test-quote

quote.md

---

template: quote.html title: "Additive manufacturing" source: "A Third Industrial Revolution | The Economist" url: http://www.economist.com/node/21552901 date: 2013-02-26 modified: 2015-05-04 review: false

#machine #things #manufacturing

And at the most recent EuroMold fair, last November, another group of machines was on display: three-dimensional (3D) printers. Instead of bashing, bending and cutting material the way it always has been, 3D printers build things by depositing material, layer by layer. That is why the process is more properly described as additive manufacturing. An American firm, 3D Systems, used one of its 3D printers to print a hammer for your correspondent, complete with a natty wood-effect handle and a metallised head.

This is what manufacturing will be like in the future. Ask a factory today to make you a single hammer to your own design and you will be presented with a bill for thousands of dollars. The makers would have to produce a mould, cast the head, machine it to a suitable finish, turn a wooden handle and then assemble the parts. To do that for one hammer would be prohibitively expensive. If you are producing thousands of hammers, each one of them will be much cheaper, thanks to economies of scale. For a 3D printer, though, economies of scale matter much less. Its software can be endlessly tweaked and it can make just about anything. The cost of setting up the machine is the same whether it makes one thing or as many things as can fit inside the machine; like a two-dimensional office printer that pushes out one letter or many different ones until the ink cartridge and paper need replacing, it will keep going, at about the same cost for each item.

visualization.md

1

Well-designed domain visualizations:

• Provide an ability to comprehend huge amounts of data on a large-scale as well as a small-scale.
• Reduce visual search time (e.g., by exploiting low level visual perception).
• Provide a better understanding of a complex data set (e.g., by exploiting data landscape metaphors).
• Reveal relations otherwise not noticed (e.g., by exploiting perception of emergent properties).
• Enable a data set to be seen from several perspectives simultaneously.
• Facilitate hypothesis formulation.
• Are effective sources of communication

2

One major key element of any successful visualization is to exploit visual perception principles

3

Visualizations help an increasingly diverse and potentially non-technical community to gain overviews about general patterns and trends and to discover hidden [semantic] structures

4

Ben Shneiderman at the University of Maryland proposed a mantra to characterize how users interact with the visualization of a large amount of information: Overview, Zoom-in (Filter), and Details on Demand (Shneiderman, 1996)

5

Visualizations of knowledge domains can help to assess scientific frontiers, to forecast research vitality, to identify disruptive events/technologies/changes, and to find knowledge carriers

6

The desire to examine large information spaces on small displays with limited resolution leads to the development of different focus and context techniques that enable users to examine local details without losing the global structure

7

The general consensus in relevant fields such as information visualization and geographic cartography is that multiple maps are preferred to a single map whenever possible. This is because each map may show different insights from the same data set

Two major problems in communicating information:

• (1) multivariate data need to be displayed on the two-dimensional surface of either paper or computer screen and
• (2) large amounts of data must be displayed in a limited space with limited resolution. The first problem is tackled by applying mathematical dimensionality reduction algorithms to map n-dimensional data into a 2-D or 3-D space. The purpose of these algorithms is to place objects that are similar to one another in n-dimensions close to each other and to place dissimilar objects far apart. This process is also called ordination. Cluster techniques can be used to further group similar objects together. Commonly used techniques are presented in section 4. The second problem is typically minimized by applying interaction (panning, filtering) and distortion techniques (fisheye) as discussed in section 4.4