### Install ExoJAX from GitHub Source

Source: https://github.com/hajimekawahara/exojax/blob/master/documents/userguide/installation.rst

Clone the ExoJAX repository and install it locally. This method is useful for development or when needing the latest unreleased features.

```sh
git clone https://github.com/HajimeKawahara/exojax.git
cd exojax
pip install .
```

--------------------------------

### Flux-based Reflection with No Emission (OpartReflectPure Setup)

Source: https://github.com/hajimekawahara/exojax/blob/master/documents/userguide/rtransfer_fbased.rst

Sets up the OpartReflectPure class for radiative transfer calculations, requiring a custom OpaLayer class. This example defines a basic OpaLayer with a specified wavenumber grid.

```python
from exojax.opacity import OpaPremodit
from exojax.rt import OpartReflectPure
from exojax.rt.layeropacity import single_layer_optical_depth
from exojax.utils.grids import wavenumber_grid
from exojax.database.exomol.api import MdbExomol
from exojax.utils.astrofunc import gravity_jupiter
import jax.numpy as jnp
from jax import config
config.update("jax_enable_x64", True)

class OpaLayer:
    # user defined class, needs to define self.nugrid
    def __init__(self, Nnus=100000):
        self.nu_grid, self.wav, self.resolution = wavenumber_grid(
            #1900.0, 2300.0, Nnus, unit="cm-1", xsmode="premodit"
            2050.0, 2150.0, Nnus, unit="cm-1", xsmode="premodit"

```

--------------------------------

### Install Specutils and JoviSpec

Source: https://github.com/hajimekawahara/exojax/blob/master/documents/tutorials/Fitting_Telluric_Lines.ipynb

Install the necessary external packages, specutils and JoviSpec, using pip and by cloning the repository and running setup.py.

```sh
pip install specutils
git clone https://github.com/HajimeKawahara/jovispec.git
cd jovispec
python setup.py install
```

--------------------------------

### Initial Installation Attempt

Source: https://github.com/hajimekawahara/exojax/wiki/Error-about-yaml

This command attempts to install exojax, which may lead to a PyYAML version error if an incompatible version is already present.

```sh
python setup.py install
```

--------------------------------

### Install ExoJAX from PyPI

Source: https://github.com/hajimekawahara/exojax/blob/master/documents/userguide/installation.rst

Use this command to install the latest stable version of ExoJAX from the Python Package Index.

```sh
pip install exojax
```

--------------------------------

### Build Documentation

Source: https://github.com/hajimekawahara/exojax/blob/master/CLAUDE.md

Build the HTML documentation using Sphinx. Ensure you are in the 'documents' directory and have Sphinx installed.

```bash
cd documents && make clean && make html
```

--------------------------------

### Import ExoJAX and Setup Precision

Source: https://github.com/hajimekawahara/exojax/blob/master/documents/tutorials/ckd_transpure.ipynb

Imports necessary libraries for ExoJAX, including modules for opacity, radiative transfer, and testing. Enables 64-bit precision for calculations.

```python
# Import required packages
import numpy as np
import matplotlib.pyplot as plt
from jax import config

# ExoJAX imports
from exojax.test.emulate_mdb import mock_mdbExomol, mock_wavenumber_grid
from exojax.opacity import OpaCKD, OpaPremodit
from exojax.rt import ArtTransPure
from exojax.test.data import get_testdata_filename, TESTDATA_CO_EXOMOL_PREMODIT_TRANSMISSION_REF

# Enable 64-bit precision for accurate calculations
config.update("jax_enable_x64", True)

print("ExoJAX CKD Tutorial: Transmission Spectroscopy")
print("=============================================")
```

--------------------------------

### Set up Stochastic Variational Inference (SVI)

Source: https://github.com/hajimekawahara/exojax/blob/master/documents/tutorials/Differentiable_Programming.ipynb

Initialize SVI with a guide, optimizer, and loss function. This is used for Bayesian inference when gradients are available.

```python
from numpyro.infer import SVI
from numpyro.infer import Trace_ELBO
import numpyro.optim as optim
from numpyro.infer.autoguide import AutoMultivariateNormal

guide = AutoMultivariateNormal(model)
optimizer = optim.Adam(0.01)
svi = SVI(model, guide, optimizer, loss=Trace_ELBO())
```

--------------------------------

### Initialize Atmospheric and Instrumental Parameters

Source: https://github.com/hajimekawahara/exojax/blob/master/documents/tutorials/Reverse_modeling_for_methane_using_MODIT.rst

Sets up atmospheric pressure layers, wavenumber grid, and instrumental spectral resolution.

```ipython3
from exojax.rt.rtransfer import pressure_layer,wavenumber_grid
from exojax.utils.constants import c
from exojax.utils.instfunc import resolution_to_gaussian_std

NP=100
Parr, dParr, k=pressure_layer(NP=NP)
Nx=5000
nus,wav,res=wavenumber_grid(np.min(wavd)-5.0,np.max(wavd)+5.0,Nx,unit="AA",xsmode="modit")

Rinst=100000. #instrumental spectral resolution
beta_inst=resolution_to_gaussian_std(Rinst)  #equivalent to beta=c/(2.0*np.sqrt(2.0*np.log(2.0))*R)
```

