Sync BMD Compute pages with Compute v1.0.0 methodology
- Advanced Options: replace "SOC under validation" with the v1 automatic SOC
behaviour; note automatic spin and DFT-D3(BJ) triggers; drop stale SOC
parenthetical in DFT+U.
- Workflows: state v1 stage chaining, k-point sampling, preview vs run-time
differences and the one-allocation/no-continuation model; remove the
"Not yet specified" list.
- Using BMD Compute: clarify that Resume does not continue a job.
Source: bmd_compute v1.0.0 (a746155) docs/methodology.md sections 1-4, 6-8.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DHgs1a4r3pZ6JVJ5dXQvoR
55d09db
Updated BMDwiki (markdown)
deb4a2c
docs: add quick routes to Home and reduce duplication
Open Home with four "Where do I go?" routes (Using BMD Compute; BMD
Agent; Computing foundations; Research methods), then three compact
groups: BMD Compute reference, Computing foundations (Tutorials 1-11
with English/Hebrew links, plus running calculations yourself) and
Research methods (by topic, incl. research practice). Each page is
linked once: 10 tables -> 3, 36 links -> 33 unique, no page dropped.
Home.md only.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DHgs1a4r3pZ6JVJ5dXQvoR
d7eec1c
docs: reorganise Home around a BMD-first learning pathway
- Home: replace the historical Tutorial 1-20 list with sections:
Start here (Using BMD Compute, Workflows, INCARs, Advanced Options;
BMD Agent and VASP fundamentals as optional background); Next steps
(Tutorials 1-11 kept numbered with English/Hebrew links, grouped as
Linux / Remote computing and POWER / Python and pymatgen, plus
running and extending calculations yourself); Research methods and
specialist tools by topic (SQS; USPEX incl. Hebrew page, XtalOpt;
high-throughput, Materials Project); Research practice. States that
sections are suggestions, not prerequisites.
- Former Tutorials 12-20: replace obsolete "Tutorial N" navigation
lines with page-name or return-to-Home links (including the stale
USPEX "Tutorial 8" and High-throughput "tutorial 9/14" references).
No page renames, URL changes or scientific content changes.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DHgs1a4r3pZ6JVJ5dXQvoR
0cf6ccd
docs: finalize access and resources wording on Using BMD Compute
- Step 0: replace the maintainer placeholder with the student-facing
address (TAU network/VPN, http://132.66.66.90:8000/, no personal
login) and simplify the shared-infrastructure note.
- Step 3: describe Execution Resources as the requested computational
resources that new users can normally leave unchanged unless a
maintainer or project workflow specifies otherwise.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DHgs1a4r3pZ6JVJ5dXQvoR
35fffdf
docs: add beginner-facing Using BMD Compute page
First-use walk-through of the current BMD Compute interface
(bmd_compute c241a64): structure input, Structure Summary and Method
Considerations, Desired Output, preview of generated inputs, Prepare
Remote vs Submit Calculation, Monitoring states, Resume by Job ID,
Load Results, when to use BMD Agent, and where to go next. No Linux,
SSH, SLURM or Python prerequisites. Access address/onboarding marked
as needing maintainer information.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DHgs1a4r3pZ6JVJ5dXQvoR
a5ab180
docs: remove redundant page-title H1s
Remove the leading H1 (and the blank line after it) from the 16 pages
where it only repeats the GitHub Wiki filename-derived page title,
including decorative emoji. Pages whose H1 adds information (Home,
VASP, band structure, high-throughput, USPEX, XtalOpt, SQS, Reading &
Editing, PyMatGen, Tutorial-Y/YY, Hebrew pages) are unchanged.
