Create informed priors for reflectometry analysis¶
avant is a python package to extract values from refl_database to create informed or uniform priors to be used in reflectometry analysis.
The aim of avant is to improve reflectometry analysis by applying Bayesian Statistics and creating ‘informed priors’ which take into account literature values in the prior probability distributions of the parameters. The priors are created in a way where they can be directly implemented into refnx to perform reflectometry analysis. Plus, the package has plotting functionalities so you can see what the prior probability distirbutions look like.
Installation¶
avant can be installed from the PyPi package manager with pip
pip install avant
The package can also be found on Github along with installation instructions.
Features¶
Currently, avant only contains priors for the following five parameters for DMPC: head volume, tail volume, head thickness, tail thickness and roughness. It can create an informed prior, Gauss, with the following methods:
pdf: probability distribution function
logpdf: natural log of the probability distribution function
cdf: cumulative distribution function
ppf: percentile point function (also known as quantile function or inverse cdf)
rvs: random variate sampling
It can also create a uniform prior which is an upper and lower bound for the prior range.
The following plotting functionalities are available:
plotGauss(name): Plot an informed Gauss prior probability distribution.
plotUniform(name): Plot a uniform prior probability distribution.
Examples¶
For example, to plot the informed prior for the head volume for DMPC:
[1]:
import avant.parameter.vh as vh
vh.plotGauss('DMPC')
---------------------------------------------------------------------------
ModuleNotFoundError Traceback (most recent call last)
/tmp/ipykernel_124/3786636547.py in <module>
----> 1 import avant.parameter.vh as vh
2 vh.plotGauss('DMPC')
ModuleNotFoundError: No module named 'avant'
For example, to plot the uniform prior for the head volume for DMPC:
[2]:
import avant.parameter.vh as vh
vh.plotUniform('DMPC')
---------------------------------------------------------------------------
ModuleNotFoundError Traceback (most recent call last)
/tmp/ipykernel_124/4090706304.py in <module>
----> 1 import avant.parameter.vh as vh
2 vh.plotUniform('DMPC')
ModuleNotFoundError: No module named 'avant'
Here is an example of setting the value of the parameter to that of the Gauss object - this is a class containing statistical information about the prior and can be fed straight into Refnx.
For example, to set a parameter for the head volume of DMPC:
[3]:
import avant.parameter.vh as vh
x = vh.Gauss('DMPC')
---------------------------------------------------------------------------
ModuleNotFoundError Traceback (most recent call last)
/tmp/ipykernel_124/3562674503.py in <module>
----> 1 import avant.parameter.vh as vh
2 x = vh.Gauss('DMPC')
ModuleNotFoundError: No module named 'avant'
Here is an example of setting the value of the parameter to a tuple containing the lower and upper bound of the uniform range of values for the parameter. This can be fed straight into Refnx.
For example, to set a parameter for the head volume of DMPC:
[4]:
import avant.parameter.vh as vh
x = vh.uniform('DMPC')
---------------------------------------------------------------------------
ModuleNotFoundError Traceback (most recent call last)
/tmp/ipykernel_124/3915618590.py in <module>
----> 1 import avant.parameter.vh as vh
2 x = vh.uniform('DMPC')
ModuleNotFoundError: No module named 'avant'