--------------------------------

### Troubleshooting Missing libcudnn.so.7

Source: https://github.com/hajimekawahara/exojax/wiki/Error-about-DNN

If the error message mentions 'libcudnn.so.7', verify that this specific version is installed and accessible. This snippet shows an example of such an error message.

```sh
2021-08-15 12:06:46.013903: W external/org_tensorflow/tensorflow/stream_executor/platform/default/dso_loader.cc:64] 
Could not load dynamic library 'libcudnn.so.7'; 
dlerror: libcudnn.so.7: cannot open shared object file: No such file or directory;

```

--------------------------------

### Load ExoMol Database and Create OpaPremodit

Source: https://github.com/hajimekawahara/exojax/blob/master/documents/userguide/history.rst

Demonstrates how to load an ExoMol database and initialize OpaPremodit with specific parameters for auto-ranging, broadening resolution, and memory policy.

```python
from exojax.database.exomol.api import MdbExomol
from exojax.opacity.premodit.api import OpaPremodit
from exojax.opacity.policies import MemoryPolicy

mdb = MdbExomol(".database/CO/12C-16O/Li2015", nu_grid)
opa = OpaPremodit.from_mdb(
    mdb, nu_grid,
    auto_trange=(500,1500),
    broadening_resolution={"mode":"manual","value":0.2},
    memory_policy=MemoryPolicy(allow_32bit=True, nstitch=2)
)
xs = opa.xsvector(T=1200.0, P=0.1)  # or opa.xsmatrix(Tarr, Parr)
```

--------------------------------

### Check Installed Version

Source: https://github.com/hajimekawahara/exojax/blob/master/CLAUDE.md

Verify the installed version of ExoJAX. This is useful for confirming installation and compatibility.

```python
import exojax
print(exojax.__version__)
```

--------------------------------

### Set up and Run frun with Specific Parameters

Source: https://github.com/hajimekawahara/exojax/blob/master/documents/tutorials/reverse_premodit.rst

Initializes parameters and calls the `frun` function to generate a test spectrum.

```python
import matplotlib.pyplot as plt
#g = gravity_jupiter(0.88, 33.2)
Rp = 0.88
Mp = 33.2
alpha = 0.1
MMR_CH4 = 0.0059
vsini = 20.0
RV = 10.0
T0 = 1200.0
u1 = 0.0
u2 = 0.0
Tarr = art.powerlaw_temperature(T0, alpha)
Ftest = frun(Tarr, MMR_CH4, Mp, Rp, u1, u2, RV, vsini)
```

--------------------------------

### Setting up Opacity Calculator with HITRAN and PreMODIT

Source: https://github.com/hajimekawahara/exojax/blob/master/documents/tutorials/Fitting_Telluric_Lines.ipynb

Initializes an opacity calculator using HITRAN molecular database and the PreMODIT method. This is suitable for low-temperature environments. Ensure the nu_grid is correctly defined.

```python
wavelength_start = 7100.0  # AA
wavelength_end = 7450.0  # AA
from exojax.database.api  import MdbHitran
from exojax.opacity import OpaDirect
from exojax.opacity import OpaPremodit
from exojax.utils.grids import wavenumber_grid
from exojax.utils.grids import wav2nu

N = 40000

margin = 10  # cm-1
nus_start = wav2nu(wavelength_end, unit="AA") - margin
nus_end = wav2nu(wavelength_start, unit="AA") + margin
#nus_start = 1.e8/wavelength_end - margin
#nus_end = 1.e8/wavelength_start + margin

mdb_water = MdbHitran("H2O", nurange=[nus_start, nus_end], isotope=1)
nus, wav, res = wavenumber_grid(nus_start, nus_end, N, xsmode="lpf", unit="cm-1")
# opa = OpaDirect(mdb_water, nu_grid=nus)
opa = OpaPremodit(mdb_water, nu_grid=nus, allow_32bit=True, auto_trange=[150.0, 300.0])
```

--------------------------------

### Install CUDA 11.5 and cuDNN on Ubuntu

Source: https://github.com/hajimekawahara/exojax/wiki/installing-or-updating-JAX

Use aptitude to install CUDA 11.5 and dpkg to install a local cuDNN repository. This is a prerequisite for installing JAX with specific CUDA/cuDNN versions.

```shell
sudo aptitude install cuda-11-5
sudo dpkg -i cudnn-local-repo-ubuntu2004-8.3.1.22_1.0-1_amd64.deb
```

--------------------------------

### Initialize AutoMultivariateNormal Guide

Source: https://github.com/hajimekawahara/exojax/blob/master/documents/tutorials/get_started_svi.ipynb

Initializes the AutoMultivariateNormal guide for a given model. This guide is suitable for models where latent variables can be approximated by a multivariate normal distribution.

```python
from numpyro.infer.autoguide import AutoMultivariateNormal
guide = AutoMultivariateNormal(model_prob)
optimizer = optim.Adam(0.01)
svi = SVI(model_prob, guide, optimizer, loss=Trace_ELBO())
```

--------------------------------

### Initialize Wavenumber Grid and Molecular Database

Source: https://github.com/hajimekawahara/exojax/blob/master/documents/tutorials/elower_setting.ipynb