No other content, code blocks, links or anchors changed.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DHgs1a4r3pZ6JVJ5dXQvoR
2fc931c
docs: correct attribution of Materials Project U values
The MP U values were fitted by the Materials Project to experimental
binary formation enthalpies following the approach of Wang, Maxisch and
Ceder (2006); they were not fitted by Wang et al. to oxidation
enthalpies. Wang 2006 remains the methodological reference.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DHgs1a4r3pZ6JVJ5dXQvoR
9688f49
docs: document BMD Compute automatic DFT+U v1 in Advanced Options
- Advanced Options: rewrite the DFT+U section to describe the adopted
BMD Compute automatic DFT+U behaviour (bmd_compute c241a64): the
Materials Project/pymatgen oxide/fluoride rule and U values (Dudarev,
J = 0), the conservative BMD d0 screen with fallback to the MP rule
when no oxidation-state guess exists, representative examples, NiO
worked example, stage placement per Desired Output (PBE stages only,
never HSE06), independence from spin/SOC/dispersion, Custom workflow
control, and why BMD energies are not strictly MP-compatible. Cite
Anisimov 1991, Dudarev 1998, Wang 2006, Jain 2011 and MP docs.
- Intro, INCARs, VASP and Workflows: stop describing DFT+U as purely
optional; point to the automatic behaviour.
SOC section, reference INCARs and POTCAR path unchanged.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DHgs1a4r3pZ6JVJ5dXQvoR
2be8069
Merge branch 'docs/separate-teaching-examples-from-compute-policy'
da488c9
docs: apply review corrections (Pulay formula, NCORE/NPAR, provenance)
- Workflows: remove the explicit Pulay-stress formula; keep the
qualitative finite-basis and repeated-relaxation explanation.
- VASP: explain that NCORE and NPAR parameterize the band parallel
decomposition differently; prefer NCORE and do not set both.
- INCARs: provenance states generation from BMD Compute on
26 September 2026; exact Compute/dependency versions not retained.
Reference INCARs, Band NCORE omission and POTCAR path unchanged.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DHgs1a4r3pZ6JVJ5dXQvoR
0d483ca
docs: record Si as the reference structure for BMD Compute INCARs
Compute commit and pymatgen/atomate2 versions remain unspecified.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DHgs1a4r3pZ6JVJ5dXQvoR
15c3c05
docs: use complete BMD Compute workflow INCARs as references
- INCARs: replace the isolated references with the complete, correctly
numbered stage INCARs of the four standard workflows (Static Energy;
Relaxed Structure; Density of States; Band Structure), reproduced
exactly. Add HSE06 Static Energy. Comparisons now: PBE GO vs PBE
Static; PBE vs HSE06 Static; HSE06 Static vs DOS vs Band (TIME
operative only with ALGO = Damped). Reference structure marked
"to be confirmed".
- Workflows: rename Geometry Optimisation to Relaxed Structure (two
identical PBE stages, Stage 2 from Stage 1 structure); separate
"BMD workflow facts" from general physics; Pulay stress presented as
one reason for repeated relaxation, not BMD's stated rationale;
remove "HSE06@PBE"; file hand-offs and KPOINTS left unspecified.
- Advanced Options: drop Si-specific wording for the base references.
Supplied INCAR settings, SOC status and POTCAR path unchanged.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DHgs1a4r3pZ6JVJ5dXQvoR
502edb0
docs: organise BMD Compute Workflows around the four desired outputs
- Workflows: rebuild around the supplied workflows (Static Energy;
Geometry Optimisation x2; PBE relax -> HSE06 static -> HSE06 DOS;
PBE relax -> HSE06 static -> HSE06 band structure). For each: the
scientific result, why prerequisite stages are needed, and the
conceptual hand-off. Substantive Pulay-stress / repeated variable-
cell relaxation discussion without claiming it as the sole rationale;
HSE06@PBE caveat; DOS (BZ integral, uniform mesh) vs E_n(k) (path,
zero-weight approach). File-level hand-offs marked unspecified.
- INCARs: comparisons renamed "vs" (not sequences); remove the stale
list of other stages; point sequence questions to Workflows.
