Mailman 3 July 2006 - Edu-sig

Crunchy Fibonaccis
by kirby urner July 28, 2006

July 28, 2006

We've many times visited the Fibonaccis on this list. In the general case, we start with any two numbers (the seed pair) and add them to get the next in series, and so on, always adding the most recent two to get the next in line. Obviously, the Python generator is perfect for this, as it remembers state between invocations (we invoke with .next() -- the standard method for iterators in general). New with Python 2.5 comes the ability to use the pause at "yield" to send in some new grist for the mill -- let's say a new (a,b) seed pair. Here's what that looks like: >>> def fibolater(a,b): while True: t = (yield a) if t is not None: a,b = t a,b = b, a + b >>> f = fibolater(0,1) >>> f.next() 0 >>> f.next() 1 >>> f.next() 1 >>> f.next() 2 >>> f.next() 3 >>> f.next() 5 >>> f.send((12,13)) 13 >>> f.next() 25 >>> f.next() 38 >>> f.next() 63 The punch line with the Phi-bonaccis is f.next()/f.next() -- and yes, that's legal Python -- approaches 1/phi. >>> from __future__ import division >>> f.next()/f.next() 0.61771561771561767 >>> f.next()/f.next() 0.61798753339269819 >>> f.next()/f.next() 0.61802721088435375 >>> f.next()/f.next() 0.6180329998700792 >>> f.next()/f.next() 0.61803384447421961 >>> f.next()/f.next() 0.61803396770035635 >>> f.next()/f.next() 0.61803398567880852 Note: phi = (1 + pow(5, 0.5))/2 Related reading: http://mathforum.org/kb/message.jspa?messageID=3867841 (re the NCLB Polynomial from my Pentagon Math). Kirby

3 3

Signature Building Blocks
by kirby urner July 26, 2006

July 26, 2006

If there's a signature Fuller polyhedron, it's probably the Coupler: >>> import syngeom >>> mypoly = syngeom.Coupler() >>> mypoly.volume 1 >>> mypoly.isaspacefiller True Some immediately object no fair gluing together already known pieces and *naming* them. But it's not a priority game, it's a getting work done game, and we need to call them something or other, so why not Coupler, since it's already nicely sorted out: >>> amightyMite = syngeom.Mite() >>> amightyMite.volume 0.125 >>> amightyMite.modules [A,-A,-B] >>> [i.volume for i in amightyMite.modules] [0.041666666666666664, 0.041666666666666664, 0.041666666666666664] MITE is an acronym from MInimum TEtrahedron, an irregular tetrahedron with a 90-90-90 XYZ corner that's able to fill the 8 corners of the XYZ coordinating apparatus, thereby putting the Coupler center right at the origin (0,0,0). However the Coupler is a bit deceptive in that it's really more of an interface between neighboring rhombic dodecahedra: >>> ballcase = syngeom.RhDodeca() >>> ballcase.volume 6 Plus it works to see it as coupling around all three equators in various ways (the Coupler being an octahedron): >>> mypoly.faces 8 At my web site: http://www.grunch.net/synergetics/modules.html More in my blog: Half-Coupler: http://worldgame.blogspot.com/2005/11/half-coupler.html Coupler: http://worldgame.blogspot.com/2004/12/more-geometry-of-thinking.html Kirby

1 1

Recanting
by kirby urner July 25, 2006

July 25, 2006

OK, so a few posts back I talked about moving away from writing Pythonic math and heading for the green fields of television. Certainly I've taken that tack in my blog some, but I can't stop myself with the gnu math writing. So for example, regarding this hexapent jag: I framed it as a CP4E web page today, at my OCN website: http://www.4dsolutions.net/ocn/hexapent.html I'm trying to escape the heat by retreating to the basement, and using the Edubuntu box (Jennifer). Portland is having a heat wave at the moment. I hope it gets cooler for OSCON, as Portland summers are usually pretty comfortable. I like those geek out-of-towners to enjoy their visits, in addition to us locals. Yes Michael, thanks for reminding me about the OSCAMP option. Good thinking. Kirby