Sets up the wavenumber grid and loads the Exomol molecular database. Ensure the database path is correct and the wavenumber grid is in ascending order.

```python
from exojax.utils.grids import wavenumber_grid
from exojax.database.exomol.api import MdbExomol

nu_grid, wav, resolution = wavenumber_grid(2200.,
                                           2300.,
                                           10000,
                                           unit="cm-1",
                                           xsmode="premodit")
mdb = MdbExomol(".database/CO/12C-16O/Li2015", nurange=nu_grid)
```

--------------------------------

### Initialize AutoBNAFNormal Guide

Source: https://github.com/hajimekawahara/exojax/blob/master/documents/tutorials/get_started_svi.rst

Initialize the AutoBNAFNormal guide for a given probabilistic model. This guide uses Block Neural Autoregressive Flow (BNAF) for more flexible dependency modeling.

```ipython3
from numpyro.infer.autoguide import AutoBNAFNormal
guide = AutoBNAFNormal(model_prob)
optimizer = optim.Adam(0.01)
svi = SVI(model_prob, guide, optimizer, loss=Trace_ELBO())
```

--------------------------------

### Setup Wavenumber Grid, Molecular Database, and Atmospheric Model

Source: https://github.com/hajimekawahara/exojax/blob/master/documents/tutorials/ckd_emispure.rst

Sets up a mock wavenumber grid, creates a mock H2O molecular database, and initializes the ArtEmisPure atmospheric radiative transfer solver with specified pressure and layer parameters.

```python
# Setup wavenumber grid and molecular database
nu_grid, wav, res = mock_wavenumber_grid()
print(f"Wavenumber grid: {len(nu_grid)} points from {nu_grid[0]:.1f} to {nu_grid[-1]:.1f} cm⁻¹")
print(f"Spectral resolution: {res:.1f}")

# Create mock H2O molecular database
mdb = mock_mdbExomol("H2O")
print(f"Molecular database: {mdb.nurange[0]:.1f} - {mdb.nurange[1]:.1f} cm⁻¹")

# Setup atmospheric radiative transfer
art = ArtEmisPure(
    pressure_top=1.0e-8, 
    pressure_btm=1.0e2, 
    nlayer=100, 
    nu_grid=nu_grid
)

print(f"Atmospheric layers: {art.nlayer}")
print(f"Pressure range: {art.pressure_top:.1e} - {art.pressure_btm:.1e} bar")
```

--------------------------------

### Install NumPyro with CUDA Support

Source: https://github.com/hajimekawahara/exojax/wiki/Installation-on-py3.9-conda-for-v2.0

Installs NumPyro version 0.13.2 with CUDA support from a specific URL. Ensure your system has compatible NVIDIA drivers and CUDA toolkit installed.

```shell
pip install numpyro[cuda]==0.13.2 -f https://storage.googleapis.com/jax-releases/jax_cuda_releases.html
```

--------------------------------

### Resolving PyYAML Version Conflict

Source: https://github.com/hajimekawahara/exojax/wiki/Error-about-yaml

This command installs the latest PyYAML version while ignoring the currently installed one, resolving the dependency conflict. It is followed by the exojax installation command.

```sh
pip install pyyaml  --ignore-installed PyYAML
python setup.py install
```

--------------------------------

### Sample Posterior Parameters with AutoBNAFNormal Guide

Source: https://github.com/hajimekawahara/exojax/blob/master/documents/tutorials/get_started_svi.rst

Sample posterior parameters using the trained AutoBNAFNormal guide. This is similar to the initial parameter sampling but uses the guide optimized by SVI.

```ipython3
predictive_posterior = Predictive(
    model_prob,
    guide=guide,
    params=params,
    num_samples=2000,
    return_sites=param_entries,
)
posterior_sample = predictive_posterior(rng_key, spectrum=None)
```

--------------------------------

### Install Core ExoJAX Dependencies

Source: https://github.com/hajimekawahara/exojax/wiki/Installation-on-py3.9-conda-for-v2.0

Installs essential Python packages for ExoJAX, including astroquery, vaex-astro, msgpack, ndindex, pybind11, vaex, cython, and shapely. Also installs pytest for testing.

```shell
pip install astroquery vaex-astro msgpack ndindex pybind11 vaex cython shapely pytest
```

--------------------------------

### Install Sphinx Dependencies

Source: https://github.com/hajimekawahara/exojax/blob/master/documents/developers/doc.rst

Installs the required Sphinx themes and emoji support for documentation.

```sh
pip install sphinx_rtd_theme sphinxemoji
```

--------------------------------

### Install JAX with CUDA 11.5 and cuDNN 8.2

Source: https://github.com/hajimekawahara/exojax/wiki/installing-or-updating-JAX

Uninstall any existing JAX installation and then install the specific JAX version compatible with CUDA 11.5 and cuDNN 8.2 using pip. The -f flag specifies the release URL.

```shell
pip uninstall jax
pip install "jax[cuda11_cudnn82]" -f https://storage.googleapis.com/jax-releases/jax_releases.html
```

--------------------------------

### Setting up spectral model components

Source: https://github.com/hajimekawahara/exojax/blob/master/documents/tutorials/Reverse_modeling_for_methane_using_MODIT.rst

