Latency Tricks. By: Clark Gaebel [web] [keybase] [email] Special thanks to reviewers: Ben Foppa [keybase], Norm Tasfi [web] [twitter] The Setup¶ Servers take time to respond to messages. In many systems, the time it takes is commonly quite low (tens of milliseconds), while rare events cause it to be very high (hundreds of milliseconds, or even seconds) in other cases. Systems that have this property aren't just annoying. This problem is so common that it has a folk solution: replicate the service to multiple servers, and send to all of them at once. How does this strategy affect our latency? Before trying to answer this question, a single server's latency must be modeled {we need to model the latency of a single server}.

Pareto Distributions¶ The probability distribution function of a Pareto distribution The cumulative distribution function of a Pareto distribution A pareto distribution has two parameters: xm and α. α is the "shape", which defines how fat the tail of the PDF is. Lyra. Composable Analytics | Flexible Business Intelligence.

Neural networks and deep learning. One of the most striking facts about neural networks is that they can compute any function at all. That is, suppose someone hands you some complicated, wiggly function, f(x): No matter what the function, there is guaranteed to be a neural network so that for every possible input, x, the value f(x) (or some close approximation) is output from the network, e.g.: This result holds even if the function has many inputs, f=f(x1,…,xm), and many outputs. This result tells us that neural networks have a kind of universality. What's more, this universality theorem holds even if we restrict our networks to have just a single layer intermediate between the input and the output neurons - a so-called single hidden layer.

The universality theorem is well known by people who use neural networks. In this chapter I give a simple and mostly visual explanation of the universality theorem. To follow the material in the chapter, you do not need to have read earlier chapters in this book. Two caveats Problem. An Interactive Introduction to Graphics Programming. This is a proposal and proof-of-concept for an interactive book about programming the graphics processor.

Modern computers come with two separate processors, two “brains”: The traditional CPU, the Central Processing Unit The newer GPU, the Graphics Processing Unit Almost all books and courses about programming only teach you how to program the CPU. GPU programming is esoteric. Yet there is a unique joy to programming the graphics processor. This book is intended to bring the wonder and joy of graphics processor programming to a wider audience. An Interactive Book This book will be interactive. The text will be extensively illustrated with manipulable diagrams and live code examples. On the left you have a graphic output and on the right you have the code that produced it. By experimenting with the code—by touching it—you can gain a much deeper understanding of how it works.

Try moving your mouse over the output. I am extremely excited about the possibilities of interactive books in general. The Thermodynamic Ice Bag | Copenhagen Suborbitals. Dear Readers, We’ve reached chapter 3 of the serial on the film cooling system of the TM65LE, and here we’ll take a closer look at the contents of the thermodynamic ice packet. Specifically, the options a rocket engine designer can chose from when it’s realized that the heat flux is far from optimal. In classic rocket engines, such as TM65LE, for a given fuel type there’s basically four methods for controlling the heat flux in an engine with a long burn time; radiation, ablative cooling, regenerative cooling, and film cooling. The methods must be used with great care since one can easily end up in a situation where you are either under or overcooling the engine.

More on that later. In the modern rocket engine school of thought there are additional cooling methods, but let us try first to consider the four classic methods separately. (Left) Merlin 1D rocket engine with radiation-cooled nozzle extension. (Left) 4th stage of a Peacekeeper missile with ablative cooled nozzle. Probabilistic Models of Cognition. IDE. IDE. Aurora (StrangeLoop 2013 demo) Visually stunning math concepts which are easy to explain.

Models help you understand why you disagree. Models help you understand why you disagree In a blog post yesterday, I advocated very strongly that rather than simply bloviating about political (and other) topics, we should instead build mathematical models to clearly express our thinking. There was a lot of disagreement on this. But quite a few people took the point to heart, and several actually decided to modify my model to illustrate their thinking. One in particular was Jeremy Scheff, who who forked my model and came up with his own. I’m going to briefly address his model to demonstrate how models help us have a rational discussion. Jeremy alters my model in several ways. In Chris’s model, basic income is paid to everyone. He is very explicit about this. This is a major point of disagreement between us.

A second place where his model disagrees with mine is that he believes a largeish number of people are not really disabled, and a Basic Income would induce them to engage in productive work: Developer API :: TheBigDB. Reinventing Explanation. The Babylonian Map of the World is one of the world's oldest extant maps, dating to 600 BCE.

