Incompatibility with latest version of numpy and matplotlib

[TobiasDuswald]

After installing abcpy manually (not via pip), I encountered issues when running the examples. Specifically, I had the latest versions of numpy (1.25.2) and matplotlib (3.7.2) installed.

Running the examples (mpirun -np 10 --bind-to core python backends/mpi/pmcabc_gaussian.py), I obtain the following error messages:

[...]
INFO:abcpy.inferences:Calculating acceptances threshold
INFO:abcpy.inferences:Calculating weights
INFO:abcpy.inferences:Calculate weights
INFO:abcpy.inferences:Calculating covariance matrix
Traceback (most recent call last):
  File "/home/testuser/abcpy/examples/backends/mpi/pmcabc_gaussian.py", line 118, in <module>
    journal = infer_parameters(backend, logging_level=logging.INFO)
  File "/home/testuser/abcpy/examples/backends/mpi/pmcabc_gaussian.py", line 73, in infer_parameters
    journal = sampler.sample([height_obs], steps, eps_arr, n_sample, n_samples_per_param, epsilon_percentile)
  File "/home/testuser/.pyenv/versions/3.9.1/lib/python3.9/site-packages/abcpy/inferences.py", line 933, in sample
    new_cov_mats = self.kernel.calculate_cov(self.accepted_parameters_manager)
  File "/home/testuser/.pyenv/versions/3.9.1/lib/python3.9/site-packages/abcpy/perturbationkernel.py", line 144, in calculate_cov
    all_covs.append(kernel.calculate_cov(accepted_parameters_manager, kernel_index))
  File "/home/testuser/.pyenv/versions/3.9.1/lib/python3.9/site-packages/abcpy/perturbationkernel.py", line 266, in calculate_cov
    (np.float, np.float32, np.float64, np.int, np.int32, np.int64)):
  File "/home/testuser/.pyenv/versions/3.9.1/lib/python3.9/site-packages/numpy/__init__.py", line 319, in __getattr__
    raise AttributeError(__former_attrs__[attr])
AttributeError: module 'numpy' has no attribute 'float'.
`np.float` was a deprecated alias for the builtin `float`. To avoid this error in existing code, use `float` by itself. Doing this will not modify any behavior and is safe. If you specifically wanted the numpy scalar type, use `np.float64` here.
The aliases was originally deprecated in NumPy 1.20; for more details and guidance see the original release note at:
    https://numpy.org/devdocs/release/1.20.0-notes.html#deprecations
--------------------------------------------------------------------------
Primary job  terminated normally, but 1 process returned
a non-zero exit code. Per user-direction, the job has been aborted.
--------------------------------------------------------------------------

and

[...]
[0.00317671]
 [0.00472121]
 [0.00478266]]
{'mu': 177.04224197656742, 'sigma': 14.735985720015503}
(array([[185.31015072,  -1.10226865],
       [ -1.10226865,  33.72395735]]), dict_keys(['mu', 'sigma']))
{'type_model': ['Normal'], 'type_dist_func': ['LogReg'], 'steps': 3, 'epsilon_init': array([0.75]), 'n_samples': 250, 'n_samples_per_param': 10, 'epsilon_percentile': 10, 'covFactor': 2, 'full_output': 0, 'epsilon_arr': [0.75, 0.7402597402597402, 0.7402597402597402]}
Traceback (most recent call last):
  File "/home/testuser/abcpy/examples/backends/mpi/pmcabc_gaussian.py", line 119, in <module>
    analyse_journal(journal)
  File "/home/testuser/abcpy/examples/backends/mpi/pmcabc_gaussian.py", line 91, in analyse_journal
    journal.plot_posterior_distr(path_to_save="posterior.png")
  File "/home/testuser/.pyenv/versions/3.9.1/lib/python3.9/site-packages/abcpy/output.py", line 847, in plot_posterior_distr
    fig, axes = scatterplot_matrix(datat, meanpost, parameters_to_show,
  File "/home/testuser/.pyenv/versions/3.9.1/lib/python3.9/site-packages/abcpy/output.py", line 663, in scatterplot_matrix
    if ax.is_first_col():
AttributeError: 'Axes' object has no attribute 'is_first_col'
--------------------------------------------------------------------------
Primary job  terminated normally, but 1 process returned
a non-zero exit code. Per user-direction, the job has been aborted.
--------------------------------------------------------------------------

Downgrading the packages as

• numpy from 1.25.2 to 1.23.5
• matplotlib from 3.7.2 to 3.5.3
  fixes the problems.

Are there any plans updating the repository to account for the API changes of these major dependencies?

[LoryPack]

Hi, thank you very much for flagging this. We'll work on this and update the code base!