Initializes components for spectral modeling, including exomol for line data, xsmatrix for cross-section matrices, and velocity grids for spectral resolution.

```python
from exojax.opacity.modit.modit import exomol,xsmatrix
from exojax.rt.rtransfer import dtauM, dtauCIA, rtrun
from exojax.rt import planck, response
from exojax.postproc.response import ipgauss_sampling
from exojax.postproc.spin_rotation import convolve_rigid_rotation
from exojax.utils.grids import velocity_grid
vsini_max = 100.0
vr_array = velocity_grid(res, vsini_max)
```

--------------------------------

### Initial Parameter Setup and Spectrum Calculation

Source: https://github.com/hajimekawahara/exojax/blob/master/documents/tutorials/Reverse_modeling_with_VALD_using_MODIT.ipynb

Sets up initial atmospheric parameters and calculates a synthetic spectrum using the `frun` function. This is useful for comparing model predictions with observed data.

```python
T0 = 3000.
alpha = 0.07
Mp=0.155 *1.99e33/1.90e30
Rp=0.186 *6.96e10/6.99e9
u1=0.0
u2=0.0
RV=0.00
vsini=2.0
mmw=2.33
log_e_H = -4.2
VMR_H = 0.09 
VMR_H2 = 0.77
VMR_FeH = 10**-8
VMR_H2O = 10**-4
VMR_OH = 10**-4
VMR_TiO = 10**-8
A_Fe = 1.5
A_Ti = 1.2
adjust_continuum = 0.99

mu = frun(T0, alpha, Mp, Rp, u1, u2, RV, vsini, \
                     mmw, log_e_H, VMR_H, VMR_H2, \
                     VMR_FeH, VMR_H2O, VMR_OH, VMR_TiO, \
                     A_Fe, A_Ti, adjust_continuum)

plt.figure(figsize = (10, 3))
plt.plot(wavd[::-1], nflux, label = "observed data")
plt.plot(wavd[::-1], mu, label="frun", ls='--')
plt.legend(); plt.grid(); plt.ylim(0.1, 1.05)
plt.show()
```

--------------------------------

### Set up Wavenumber Grid and Atmosphere

Source: https://github.com/hajimekawahara/exojax/blob/master/documents/tutorials/jupiters/Jupiter_cloud_model_using_amp.ipynb

Initializes the wavenumber grid and atmospheric parameters for reflection modeling. Ensure the wavenumber grid is in ascending order.

```python
from exojax.utils.grids import wavenumber_grid

N = 10000
nus, wav, res = wavenumber_grid(10**3, 10**4, N, xsmode="premodit")

from exojax.rt import ArtReflectPure
from exojax.utils.astrofunc import gravity_jupiter

art = ArtReflectPure(nu_grid=nus, pressure_btm=1.0e2, pressure_top=1.0e-3, nlayer=100)
art.change_temperature_range(80.0, 400.0)
Tarr = art.powerlaw_temperature(150.0, 0.2)
Parr = art.pressure
mu = 2.3  # mean molecular weight
gravity = gravity_jupiter(1.0, 1.0)
```

--------------------------------

### Install JAX CPU Version

Source: https://github.com/hajimekawahara/exojax/blob/master/documents/userguide/installation.rst

Installs the CPU-only version of JAX. Use this if you do not have a compatible GPU or do not require GPU acceleration.

```sh
pip install --upgrade jax
```

--------------------------------

### Install JAX with CUDA 11 Support

Source: https://github.com/hajimekawahara/exojax/blob/master/documents/userguide/installation.rst

Installs JAX with support for CUDA 11. Ensure your CUDA toolkit is compatible.

```sh
pip install --upgrade "jax[cuda11]" -f https://storage.googleapis.com/jax-releases/jax_cuda_releases.html
```

--------------------------------

### Initialize Cloud Model and Load Data

Source: https://github.com/hajimekawahara/exojax/blob/master/documents/tutorials/jupiters/Jupiter_Hires_Modeling.ipynb

Initializes the PdbCloud for ammonia and sets up an atmospheric model with background gas. It also checks the temperature range and prepares for condensate calculations.

```python
from exojax.database.pardb  import PdbCloud
from exojax.atm.atmphys import AmpAmcloud

pdb_nh3 = PdbCloud("NH3")
amp_nh3 = AmpAmcloud(pdb_nh3, bkgatm="H2")
amp_nh3.check_temperature_range(Tarr)
```

--------------------------------

### Install JAX with CUDA 12 Support

Source: https://github.com/hajimekawahara/exojax/blob/master/documents/userguide/installation.rst

Installs JAX with support for CUDA 12. Ensure your CUDA toolkit is compatible.

```sh
pip install --upgrade "jax[cuda12]" -f https://storage.googleapis.com/jax-releases/jax_cuda_releases.html
```

--------------------------------

### Initialize Wavenumber Grid and Load CH4 Database (Exomol)

Source: https://github.com/hajimekawahara/exojax/blob/master/documents/analysis/freq_elower.ipynb