Reference INCAR blocks, SOC status and POTCAR path unchanged.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DHgs1a4r3pZ6JVJ5dXQvoR
f77fb7f
docs: add BMD Compute Workflows page and four-page structure
- VASP: expand into DFT fundamentals (Kohn-Sham SCF, exchange-
correlation and hybrids, plane-wave truncation and PAW, Brillouin-
zone sampling, occupations/smearing, forces and relaxation); FCC-Si
teaching example unchanged.
- BMD-Compute-Workflows (new): why calculations are staged, what can
pass between stages, relax->static rationale incl. Pulay stress,
repeated relaxation, PBE non-SCF vs HSE06 zero-weight approach;
BMD Compute-specific sequences and hand-offs marked "to be provided".
- BMD-Compute-INCARs: keep stage-level explanations only; move Pulay
and sequence reasoning to Workflows. Reference INCAR blocks unchanged.
- Advanced Options, Home: navigation for the four-page sequence.
POTCAR path and SOC status unchanged.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DHgs1a4r3pZ6JVJ5dXQvoR
d01e4cf
docs: tighten BMD Compute INCARs commentary
- Fix "four kinds" -> "five kinds" of settings.
- Soften Pulay-stress text; drop unsourced Si ENMAX value.
- Reword PBE/HSE06 band-gap claim.
- State that TIME is retained but not operative with ALGO = Normal.
- Explain LORBIT/NEDOS in the static stage without conflating it with
the dedicated DOS stage.
- Add provenance placeholders (Compute commit, pymatgen, atomate2).
Reference INCAR blocks unchanged.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DHgs1a4r3pZ6JVJ5dXQvoR
e530d07
docs: add BMD Compute INCAR teaching pages and fix Agent CLI
- VASP: rework into a fundamentals page (inputs, SCF/ionic loops,
ENCUT, k-points, smearing, EDIFF/EDIFFG, relaxation, restarts,
parallelisation); FCC-Si example kept and labelled as teaching.
- BMD-Compute-INCARs (new): the four supplied BMD Compute Si reference
INCARs reproduced exactly, with settings grouped by role and
Relax->Static, PBE->HSE06 and HSE06 DOS->Band comparisons.
- BMD-Compute-Advanced-Options (new): concept pages for spin, DFT+U,
dispersion, Gamma-only and ions-only; reference deltas pending;
SOC marked under validation.
- BMD-Agent: correct bmd-agent / JOB_ID / PATH semantics, document
--verbose, keep expert subcommands secondary; adopt agreed authority
wording.
- Manual HSE band-structure and USPEX pages: keep as teaching/manual
workflows and link to the reference page; Home links new pages.
POTCAR path unchanged pending verification.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DHgs1a4r3pZ6JVJ5dXQvoR
feed7c6
docs: separate teaching examples from BMD Compute policy
- VASP: label FCC-Si INCAR as a teaching example, not BMD Compute
production policy; fix Unicode minus in EDIFF; drop EDIFFG line
that has no effect with NSW = 0.
- HSE06 band structure: mark the workflow and INCAR settings as an
educational/manual example; BMD Compute takes precedence.
- USPEX: relabel "BMD's Recommended INCAR Files" as workflow examples.
- BMD Agent: make bmd-agent, bmd-agent JOB_ID and bmd-agent PATH the
primary interface, with expert subcommands under Advanced usage;
BMD Compute owns calculation methodology, BMDex supplies curated
supporting evidence and tools.
POTCAR path intentionally unchanged pending verification on POWER.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DHgs1a4r3pZ6JVJ5dXQvoR
6d31432
docs: recommend protected SSH keys
37e4b1e
docs: canonicalize BMDwiki wiki links
b565e34
docs: clarify BMD Agent purpose
16c6009
docs: add BMD Agent wiki page
3907cc0
Updated Calculating Band Structure in VASP (markdown)
4132269
Updated Navigating with Linux (markdown)
49c4b5f
Updated Calculating Band Structure in VASP (markdown)
ae90cb8
Updated Publishing results (markdown)
4e9838f
Updated High‐Throughput Computing (markdown)
d18756c
Updated The Materials Project (markdown)
3b47f52