3 11

Python Editor / IDE for HS students
by Kevin Driscoll July 24, 2006

July 24, 2006

I'm preparing a new diskimage for a WinXP-based computer lab and deciding the editor or IDE I'll use to teach Python this year. I always include my preference, Vim, but typically my students have used either Textpad or IDLE. Although I like Textpad, I want my lab to be as free as possible so I'm leaning towards SciTE. What are your experiences with editors / IDEs and Python? I've used Eclipse with Java, how is the PyDev plugin? Is there a good editor with method completion and collapsible codeblocks? Flexibility is valuable, because they'll also be using this editor for XHTML / CSS but Python is the most demanding unit. Kevin

4 6

Project HexaPent
by kirby urner July 23, 2006

July 23, 2006

I think it's important that the civilian open source community establish beach heads in as many pioneering areas as we dare to boldly go, otherwise we surrender to that old urban legend: that the military does all the real R&D while civilians just get the "spin offs". That may have been true in Da Vinci's day (depends on whether we label him civilian), but today the leading edge is so often right out in the open, where everyone can see and document for themselves, that civilians have what it takes to do their own innovating, minus this indirect pass through a killingry chapter (i.e. make it kill first, then commercialize it (often a very inefficient way to go)). Along these lines, I've opted out of some too-stupid-for-words military projects, but retain access to the civilian aspects. I'm speaking of Fuller's Geoscope in particular, now enhanced with new thinking from Glenn Stockton's group (of which I'm a member), Patrick Barton and others (see wwwanderers.org & isepp.org -- the former especially in need of some maintenance). I'm finding the new wxPython book very readable and a new romance with that GUI development tool has commenced on my end. I've downloaded the latest beta of Python 2.5, and the corresponding wx, and am happy as a clam (are clams really happy?), relearning some of these ropes, plus discovering new ones. One feature of wx that most excites me is this ability to have non-rectangular top windows. Like the music industry, I don't want all my most important tools forced into 90 degree motifs. In particular, our HexaPent Matrix Disply (HMD) should be hexagonal in conception, with obvious click points for N, NE, E, SE, S, SW, W, NW (going clockwise). What's on your display to begin with is your "home hexagon" (a region of the earth's surface). When you click in any of these directions, you go to a neighboring hexagon (through points west and east, across edges in the six other directions, com sa: oooo SAT sig N _ _ _ / \ / \ \ x / \ _ _ _ / S <zoom> <rotate> x = self ). The hand-held units will be somewhat different, more like iPods in some respects, but the hexagonal motif is the same. The link to Fuller's Geoscope: the hexapent design, dual to an omnitriangulated geodesic, characteristic of turtle shells and fullerene, other natural geometries (not patentable); the use of same of a non-Mercator view of the earth; individual tilings, consisting of irregular (but many almost regular) hexagons and 12 pentagons (radially at the vertices of an original anchoring icosahedron and/or pentagonal dodecahedron). Here's an excerpt of a recent email to Glenn: """ The problem with the XY lat/long approach is it produces singularities at the poles, with the fictitiously square regions mostly occuring in places like Kansas (very far from Oz). But in reality, none of those XY squares are really square, because the Earth isn't really a plane, but a ball. Those squares were approximations. When we go from spherical to flat displays, our mathematics has to accomodate these approximations by means of various tradeoffs and compromises (the so-called "cartographic projections"). Another solution to the lat/long problem is to complement it with a more spherically sophisticated grid patterns, the hexapent being a good example, which is where the Global Matrix comes in. ... Since Fuller's day, the hardware for projecting in the round has come a long way, coming from the planetarium tradition. Joe Clinton, an early student of Fuller's, has made a distinguished name for himself in this arena. The geoscope idea itself has never been lost. Thanks to a convergence of technology and data, the applications for real time spherical and flat screen monitoring have never been greater. We're in a time of accelerating interest in this whole field of GIS/GPS and global data displays. Which is where the hexapent comes in, as an alternative to lat/long or XY, as a potentially more spherically adept geometry. It's users will have a strategic advantage over those clinging to the old ways. """ It's going to take me awhile to get up to speed on wx with all this, but in making this an open source project, I'm at least saving a stub on which others might build (without waiting for me if necessary, although you don't get far with a reputation for rip off artist, even when the source is open (something many gnubees fail to sufficiently appreciate (so we cut them some slack))). More info: http://mathforum.org/kb/thread.jspa?threadID=1416622 If any of your OSCON people want to visit the Linus Pauling House on Hawthorne, and/or have beers in my neighborhood, give me a holler. The only OSCON access I've purchased to date is an Exhibit Hall pass, and given the nearness of the deadline, that's probably going to be it for this year. Kirby