Sets up a wavenumber grid and loads the CH4 molecular database from Exomol. Ensure the database path is correct for your system.

```python
from exojax.database.exomol.api import MdbExomol
from exojax.utils.grids import wavenumber_grid
import numpy as np


nus, wav, r = wavenumber_grid(16370, 16390, 20000, unit="AA",xsmode="premodit")
mdbCH4 = MdbExomol("/home/kawahara/.database/CH4/12C-1H4/YT34to10/",nus)
```

--------------------------------

### Install libcst for Script

Source: https://github.com/hajimekawahara/exojax/blob/master/documents/userguide/history.rst

Install the libcst library to run the script for updating import statements. Ensure you back up your files before execution.

```sh
pip install libcst
```

--------------------------------

### CH4 Setting (PREMODIT)

Source: https://github.com/hajimekawahara/exojax/blob/master/documents/tutorials/reverse_premodit.ipynb

Initializes the Exomol database for CH4 and sets up the OpaPremodit object for spectral calculations. Ensure the database path is correct and the nu_grid is defined.

```python
mdb = MdbExomol('.database/CH4/12C-1H4/YT10to10/',
                nurange=nu_grid,
                gpu_transfer=False)
print('N=', len(mdb.nu_lines))
diffmode = 0
opa = OpaPremodit(mdb=mdb,
                  nu_grid=nu_grid,
                  diffmode=diffmode,
                  auto_trange=[Tlow, Thigh], 
                  dit_grid_resolution=1.0,allow_32bit=True)
```

--------------------------------

### Minimal OpaPremodit Example (ExoMol)

Source: https://github.com/hajimekawahara/exojax/blob/master/documents/userguide/opacity_di.rst

Builds OpaPremodit opacity from a snapshot or directly from an mdb. Use `from_snapshot` to avoid dependencies on concrete mdb classes, or `from_mdb` for legacy usage.

```python
from exojax.database.exomol.api import MdbExomol
from exojax.opacity import OpaPremodit
from exojax.utils.grids import wavenumber_grid

# Make a grid and load an mdb
nu_grid, _, _ = wavenumber_grid(4200.0, 4300.0, 20000, xsmode="premodit")
mdb = MdbExomol(".database/CO/12C-16O/Li2015", nurange=nu_grid)

# 1) Build from a snapshot (data-only DTO)
snap = mdb.to_snapshot()
opa1 = OpaPremodit.from_snapshot(
    snap, nu_grid, manual_params=(5.0, 1000.0, 1200.0)
)

# 2) Back-compat: build directly from an mdb
opa2 = OpaPremodit.from_mdb(
    mdb, nu_grid, manual_params=(5.0, 1000.0, 1200.0)
)
```

--------------------------------

### Initialize plotting and get VIRGA gas recommendations

Source: https://github.com/hajimekawahara/exojax/blob/master/documents/userguide/amclouds_comparison_virga.ipynb

Sets up the Bokeh plotting environment for notebooks and uses VIRGA's `recommend_gas` function to determine relevant gases for the given atmospheric conditions.

```python
from bokeh.io import output_notebook
from bokeh.plotting import show, figure
from bokeh.palettes import Colorblind
output_notebook()

metallicity = 1 #atmospheric metallicity relative to Solar

#get virga recommendation for which gases to run
recommended = jdi.recommend_gas(pressure, temperature, metallicity,mu,
                #Turn on plotting
                 plot=True)
#print the results
print(recommended)
```

--------------------------------

### Install Conda-Managed Libraries

Source: https://github.com/hajimekawahara/exojax/wiki/Installation-on-py3.9-conda-for-v2.0

Installs HDF5, pytables, and GEOS using Conda-Forge. These are often required for data handling and geospatial operations.

```shell
conda install -c conda-forge hdf5 pytables geos
```

--------------------------------

### Initialize Wavenumber Grid and Opacity Database

Source: https://github.com/hajimekawahara/exojax/blob/master/documents/tutorials/jupiters/Jupiter_Hires_Modeling.ipynb

Sets up the wavenumber grid and loads the Hitran database for H2O opacity. Ensure wavelength units are correctly converted to wavenumbers.

```python
from exojax.database.api  import MdbHitran
from exojax.opacity import OpaDirect
from exojax.opacity import OpaPremodit
from exojax.utils.grids import wavenumber_grid
from exojax.utils.grids import wav2nu

N = 40000

margin = 10  # cm-1
nus_start = wav2nu(wavelength_end, unit="AA") - margin
nus_end = wav2nu(wavelength_start, unit="AA") + margin

mdb_water = MdbHitran("H2O", nurange=[nus_start, nus_end], isotope=1)
nus, wav, res = wavenumber_grid(nus_start, nus_end, N, xsmode="lpf", unit="cm-1")
opa_telluric = OpaPremodit(mdb_water, nu_grid=nus, allow_32bit=True, auto_trange=[80.0, 400.0])
```

--------------------------------

### Downloading ExoMol Transition Files

Source: https://github.com/hajimekawahara/exojax/blob/master/documents/tutorials/Comparing_HITEMP_and_ExoMol.rst

