darothen / pyrcel Goto Github PK

• Re-factor code
• Re-name repository
• Update github
• Update readthedocs
• Update travis
• Update Zenodo

Dear Dr. Rothenberg,
I am currently utilizing the pyrcel package for my research on aerosol activation. I have a specific question regarding the output of the model. I would like to know if pyrcel can provide the number concentration of aerosols at each integration time step. Specifically, does the model output the number concentration corresponding to the new sizes after hygroscopic growth for each bin, or does it maintain the original number concentration for each bin despite changes in particle size?
Thank you very much for your assistance and for developing such a valuable tool for the research community. I look forward to your response.

Best regards,
Feiya

Currently, the air temperature is diagnosed in both the thermodynamics and RHS routines by using the virtual temperature, $$T_v \equiv T(1 + 0.61w)$$ Rather - at least and most importantly in the RHS evaluation - it should be diagnosed using density temperature, $$T_\rho \equiv T(1 - q_t + q/\epsilon)$$ where $q_t$ is the total vapor/condensate mixing ratio.

Right now, you can't import any of the model components without a working build of Assimulo. That shouldn't be a requirement - the integrators.pypackage should be refactored to accommodate this.

Dear Daniel,

This is Tianning Zhao. I'm currently using PYRCEL for a course project. Thanks a lot for all the efforts to put together parcel model in python, it was really helpful. I was having trouble to run ARG activation for customized multimode lognormal aerosols. The error I was getting is like this: 'MultiModeLognorm' object has no attribute 'mu'. Could you please help me fix this? Thank you.

A litany of things to do in prep for a nice, new v1.4.0 release:

• Update documentation to include more references to work using this paper
• Deprecate all Cython for Numba
• (potentially) re-factor integration solvers to use a different ODE solver library (either call out to Julia's DifferentialEquations or possibly scikit-odes
• Re-factor internal arrays to wrap behind xarray
• Create new Factory-style initialization procedure for easier customizability of the core simulation
• black formatting
• More complete type annotations throughout code
• Deprecate Python <= 3.5 support
• Move from TravisCI to CircleCI

On this page

https://pyrcel.readthedocs.io/en/latest/install.html

it says to run

python setup.py build_ext --inplace

but there is no setup.py file in the repository (it was removed in commit 58a34c7).

Proposing a user-extensible output module. The basic idea is that a user can pass either a set of default or derived-type classes which encapsulates how to record data to disk. For instance, suppose we have a default writer type

class OutputWriter(object):

    def __init__(self, output_filename):
        self.output_filename = output_filename

    def write_step(self, model_state)
    def finalize(self)

The __init__ and finalize methods take care of maintaining a reference to the output file. The write_step method would be what is called during model runtime. E.g., in a proposed model physics loop:

output_handler = OutputWriter("my_output.csv")

while t < t_end:
    model.step()
    if ( output_step ):
        output_handler.write_step(model.state)
    t += dt

Then any output handlers simply need to implement the interface. Ideally, the write_step file would incrementally add to a file.

Need a library of convenience methods for producing common plots on

• supersaturation, temperature, other variable traces
• diagnosis of time-dependent activation
• initial / dry / wet vs final aerosol / droplet spectra
• more as identified

see https://pypi.org/project/pyrcel/#history
Thanks,
Sylwester