2 3

More Pythonic Soccer Ball Geometry (was HexaPent)
by kirby urner July 23, 2006

July 23, 2006

So here's a simple CS approach to the first/simplest hexapent: the soccer ball (at least until Adidas messed with it). You've got the 12 pentagons and the 20 hexagons. This lesson plan should fly easily in cultures which already have that image deeply engrained (USAers still lag in this respect, probably always will, given their "pig skin" fascination). I'm using what I'd call intermediate-to-beginner Python here. No class definitions, everything procedural. However, I do use globals, and take for granted the set, list comprehension, slicing. Lots of basic syntax is already in play. It's a script for building a data structure, step by step, starting from an encoding of the icosahedron's 20 faces, in terms of 3-tuples. The letters A-L label the 12 vertices. Those 12 vertices transform into the 12 pentagons of any hexa-pent, leaving the triangles to become 20 hexagons (Glenn calls this a 12/20). Then I go through a lot of selecting and sorting with the following goal in mind: build a dictionary such that every face has a name (the keys) and every face is paired with its five or six neighbors (the values). Note that the tricks I use to build this data structure (globaldata) aren't necessarily going to scale when we move to the higher frequency hexapents (12 pentagons, more hexagons). Here' we go: =================================================================== # here is a topological icosahedron, expressed as openings (faces) # Note: I recycled this from my rbf.py but changed the letters to # something friendlier baseicosa = [('A', 'C', 'I'), ('A', 'L', 'B'), ('B', 'D', 'K'), ('D', 'C', 'J'), ('I', 'E', 'G'), ('G', 'J', 'H'), ('K', 'F', 'H'), ('L', 'F', 'E'), ('A', 'B', 'D'), ('A', 'C', 'D'), ('A', 'L', 'I'), ('I', 'E', 'L'), ('C', 'I', 'G'), ('C', 'G', 'J'), ('L', 'B', 'F'), ('K', 'B', 'F'), ('D', 'K', 'J'), ('H', 'K', 'J'), ('F', 'E', 'H'), ('G', 'E', 'H')] # we will consider these the names of 12 pentagons then add a # 'soccer ball' pattern of 20 hexagons, named for the pentagons # at their 3 alternate edges, e.g. hexagon ACI. # Every hexagon will have six neighbors: three pentagons with # one letter in common, three hexagons with two letters is common # Every pentagon with have five neighbors: five hexagons with # one letter in common. pents = ['A','B','C','D','E','F','G','H','I','J','K','L'] # 20 print "Pents: %s" % pents def gethexes(): thelist = [] for t in baseicosa: thehex = ''.join(sorted(t)) thelist.append(thehex) return thelist hexes = gethexes() print "Hexes: %s" % hexes # and let's get the edges while we're at it. Every edge of every # triangle is an edge. We'll use the set data structure to police # against duplicates def getedges(): theset = set() for t in baseicosa: for i in (t[:2], t[1:], t[-1]+t[0]): theset.add(''.join(sorted(i))) return sorted(list(theset)) edges = getedges() print "Edges: %s" % edges # now we'd like a global dictionary in which every pentagon is paired # with its 5 neighbors. These will simply be all hexagons with # a letter in common. globaldata = {} def buildpents(): result = {} for p in pents: neighbors = [] for h in hexes: if p in h: neighbors.append(h) result[p] = neighbors return result result = buildpents() print "Pent neighbors: %s" % result globaldata.update(result) # adding to our globaldata, will be all the hexagons paired with # their six neighbors: 3 pentagons, and 3 hexagons across an edge. # We detect edge sharing as a boundary condition, i.e. if a panel # has two (and only two) letters in common, it's across an edge def buildhexes(): result = {} for h in hexes: neighbors = list(h) # 3 pentagons for candidate in hexes: votes = 0 for i in range(3): if h[i] in candidate: votes += 1 if votes == 2: neighbors.append(candidate) result[h] = neighbors return result result = buildhexes() print "Hexa neighbors: %s" % result globaldata.update(result)