This snippet demonstrates downloading transition files for a specific molecule and isotopologue from ExoMol. It also mentions the caching mechanism for faster subsequent downloads and the option to delete the compressed file.

```text
Molecule:  CO
Isotopologue:  12C-16O
ExoMol database:  None
Local folder:  CO/12C-16O/Li2015
Transition files: 
	 => File 12C-16O__Li2015.trans
Total download size ['12C-16O__Li2015.trans.bz2'] is: 0.001307 GB
		 => Downloading from https://www.exomol.com/db/CO/12C-16O/Li2015/12C-16O__Li2015.trans.bz2
		 => Caching the *.trans.bz2 file to the vaex (*.h5) format. After the second time, it will become much faster.
		 => You can deleted the 'trans.bz2' file by hand.
Broadener:  H2
Broadening code level: a0
```

--------------------------------

### Clone and Install JoviSpec Repository

Source: https://github.com/hajimekawahara/exojax/blob/master/documents/tutorials/jupiters/Jupiter_Hires_Modeling_Exomol.ipynb

Instructions to clone the JoviSpec repository from GitHub and install it locally. This is necessary to access Jupiter spectral data.

```shell
git clone https://github.com/HajimeKawahara/jovispec.git
python setup.py install
```

--------------------------------

### Install with Pip in Development Mode

Source: https://github.com/hajimekawahara/exojax/blob/master/CLAUDE.md

Alternative method to install the package in development mode using pip. This is a common practice for Python projects.

```bash
pip install -e .
```

--------------------------------

### Set Up Initial Guess for Fitting Parameters

Source: https://github.com/hajimekawahara/exojax/blob/master/documents/tutorials/Fitting_Telluric_Lines.ipynb

Defines the initial guess for fitting parameters including temperature, pressure, column density, continuum, and instrumental broadening. Requires exojax.utils.instfunc and numpy.

```python
from exojax.utils.instfunc import resolution_to_gaussian_std

T_b = 200.0
P_b = 0.5
nl_b = 2.0e22
a_b = 0.52
Rinst = 100000.0
beta_inst_b = resolution_to_gaussian_std(Rinst)

initial_guess = np.array([T_b, P_b, nl_b, a_b, beta_inst_b])
initpar = np.ones_like(initial_guess)
```

--------------------------------

### Initialize PreMOLIT Opacity Calculator

Source: https://github.com/hajimekawahara/exojax/blob/master/documents/tutorials/jupiters/Jupiter_Hires_Modeling.ipynb

Initializes the PreMOLIT opacity calculator with the reduced methane data. This method is memory-efficient and suitable for high-resolution spectra.

```python
opa = OpaPremodit(mdb_reduced, nu_grid=nus, allow_32bit=True, auto_trange=[80.0, 300.0]) #this reduced the device memory use in 5/6.
```

--------------------------------

### Verify JAX sparse module import in Python

Source: https://github.com/hajimekawahara/exojax/wiki/installing-or-updating-JAX

After installation, verify that the JAX sparse module can be imported successfully within a Python interpreter. This confirms the installation and CUDA/cuDNN compatibility.

```python
from jax.experimental.sparse import coo
```

--------------------------------

### Initialize PreMoidit Opacity with Mock Data

Source: https://github.com/hajimekawahara/exojax/blob/master/documents/tutorials/ckd_principle.rst

This snippet demonstrates how to initialize the OpaPremodit class using mock molecular database and wavenumber grid data. It sets up the opacity object for a specific molecule (H2O) and wavenumber range, with automatic temperature range handling.

```python
from exojax.test.emulate_mdb import mock_mdbExomol, mock_wavenumber_grid
from exojax.opacity.premodit.api import OpaPremodit
from exojax.opacity.ckd.api import OpaCKD
from exojax.opacity.ckd.core import compute_g_ordinates
import jax.numpy as jnp
import matplotlib.pyplot as plt
from jax import config

config.update("jax_enable_x64", True)

nus, wav, res = mock_wavenumber_grid(lambda0=22930.0, lambda1=22940.0, Nx=20000)
mdb = mock_mdbExomol("H2O")
opa = OpaPremodit(mdb, nus, auto_trange=[500.0, 1500.0])
```

--------------------------------

### Installing cuDNN for Exojax

Source: https://github.com/hajimekawahara/exojax/wiki/Error-about-DNN

This command sequence provides instructions for installing cuDNN, including copying library files and headers to the appropriate CUDA directories. This is a common solution for cuDNN version mismatch errors.

```sh
tar xvfz cudnn-10.1-linux-x64-v7.6.5.32.tgz
sudo cp cuda/lib64/libcudnn* /usr/local/cuda/targets/x86_64-linux/lib/
sudo cp cuda/include/cudnn*.h /usr/local/cuda/include
sudo cp cuda/lib64/libcudnn* /usr/local/cuda/lib64
sudo chmod a+r /usr/local/cuda/include/cudnn*.h /usr/local/cuda/lib64/libcudnn*

```

--------------------------------

### Set up T-P profile and partial pressures

Source: https://github.com/hajimekawahara/exojax/blob/master/documents/tutorials/Forward_modeling_for_Fe_I_lines_of_Kurucz.rst

