Changelog¶
v 0.4.0 (in progress)¶
- Added
slater.tressian_dot method returning third derivatives pre-contracted with the backflow Jacobian, and its documentation in slater.rst
Quick start guide and MPI parallelisation page in documentation
Documented all previously undocumented input keywords (GAUTOL, OPT_DTVMC, OPT_MAXITER, OPT_NOCTF_CYCLES, POSTFIT_VMC, OPT_FIXNL, VM_SMOOTH_LIMIT, VM_W_MAX, VM_W_MIN, EMIN_XI_VALUE, DMC_METHOD, LIMDMC, ALIMIT, NUCLEUS_GF_MODS, EBEST_AV_WINDOW, BACKFLOW)
drift_velocity and energy_parameters_d1 sections in wfn.rst
OPT_ORBITALS and USE_TMOVE keywords in config.rst
DMC tutorial page (dmc.rst): drift-diffusion, velocity limiting, branching, population control, T-move, load balancing
- Fixes
Input keywords parsed with wrong type: VM_W_MAX, VM_W_MIN, EMIN_XI_VALUE read as booleans instead of floats, NON_LOCAL_GRID as boolean instead of integer; LIMDMC misspelled as “limdmct” so the keyword was silently ignored
Misleading NUMBA_NUM_THREADS advice in parallel.rst: JIT kernels are single-threaded
Cusp initialisation description in cusp.rst: Cusp is created by CuspFactory, not by Slater
Missing underscores in DMC_EQUIL_NBLOCK and DMC_STATS_NBLOCK keyword names in config.rst
Third derivatives of Slater orbitals: yyy and zzz components of tressian_matrix used the x-component of the angular gradient, and tressian doubled one ordering of the triple-product trace instead of summing both (verified against numerical derivatives on Ne)
False trace-reversal identity in tressian snippets of slater.rst
Broken Numba documentation link in index.rst
Incorrect output description for kinetic_energy_parameters_d1 in wfn.rst
Missing backflow-parameters formula in kinetic_energy_parameters_d1 section of wfn.rst
Incorrect output description for nonlocal_energy_parameters_d1 and missing sign in kinetic energy derivative in wfn.rst
Methods summary in wfn.rst attributed to Slater class instead of Wfn
Duplicate Sphinx anchor
intermediate databetween jastrow.rst and backflow.rstDuplicate Sphinx anchor
valuebetween wfn.rst and slater.rstTypo “Input parameter are supported” → “Input parameters supported” in config.rst
- Improved
Speed up backflow optimization: kinetic_energy_parameters_d1 uses slater.tressian_dot to avoid materializing the (3N)³ tressian tensor, vectorized bf_d1 loop and fused reduction loop in backflow_laplacian_parameters_d1
Cusp correction page expanded with physical motivation, correction scheme, and API reference
v 0.3.0 (2025-10-30)¶
- Added
Documentation on https://casinoqmc.readthedocs.io/en/latest/
- Improved
Improve Jastrow speed by 20%
scipy.optimize.least_squares optimization in one thread
Add tests
- Fixes
Remove
np.float_as it not supported in Numpy 2.XFix reweighted VMC variance optimization.
v 0.2.0 (2024-10-30)¶
- Added
Pseudopotential for VMC and DMC (with T-move)
- Improved
Use Horner Scheme in Jastrow term
Electron-by-Electron (EBES) sampling in DMC
Use structrefs instead of jitclasses
- Fixes
System with only up- or only down- electrons (i.e H-atom)
v 0.1.0 (2023-10-30)¶
- Added
Orbital file formats: gwfn.data, stowfn.data
Multi-determinant expansions
3-term Jastrow factor and 3-term backflow transformation
Configuration-by-configuration (CBCS) sampling
Partial cusp correction according to A. Ma, M. D. Towler, N. D. Drummond, and R. J. Needs, without rc optimization
MPI parallelization
VMC and DMC energy calculation, varmin and emin optimization