andrea-mitridate / ptarcade Goto Github PK

I test the code with a very simple model file where $\Omega_{\rm gw} h^2 = 0$

import numpy as np
from ptarcade.models_utils import prior

parameters = {
            'test' : prior("Uniform", 0 , 1)
            }

smbhb = False
name = "test"


def spectrum( f , test ):
    Omegah2 = 0
    return 0*f

and I run it with
ptarcade -m test_model.py

it runs quickly and I get the chain. The posterior is

which I believe does not reflect the uniform prior from 0 to 1. The issue is still here when I add Nsamp or burn a lot of samples in the beginning.

However when I change the prior from 1 to 2 the issue is gone:

So I suppose that something happened in the definition of the uniform prior?

Is your feature request related to a problem? Please describe.
Some models are bounded by BBN or CMB bounds on cosmological energy density. Having a switch to include these bounds in the prior function would be useful for constraining posteriors

Describe the solution you'd like

1. Create a utility function to calculate the integrated energy density across frequency for a model spectrum supplied with input parameters
2. Use that utility function in the log-prior function. If the integrated energy density is greater than a chosen bound, return -inf

Describe alternatives you've considered
We could just take the posterior and remove samples that result in energy densities greater than the bound? How would these two methods change the posterior?

Additional context
I'd like to compute the integrated energy density of the different spectra I'll be using to compare each other

Our full documentation is great, but we should add some more information, images, etc. to the README.

Attribute error for params in 'plot_utils.plot_posteriors'. I am using the conda environment method to run PTArcade.

Following the quick start section in from the arXiv documentation I can run the mcmc step but when I run the commands for plotting

p_utils.plot_posteriors([chain] , [params]) 

I get the following error

File ~/miniconda3/envs/ptarcade/lib/python3.10/site-packages/ptarcade/plot_utils.py:827, in <listcomp>(.0)
    824 N_chains = len(chains)
    826 priors = params
--> 827 params = [np.array(list(par.keys())) for par in params]
    829 if par_to_plot is None:
    830     par_to_plot = [None] * N_chains

AttributeError: 'numpy.ndarray' object has no attribute 'keys'

The chains seem to import no problem

after running

chain, params = c_utils.import_chains('./chains/np_model/') 

the output is

Starting import from ./chains/np_model/
Finished importing   chains/np_model in 0.13s

Perhaps I am doing something basic wrong, but I'm unsure.

Any help you could give would be greatly appreciated. Thanks so much for making this very exciting tool.

Cheers,
Andrew

Hello, I am sorry if this is a bit naive or silly. But I was wondering if PTArcade can be somehow made compatible with Windows, or can be made possible to be installed on Google Colab?

Before doing pip install, one needs to install Tempo2. And from what I gathered, it can not be installed on Windows. Due to such linux/ mac dependencies, I am unable to install PTArcade on Google colab.

Is it possible to have a version which works on windows too? Or if there's already a way to do it, I will be grateful for the help.

Also, this is the first time I am posting/ commenting on GitHub, so I am open to any suggestions/ feedback.

Thank you!

For NG15 the user can choose between the data with and without spatial correlations in Ceffyl mode. There is not such a choice for IPTA2 and NG12.

Solution

Do not error out. Capture the user's config, print a statement to STDOUT that we are falling back to the available mode for the given dataset.

Describe the bug
a) Is there a way to impose constraints on parameters using specific settings? For instance, assuming I have parameters
a and b with a Uniform prior ranging from [0,1], I'd like to enforce a condition where x≥y. This is reminiscent of the example found in Bilby's documentation.

b) The documentation provides a method for importing a spectrum:

log_spectrum = spec_importer(path)
def spectrum(f, p):
    return 10**log_spectrum(np.log10(f), p=p)

Given this, is it feasible to modify the log_spectrum function to accommodate additional parameters and subsequently retrieve the final spectrum as demonstrated below?

def spectrum(f, p, A):
    return A * 10**log_spectrum(np.log10(f), p=p)

c) Is this package capable of parallel execution on both High-Performance Computing (HPC) systems and personal PCs? If so, how can one configure the number of threads or cores utilized by the package?

Useful resources:

• https://github.com/singularityhub/github-ci/blob/master/.github/workflows/manual-deploy.yml
• https://github.com/marketplace/actions/setup-singularity
• https://docs.github.com/en/packages/working-with-a-github-packages-registry/working-with-the-container-registry
• https://docs.github.com/en/actions/using-workflows/triggering-a-workflow#using-filters-to-target-specific-paths-for-pull-request-or-push-events

For testing locally:

• https://github.com/nektos/act

Hi, I am trying to study the joint analysis of PTA data with other cosmological data such as Planck using the MCMC sampler Cobaya. Can the PTArcade be integrated with cosmological samplers? Please let me know how to do it if it is possible.

plot_posteriors crashes when save = False if it does not find a directory named plots in the local directory.

Possible solutions:

• change it so that a plots directory is created if none is found in the local directory.
• save the plot in the local directory by default

Related to this we should allow the user to specify the output directory.

Describe the bug
A clear and concise description of what the bug is.

To Reproduce
Steps to reproduce the behavior:

1. Go to '...'
2. Click on '....'
3. Scroll down to '....'
4. See error

Expected behavior
A clear and concise description of what you expected to happen.

Screenshots
If applicable, add screenshots to help explain your problem.

Desktop (please complete the following information):

• OS: [e.g. Ubuntu 22.04]
• Python version [e.g. 3.10.11]
• Python environment [e.g. conda, venv]
• Version [e.g. 0.1.2]

Additional context
Add any other context about the problem here.

Is your feature request related to a problem? Please describe.
Input spectrum models have a peak frequency. It'd be nice to have a separate method to compute it for given spectral parameters

Describe the solution you'd like
Add a separate method in the input models to compute f0

Describe alternatives you've considered
I can just copy the code out of model.py and add my own f0 function for each input model. But it'd be nice to have it as a formal part of each input spectrum

Additional context
I want to see where the peak frequency compares to the PTA band