Initializes an atmospheric model with a specified number of layers and defines temperature and pressure profiles. It also calculates partial pressures for different gas species.

```ipython3
from exojax.rt import ArtEmisPure
nlayer = 100
T0 = 3000.0  # 10000. #3000. #1295.0 #K
alpha = 0.1
Tlow = 500.0
Thigh = 4500.0
art = ArtEmisPure(nu_grid=nu_grid, pressure_top=1.e-8, pressure_btm=1.e2, nlayer=nlayer)
art.change_temperature_range(Tlow, Thigh)
Tarr = art.powerlaw_temperature(T0, alpha)
Parr = art.pressure
dParr = art.dParr

H_He_HH_VMR = [0.0, 0.16, 0.84]  # typical quasi-"solar-fraction"
PH = Parr * H_He_HH_VMR[0]
PHe = Parr * H_He_HH_VMR[1]
PHH = Parr * H_He_HH_VMR[2]

fig = plt.figure(figsize=(6, 4))
plt.plot(Tarr, Parr, label="$P_\mathrm{total}$")
plt.plot(Tarr, PH, "--", label="$P_\mathrm{H}$")
plt.plot(Tarr, PHH, "--", label="$P_\mathrm{H_2}$")
plt.plot(Tarr, PHe, "--", label="$P_\mathrm{He}$")
plt.plot(Tarr[80], Parr[80], marker="*", markersize=15)
plt.yscale("log")
plt.xlabel("temperature (K)")
plt.ylabel("pressure (bar)")
plt.gca().invert_yaxis()
plt.legend()
plt.show()
```

--------------------------------

### Chemical Setup and Species Indexing

Source: https://github.com/hajimekawahara/exojax/blob/master/documents/tutorials/equilibrium_chemistry.rst

Sets up chemical equilibrium calculations using `chemsetup` from `exogibbs.presets.fastchem`. It prints the indices and JANAF names for CO and H2, and lists the available elements.

```ipython3
from exogibbs.presets.fastchem import chemsetup
# chemical setup
chem = chemsetup()
idx_co = chem.species.index("C1O1")
print("idx for CO=",idx_co, "JANAF name", chem.species[idx_co])  # check index of CO
idx_h2 = chem.species.index("H2")
print("idx for H2=",idx_h2, "JANAF name", chem.species[idx_h2])  # check index of H2
print("element:", chem.elements)
```

--------------------------------

### Initialize and Setup CKD Opacity Calculator

Source: https://github.com/hajimekawahara/exojax/blob/master/documents/tutorials/ckd_emispure.rst

Sets up the Correlated-k (CKD) opacity calculator using a base opacity calculator. It configures parameters like the number of g-ordinates and spectral band width. The setup also includes printing the calculator's configuration details.

```python
# Initialize CKD opacity calculator
opa_ckd = OpaCKD(
    base_opa,           # Base opacity calculator
    Ng=32,              # Number of g-ordinates for quadrature
    band_width=0.5      # Spectral band width
)

print(f"CKD Opacity Calculator Setup:")
print(f"  Number of g-ordinates (Ng): {opa_ckd.Ng}")
print(f"  Band width: {opa_ckd.band_width}")
print(f"  Number of spectral bands: {len(opa_ckd.nu_bands)}")
print(f"  Spectral range: {opa_ckd.nu_bands[0]:.1f} - {opa_ckd.nu_bands[-1]:.1f} cm⁻¹")
```

--------------------------------

### Get Molecular Mass

Source: https://github.com/hajimekawahara/exojax/blob/master/documents/tutorials/Forward_modeling_using_PreMODIT.ipynb

Retrieves the molecular mass of CO from the loaded database.

```python
molmassCO=mdbCO.molmass
```

--------------------------------

### Initialize ExoMol with Optional Quantum States

Source: https://github.com/hajimekawahara/exojax/blob/master/documents/userguide/api.rst

Illustrates initializing an ExoMol database with optional quantum states enabled and activation turned off. This setup is necessary for filtering lines based on vibrational or electronic states using the DataFrame.

```ipython
>>> from exojax.utils.grids import wavenumber_grid
>>> from exojax.database.exomol.api import MdbExomol 
>>> 
>>> nus, wav, res = wavenumber_grid(24000.0, 26000.0, 1000, unit="AA")
>>> mdb =MdbExomol("CO/12C-16O/Li2015/", nus, optional_quantum_states=True, activation=False)
```

--------------------------------

### Get Molecular Mass

Source: https://github.com/hajimekawahara/exojax/blob/master/documents/tutorials/Forward_modeling_using_the_DIT_Cross_Section_for_methane.ipynb

Fetches the molecular mass for CH4 using the molinfo module.

```python
from exojax.database import molinfo 
molmassCH4=molinfo.molmass("CH4")
```

--------------------------------

### Initialize Cloud and Microphysics Models

Source: https://github.com/hajimekawahara/exojax/blob/master/documents/tutorials/jupiters/Jupiter_cloud_model_using_amp.ipynb

Sets up the particulate database for NH3 clouds and the atmospheric microphysics model (AM01). Ensure the temperature range of the atmosphere is compatible with the cloud model.

```python
from exojax.database.pardb  import PdbCloud
from exojax.atm.atmphys import AmpAmcloud

pdb_nh3 = PdbCloud("NH3")
#pdb_nh3.generate_miegrid() # when you use the miegrid
#pdb_nh3.load_miegrid() # when you use the miegrid

amp_nh3 = AmpAmcloud(pdb_nh3,bkgatm="H2")
amp_nh3.check_temperature_range(Tarr)
```

