My Coding >
Numerical simulations >
Stochastic methods >
MonteCarlo methods >
MonteCarlo method for sphere volume calculation
MonteCarlo method for sphere volume calculationIn this tutorial. I will show, how to use the MonteCarlo technique for the calculation of the volume of the ndimensional sphere, or, socalled, nball. This method is great for cases when you don't know how to use equations, but you can find a way to classify your data by good or bad. For example for nball, we can think that you don't know the equation for the volume calculation, but can easily say if this dot stays within the sphere or outside. So the main idea of the Monte Carlo method is to use randomly casted points and compare the filling of the test figure with the unknown figure and then do a proportional ratio of their properties. Sphere volume2D sphere volumeFor the 2D sphere volume we use different words and call them circle surface, but generally speaking, this is the volume of this object. So, let's calculate it with the Monte Carlo technique. First of all, we will load numpy for calculation and matplotlib for presentation.
Next step  we need to define the initial parameters. Square size  square, which will have a circle inside. By comparing a number of random dots within the square and circle we can say about the relations of the square and circle area. N=100  number of random trials.
Now we do this Monte Carlo random experiment. to generate N random dots within the square, Then, for every tod we calculate the distance from the centre, and then if this distance is less than the standard radius, we will classify this dot as a dot inside the circle
Now we just need to compare our calculations. This is a very important step! Estimated area  number of dots inside the circle, divided by the total number of trials (or dots) and everything is multiplied by the area of the square. This is the main idea of the Monte Carlo method!
This will give you a very approximate volume of the 2D sphere, or an area of a circle. For more accurate results we need to use N=100_000.
This image shows an approximate result of the program calculation. Because I do not fix the start value for a random generator, your picture will be different. 5dimensional spherenothing will change if we will go to any other dimensions, so let's look at the 5D sphere. Really, the code will be almost the same, but I will show you some changes
The first change is to calculate the distance from the centre. It is a boring and not very scalable way to write squares for every dimension. We can use numpy features! So, we generate a set of random dots, and then for every dot we calculate the distance from the centre. It can be possible to do with numpy by doing this along the axis.
Also, I prefer to see, how results are approaching the ideal value and therefore I will recalculate the result every 100 trials
The last step is to show the result. Also, I do prefer to use a logarithmic scale along the Xaxis.
Ndimensional ballNow, because we are pretty confident in our MonteCarlo technique, let's apply it to the calculation of the volumes of spheres of different dimensions. The code will be the same with minor modifications. First of all, let's define parameters, with selfexplanatory code.
At this step, we will calculate the volume for each dimension from 2 towards Ndim (20 in our case). For each dimension, we will generate N dots in dimdimensional space. It is easy to do with the Numpy random.uniform function. Then we will check the location of every dot follow our algorithm as usual, and store the results.
And now, let's plot the results.
In fact, it is not very correct to compare the volumes of spheres of a different dimension, because the units are different, and depends on the dimension of the system. But you can compare coefficients and also see, that the MonteCarlo method can produce good results without knowing the exact equations. These are absolutely amazing results!

Last 10 artitles
9 popular artitles


© 2020 MyCoding.uk My blog about coding and further learning. This blog was writen with pure Perl and frontend output was performed with TemplateToolkit. 