pyqcm-skilllisted
Install: claude install-skill pyqcm-project/pyqcm-skill
# pyqcm
Pyqcm implements three related quantum cluster methods, Cluster Perturbation Theory (CPT), the
Variational Cluster Approach (VCA), and Cellular/Cluster Dynamical Mean Field Theory (CDMFT), for
approximating strongly correlated (Hubbard-like) lattice models. The impurity solver is exact
diagonalization (Lanczos and variants) on sparse matrices. Core numerics are C++; the day-to-day
interface is Python.
## First: locate the pyqcm source tree
This skill ships no pyqcm source. Nearly everything below refers to a **pyqcm checkout**, written
`$PYQCM_ROOT`. Resolve it before doing anything else, in this order:
1. `$PYQCM_ROOT`, if the user has it set.
2. Derive it from an installed package:
`python3 -c "import pyqcm, pathlib; print(pathlib.Path(pyqcm.__file__).resolve().parent.parent)"`,
then **confirm `src_ed/` exists there**. A pip-installed pyqcm ships only the Python package plus
the compiled extension, so this check fails for wheel installs and that is the expected outcome,
not an error.
3. Ask the user for the path.
If no checkout is reachable, say so plainly and carry on: the scripting and physics jobs still work
from the reference files alone. Do not hand out paths under `$PYQCM_ROOT` that you have not confirmed
exist.
The checkout is also the install source. pyqcm on PyPI is **sdist-only**, so every install compiles
from source and needs CMake, a C++ compiler, and BLAS regardless of how it is invoked.
## The four jobs
Work out which one the user