--------------------------------

### Define Wavelength Range

Source: https://github.com/hajimekawahara/exojax/blob/master/documents/tutorials/jupiters/Jupiter_Hires_Modeling.ipynb

Sets the start and end wavelengths for spectral analysis in Angstroms.

```python
wavelength_start = 8500.0 #AA
wavelength_end = 8800.0 #AA
```

--------------------------------

### Plot Temperature-Pressure Profile

Source: https://github.com/hajimekawahara/exojax/blob/master/documents/tutorials/Forward_modeling.ipynb

Visualizes the generated temperature-pressure profile. Ensure matplotlib is installed for plotting.

```python
import matplotlib.pyplot as plt
plt.plot(Tarr,Parr)
plt.yscale("log")
plt.gca().invert_yaxis()
plt.show()
```

--------------------------------

### CH4 Setting (PREMODIT)

Source: https://github.com/hajimekawahara/exojax/blob/master/documents/tutorials/reverse_premodit.rst

Initializes the MdbExomol database and OpaPremodit for CH4 opacity calculations. Ensure the database path is correct and specify the desired nu_grid.

```ipython3
### CH4 setting (PREMODIT)
mdb = MdbExomol('.database/CH4/12C-1H4/YT10to10/',
                nurange=nu_grid,
                gpu_transfer=False)
print('N=', len(mdb.nu_lines))
diffmode = 0
opa = OpaPremodit(mdb=mdb,
                  nu_grid=nu_grid,
                  diffmode=diffmode,
                  auto_trange=[Tlow, Thigh],
                  dit_grid_resolution=1.0,allow_32bit=True)
```

--------------------------------

### Get Exact Isotope Name

Source: https://github.com/hajimekawahara/exojax/blob/master/documents/userguide/api.rst

Retrieve the exact name of an isotope given its numerical identifier.

```ipython
>>> mdb.exact_isotope_name(1) #-> (12C)(16O)
```

--------------------------------

### Set up Wavenumber Grid and Load Molecular Database

Source: https://github.com/hajimekawahara/exojax/blob/master/documents/analysis/freq_elower.ipynb

Initializes a wavenumber grid and loads a molecular database for CH4. Ensure the database path is correct for your system.

```python
nusa, wava, ra = wavenumber_grid(16350, 16450, 1000, unit="AA",xsmode="premodit")
mdbCH4a = MdbExomol("/home/kawahara/.database/CH4/12C-1H4/YT34to10/",nusa)
```

--------------------------------

### Get Methane Molecular Mass

Source: https://github.com/hajimekawahara/exojax/blob/master/documents/tutorials/Forward_modeling_using_PreMODIT_Cross_Section_for_methane.ipynb

Retrieves the molecular mass of methane from the loaded molecular database.

```python
molmassCH4=mdbCH4.molmass
```

--------------------------------

### Setting up and Running the MCMC Sampler

Source: https://github.com/hajimekawahara/exojax/blob/master/documents/tutorials/Reverse_modeling_with_VALD_using_MODIT.ipynb

Initializes the NUTS kernel and MCMC object for sampling from the posterior distribution defined by `model_c`. Requires a PRNGKey for reproducibility.

```python
rng_key = random.PRNGKey(0)
rng_key, rng_key_ = random.split(rng_key)
num_warmup, num_samples = 100, 200 #500, 1000
kernel = NUTS(model_c,forward_mode_differentiation=True,max_tree_depth=7)
mcmc = MCMC(kernel, num_warmup=num_warmup, num_samples=num_samples)
```

--------------------------------

### Get molecular mass

Source: https://github.com/hajimekawahara/exojax/blob/master/documents/tutorials/Forward_modeling_using_DIT.rst

Retrieves the molecular mass for CO, which is needed for Doppler broadening calculations.

```ipython3
from exojax.database import molinfo 
molmassCO=molinfo.molmass("CO")
```

--------------------------------

### Uninstall ExoJAX

Source: https://github.com/hajimekawahara/exojax/blob/master/documents/userguide/installation.rst

Recommended command to run before installing a new version to prevent conflicts with removed modules.

```sh
pip uninstall exojax
```

--------------------------------

### Initialize Wavenumber Grid

Source: https://github.com/hajimekawahara/exojax/blob/master/documents/tutorials/branch.ipynb

Sets up the wavenumber grid for spectral analysis. Ensure the grid is in ascending order; descending order will raise an error.

```python
from exojax.database.exomol.api import MdbExomol
from exojax.utils.grids import wavenumber_grid
nus,wave,resolution = wavenumber_grid(1900.0,2300.0,150000,xsmode="lpf")
```

--------------------------------

### Generate Sphinx Documentation

Source: https://github.com/hajimekawahara/exojax/blob/master/documents/developers/doc.rst

Installs the package, generates API documentation from source, and builds the HTML documentation.

```python3
python setup.py install
rm -rf documents/exojax
sphinx-apidoc -F -o documents/exojax src/exojax
cd documents
make clean
make html
```