dylanljones / exactdiag

Exact diagonalization of fermionic (many-body) systems

Hi, I'm Dylan 👋

Currently pursuing a PhD in strongly-correlated condensed matter physics at the University of Augsburg. Interested in computational physics, quantum computing, machine learning and programming in general.

Projects

Computational physics

• lattpy - Simple and efficient Python package for modeling d-dimensional Bravais lattices in solid state physics.
• exactdiag - Exact diagonalization of fermionic many-body systems.

Music

• pyrekordbox - Inofficial Python package for interacting with the database and other files (XML, ANLZ, MySettings) of Pioneers Rekordbox DJ software.

Other

• ftmplt - Simple string parsing and formatting using Python's f-string templates.
• mplstyles - Collection of scientific Matplotlib styles.

exactdiag's Issues

Green's function using Lehmann sum does not return the expected result for U=0 and N>2

Describe the bug
The Lehmann Green's function does not match the non-interacting GF for U=0 and bulk sites (not 0 or N-1)

To Reproduce
Steps to reproduce the behavior:

import numpy as np
import matplotlib.pyplot as plt
import exactdiag as ed


def greens0(model, z):
    nsites = model.num_sites
    ham = np.zeros((nsites, nsites), dtype=np.float64)
    np.fill_diagonal(ham, model.eps)
    for i, j in model.neighbors:
        ham[i, j] = ham[j, i] = model.hop

    return ed.gf0_pole(ham, z=z, mode="full")


model = ed.models.HubbardModel.chain(3, inter=0.0, beta=200.0).hf()
z = np.linspace(-5, +5, 1001) + 1e-1j

gfmat = ed.greens.compute_gf_full(model, z)
gfmat0 = greens0(model, z)

fig, ax = plt.subplots()
ax.plot(z.real, -gfmat0[:, 1, 1].imag)
ax.plot(z.real, -gfmat[:, 1, 1].imag, ls="--")
plt.show()

Expected behavior
The Green's function should match the non-interacting GF

Numpy/Scipy import error

Running the tests in a GitHub Workflow on Windows using Python 3.10 results in Numpy/Scipy import errors:

ImportError while importing test module 'D:\a\exactdiag\exactdiag\tests\test_basis.py'.
Hint: make sure your test modules/packages have valid Python names.
Traceback:
C:\hostedtoolcache\windows\Python\3.10.7\x64\lib\importlib\__init__.py:1[26](https://github.com/dylanljones/exactdiag/actions/runs/3226005826/jobs/5279006230#step:6:27): in import_module
    return _bootstrap._gcd_import(name[level:], package, level)
tests\test_basis.py:11: in <module>
    from exactdiag import basis
exactdiag\__init__.py:[27](https://github.com/dylanljones/exactdiag/actions/runs/3226005826/jobs/5279006230#step:6:28): in <module>
    from .matrix import (
exactdiag\matrix.py:11: in <module>
    from scipy.sparse import spmatrix
C:\hostedtoolcache\windows\Python\3.10.7\x64\lib\site-packages\scipy\sparse\__init__.py:267: in <module>
    from ._csr import *
C:\hostedtoolcache\windows\Python\3.10.7\x64\lib\site-packages\scipy\sparse\_csr.py:10: in <module>
    from ._sparsetools import (csr_tocsc, csr_tobsr, csr_count_blocks,
E   ImportError: numpy.core.multiarray failed to import
------------------------------- Captured stderr -------------------------------
RuntimeError: module compiled against API version 0xf but this version of numpy is 0xe

Running the tests local with the same dependencies from the requirements.txt file works as expected for Python versions 3.7-3.10