Home

Awesome

Gaussian

This library was developed targeting Arduino applications. However, works just great with any C++.

The use of Gaussians is really hard because requires a deep understanding about complex math.

Not anymore. We have done it for you, fell free to use the fully-implemented class of Gaussian.

Do you want to do a Moving Average? GaussianAverage is just what you are looking for. Simple or with Gaussians, it's already implemented for you.

Find out more about Gaussians here: Wikipedia::Gaussian Function.

Installation

  1. Download the Latest release from gitHub.
  2. Unzip and modify the Folder name to "Gaussian" (Remove the '-version')
  3. Paste the modified folder on your Library folder (On your Libraries folder inside Sketchbooks or Arduino software).
  4. Download and Install LinkedList class as well, if you are going to use GaussianAverage class.
  5. Restart the Arduino Software

Getting started

Gaussian class

A class that contains a mean and a variance attribute.

Also contains methods to do sums with others Gaussians.

To declare a Gaussian object:

// Simple Gaussian initialization and instantiation
Gaussian myGaussian = Gaussian(mean, variance);

// This will declare a Gaussian with mean = 0 and variance = MAX_VARIANCE (A really big value)
Gaussian myGaussian = Gaussian();

// This will do the same as above
Gaussian myGaussian();
Gaussian myGaussian(0);

Dealing with mean is very easy. You can either access it direcly (attribute mean), or change it with the methods move and setMean. Both methods returns the self Gaussian object, allowing you to do actions consecutively in one single line.

Set, Change and get mean values

// Set or Get the attribute direcly to the object
myGaussian.mean = 10;

// Increase the mean by `x`
myGaussian.move(10);

// Set to 10 and move by -10
myGaussian.setMean(10).move(-10);

When dealing with the variance, you can also do the same as the mean. Access it direcly from the object, or use the methods vary and setVariance to change the variance.

Set, Change and get variance values

// Set or Get the attribute direcly to the object
myGaussian.variance = 4;

// Increase the variance by `y`
myGaussian.vary(3);

// Set to 3 and then vary 10 more
myGaussian.setVariance(3).vary(10);

Ok, now that we can change both mean and variance, we can now go to the next step: Summing Gaussians.

It's already implemented all the math to do that, and it's nos so hard to understand it, after resuming sum of Gaussians to 'sum of pondered values'.

The operators +, +=, = works just good with the class Gaussian.

Sum, Set and do Math with Gaussians

Gaussian g1 = Gaussian(0, 10);
Gaussian g2(10, 10);

// Simply sum both of them (the + method is overloaded on the class)
Gaussian result = g1 + g2;

// Or do like this
result = g1;
result += g2;

There are also two methods very usefull for plotting and generating normal distributed random values.

Check the examples GaussianUniformPlot and GaussianRandomPlot

Plotting Gaussians and randomizing values

// In case you want to find out the probability of 'x' of the Gaussian,
// make use of the plot(x) method.
Gaussian myGaussian(3, 5.4);

double Y1 = myGaussian.plot(3);
double Y2 = myGaussian.plot(7.3);

// Also, if you want to generate a normaly distribute value for the
// given Gaussian, make use of the random() method.

// This will generate a random value, normaly distribute around the mean
double myRandomX = myGaussian.random();
double otherRandomX = myGaussian.random();

============================

GaussianAverage class

This class provides a simple but really powerfull filter called Moving Average.

It's an average of the last n values, but using Gaussians to incorporate the power of uncertainty.

You can also use it as a simple Moving Average by setting the variance to a fixed value, or just not setting it (The default value when a Gaussian is created is really High, and is a constant).

Also, GaussianAverage class EXTENDS from Gaussian. Instead of calculating only the average (mean), you have access to the variance as well.

ATTENTION: This class REQUIRES LinkedList class as well. It MUST be included BEFORE GaussianAverage.

#include <LinkedList.h>
#include <GaussianAverage.h>

To declare a GaussianAverage object

// The default number of samples is 4
GaussianAverage myAverage = GaussianAverage();

// If you want to change it pass to constructor
// 30 samples will be stored
GaussianAverage myAverage = GaussianAverage(30);

// The same as above
GaussianAverage myAverage(30);

Now that we have our class instantiated, let's proceed to add Values (or Gaussians) to it.

There are two main ways to add values to the Average. Using the sum(Gaussian) method, or += overloaded operator.

To add values to the Moving Average just use the operator +=

// Add it through the add method
myAverage.add(Gaussian(32,20));
myAverage.add(myGaussian);

// Add it with the overloaded operator +=
myAverage += Gaussian(32, 20);
myAverage += myGaussian;

// You can also add an double/integer to it
// (note that the Variance will be always the same, and very HIGH)
myAverage += 32;
myAverage += 70;

Ok, we have created and added... What is missing? Yes, Process.

The process of calculating the average is not done every time you add something new (you migth want to add three samples before calculating). So, to calculate it you can use the method process().

Once you have processed, the values of the mean and variance will be stored inside the object. Note that process method returns the self object.

To minimize the CPU usage, we have implemented a cache detector. If you use the method process more than once before adding something new, the new average will not be calculated.

Process the new Average

// This will run the calculation and store the result inside the myAverage object
// (remember that it's also a Gaussian!)
myAverage.process();

// Or perhaps, you want to save it
mySavedAverage = myAverage.process();

// Or you want to process, change it, and then save
myAverage.process();
myAverage.vary(30).move(10);

mySavedAverage = myAverage;

Library Reference

Gaussian class

=======================

GaussianAverage class