savoy082

Pt En < change language In this post I want to showcase the beginning of what I think will be a really cool project. After reading a really nice book (that Bill Gates himself recommends) on machine learning, I decided to experiment with genetic algorithms. For that matter, the main goal here will be to develop a genetic algorithm that teaches a vacuum cleaner how it should move in a dirty room to clean it in the best way possible (that is what is happening in the animation above). The first step is to define what I mean by room : a room will be a rectangular grid where each cell has a number from $0$ to $1$. A cell with a $0$ is perfectly clean and a cell with a number $1$ is as dirty as a piece of floor can get. Below we see a $5\times 5$ room with a robot (in red) already in the middle of the room: As I read in The Master Algorithm (and frankly I found it very enlightening), when one is going to use genetic algorithms we must previously define the structure of the solutio...

Random walks (check my older post here ) and Diophantine equations are two simple mathematical beasts. After a college seminar, I tried putting them together to make something neat, and came up with this: just pick a Diophantine equation, simulate a random walk, and try to see if the random walk went over any solutions! In this report I briefly go over how I came up with this, and show the code I wrote/some results. The Matlab code is all in here (I also have some Python code about random walks in general here , check my post !). In the image above, the blue line represents a path taken by the random walk and in orange/red the nodes tell us how far we are from a solution (the smaller the node, the closer we are). Note that this notion of proximity isn't given by actually computing the distance to a known solution, but just by looking at how similar the two sides of the equation are. The image below shows the evolution of three independent random walks trying to find an example ...

Think of a drunk man that continuously tumbles left and right, back and forth, with no final destination. That can be thought of as a random walk on the plane... Now imagine that the drunk man has a shorter leg and tumbles more to one of the sides: that is a biased random walk. Now imagine the drunk man can teleport to a nearby location. That is (kind of) a Lèvy flight ! All those are quite interesting to observe in motion and I implemented them in Python. (the code can be found here ). From those, only the tilted walks and the one with the trail are such that the particle's movement wraps around the borders (leaving the green area makes the particle appear in the opposite side, just like the snake from the game Snake). The image above is a screenshot of the execution of the random walk that leaves a coloured trail. I also implemented an animation where the screen gets progressively filled with circles of different colours, as if being splattered with paint. Not sure why... But it...

In this first post I want to share with you guys a piece of code I wrote to "solve" a problem where geometry meets optimization. I say "solve" because I didn't actually do anything that fantastic regarding the actual problem I address, but rather developed a small tool to help visualize the geometrical part of the problem. Even so, I do believe that for the smaller cases my tool can solve the problem. The problem is along the lines of: define an energy function whose value depends on the positions of points in a sphere, and now try to minimize/maximize it (depending on a parameter). That is it. I used my coding skills to write an algorithm that solves this when the number of points is small, and that lets me see the creation of the solution: I create a random possible distribution of points and then let them adjust themselves to their desired positions, hopefully reaching the desired minimum/maximum. In here you can find the report I wrote for this, in English....