It's a crude map, difficult to read at a glance, but fortunately an accompanying cuneiform text describes the features on the map, including Babylon, seven other cities, a canal, and a mountain: Modern maps are, of course, far better than this early map. They improve on it by taking advantage of the many map-making techniques developed since 600 BCE, such as: surveying to get proportions correct; projections to correct for the curvature of the Earth; methods to depict topographic features; and so on. Even ideas such as showing roads and nautical routes were not a priori obvious, but had to be invented. This agglomeration of ideas has turned maps into a powerful medium for thought.

Using this map, an ordinary person can walk into the Underground for the first time, and within minutes know how to find their way from place to place. Why go the trouble of constructing these prototypes? Storytelling with Data. Feb. 28, 2013 This is a condensed version of my opening keynote at the Tapestry Conference, which was held yesterday in Nashville’s beautiful Union Station Hotel. I’m writing this from memory so at best it will only be an approximation of what I said. Thanks to all the organizers and attendees for a great event. Update: a video of my talk is now available on the Tapestry blog. Storytelling with data I was happy to see that the theme of this conference was storytelling, because as we develop new ways of gathering, processing, visualizing and presenting data, we sometimes risk focusing so much on techniques that we forget to tell stories.

Storytelling is a basic human activity. Have an audience We naturally adjust our stories to fit our audience, but with graphics we don’t have that luxury. So whenever I look at a data set or try to build a graphic, I need to know who my audience is. I work at The New York Times, so the question is: Who is my audience? Design for someone else Respect the reader. Scrubbing Calculator. Bret Victor / May 31, 2011 This page presents an idea for exploring practical algebraic problems without using symbolic variables. I call this tool a "scrubbing calculator", because you solve problems by interactively scrubbing over numbers until you're happy with the results.

Background: This work assumes a Soulver -like environment for interactive arithmetic, and picks up where Soulver leaves off. If you haven't seen Soulver, go see it. You'll be glad you did. Scrubbing I recently designed a series of data graphics for a book . I had chosen heights for the top margin, bottom margin, and gap between the bars, and I needed to know the height of the bars themselves. And then perform algebraic manipulation to invert the equation and solve for x . Here's a different way of finding the answer: What happened here?

I performed the adjustment simply by dragging horizontally on the value ("scrubbing" the number). Notice that no algebra-style variables were used. Double Scrubble Connecting Unlocking. Nbviewer.ipython.org/url/norvig.com/ipython/Economics. Now let's describe the code to run the simulation and summarize/plot the results. The function simulate does the work; it runs the interaction function to find two actors, then calls the transaction function to figure out how to split their wealth, and repeats this T times. The only other thing it does is record results. Every so-many steps, it records some summary statistics of the population (by default, this will be every 25 steps). What information do we record to summarize the population? (Note that we record the median, which changes over time; the mean is defined to be 100 when we start, and since all transactions conserve wealth, the mean will always be 100.) What do we do with these results, once we have recorded them?

Data Structure Visualization. Calca - The Text Editor that Loves Math. Andart: Torus–Earth. Torus–Earth One question at Io9 that came up when they published my Double Earth analysis was "What about a toroidal Earth? " This is by no means a new question, and there has been some lengthy discussions online and earlier modelling. But being a do-it-yourself person I decided to try to analyze it on my own. Can toroid planets exist? It is not obvious that a toroid planet is stable. For all practical purposes planets are liquid blobs with no surface tension: the strength of rock is nothing compared to the weight of a planet. The equilibrium shapes of self-gravitating rotating ellipsoidal planets have been extensively analyzed. Similarly, equilibrium states of self-gravitating toroid shapes have been examined by Poincare, Kowalewsky and Dyson (Dyson 1893, Dyson 1893b). It looks like a toroid planet is not forbidden by the laws of physics. Directions I will call the two circles along the plane of rotation the equators (the inner and outer).

Toroid gravity Donut One day is 2.84 hours long. Hoop. Starship Radiators. Looking at the Daedalus starship, it might appear at first glance that the ship has no radiators. So why have these cumbersome appendages suddenly reappeared on the Icarus starships? Are they absolutely required? And if so, how do we design them, and how much of a mass penalty will they add to the ship? Figure 1. Why do we need this? Heat Sources The Daedalus starship’s thermal control requirements depended on the assumption that the neutrons produced by side fusion reactions were absorbed directly in the fuel pellets[i]. The situation for the designs that have emerged from the Icarus workshop sessions is dramatically different. The portion of the radiation that streams forwards, striking the tanks and the payload bay, will boil away the fuel and fry the payload in no time.

Fortunately, there are x-rays absorbers, such as tungsten, and neutron absorbers, such hydrogen and other light elements. Figure 2. Radiator Capacity To determine the absorbed power we can use Where: P=Source Power (W)