>
</ul>
<p><strong>Closed issues:</strong></p>
<ul>
<li>Struct for atoms (#118)</li>
<li>GPU discussion (#350)</li>
<li>SCF default tolerance should be on the density (#780)</li>
<li>Run GPU CI only after activation (#803)</li>
</ul>
<p><strong>Merged pull requests:</strong></p>
<ul>
<li>Bugfix for computations of <code>δψ</code> and supercells (#775) (@epolack)</li>
<li>Switch default scf solver to builtin Anderson (#791) (@epolack)</li>
<li>Add developper documentation for GPU computations (#793) (@GVigne)</li>
<li>Correct PW parameterization comment (#795) (@epolack)</li>
<li>Add GPU CI via gitlab (#796) (@mfherbst)</li>
<li>Switch to density tolerance by default (#800) (@mfherbst)</li>
<li>Make cigpu badge a link to ci pipelines in gitlab (#801) (@carstenbauer)</li>
<li>Rely on AtomsIO for structure parsing (#802) (@mfherbst)</li>
<li>Remove deprecated features (#804) (@mfherbst)</li>
</ul>
Cite this paper whenever you use DFTK.
https://doi.org/10.5281/zenodo.7464563
oai:zenodo.org:7464563
Zenodo
https://github.com/JuliaMolSim/DFTK.jl/tree/v0.6.0
https://doi.org/10.5281/zenodo.3541724
info:eu-repo/semantics/openAccess
MIT License
https://opensource.org/licenses/MIT
DFTK: The Density-functional toolkit
info:eu-repo/semantics/other