1 3

Python Pedagogy
by kirby urner July 23, 2006

July 23, 2006

Although we all agree to use 'self' as the "me object" proxy, it's not a keyword, and we could use a different stand in, e.g.: class Human (object): def __init__(me, name): me.name = name def __repr__(me): return 'Hi, my name is %s' % me.name >>> import subgenius >>> aguy = Human('Bob') >>> aguy 'Hi, my name is Bob' It occurs to me that I could test "me" in place of "self" as it's shorter, and as it might encourage a first person identification which I think needs to happen when tackling OO. We explored earlier on this list the difference between CivBuilder 3rd person games, like SimCity and Civilization IV (many others), and 1st person shooters e.g. Quake and Doom. Of course many games give both 1st and 3rd, though we should distinguish between two kinds of 3rd: 3rd as in "I am that character (avatar, action figure or whatever, as in 'Alice')" vs. "I have some god's eye view" (incorporeal flyer, as in Google Earth and most of those WarCrafty type games, also Sims). Likewise, I think when coming to think formally in terms of objects (vs. informally, which begins with the emergence of language), it helps to personally project a "self" into various household objects, houses themselves. Like in the movie 'Cars' we need to *become* a thing, then ask (in the first person): what are my behaviors/methods, what are my attributes/properties? It's a game of "who am I" (or "who I am") and is already a natural feature of childhood play (fantasy self projections). I'm thinking the word 'self', at least in English, is too '3rd person' in some ways, and looking down on a lot of objects, each with a 'self', you have only a god's eye view. However, the grammar around 'me' is different -- there's only one of them (one first person), and therefore thinking "me" promotes a kind of first person instrospective attitude. And we *want* that, as an option, when modeling in OO. So I'd be accomplishing two things in this lesson (involving temporarily substituting "me" for "self" in some class definitions): (a) I'd be communicating the subtle teaching that 'self is not a keyword in Python' and (b) helping with the subliminal process of personally identifying with various objects, in order to become a better object-oriented programmer. At the end of the lesson, I'd reinforce the canonical accepted 'self' (i.e. the god's eye view) as the proper one, but hopefully students would have taken to heart the point of this lesson. Note that I'm not proposing this as a "for kids only" type lesson plan experience. I've been brainstorming a lot about what a Computer Science for Liberal Arts Majors might look like (recent link below), and this whole idea of "point of view" is already standard fare in literature courses, as well as film theory. We can bridge to OO through this "pronouns" discussion. Note about 2nd person: many multi-player Internet games, plus single-user games, do have a "we" concept, i.e. you're a part of a team with a shared objective, up against other players, or up against the computer, as the case may be. Related topic: me.__dict__ is a good intro to the idea of a "personal namespace" in addition to basic Python -- a helpful notion in psychology and diplomacy, where world views may start far apart, but grow closer through growing familiarity with the others' operations. It's in the tradition of Leibniz to want to use some "machine language" as a basis for diplomacy (American Transcendentalism has echoes of that, e.g. in Fuller's 'cosmic computer' meme)). Kirby Liberal Arts Compsci (except selling as Maths in this context): http://mathforum.org/kb/message.jspa?messageID=4931704

6 21

Ann: Crunchy Frog 0.6
by Andre Roberge July 22, 2006

July 22, 2006

[note to edu-sig people: the 0.6 version looks much different from the 0.4.9. Much!] Crunchy Frog (version 0.6) has been released. Crunchy Frog is an application that transforms an html-based Python tutorial into an interactive session within a browser window. The interactive embedded objects include: * a Python interpreter; * a simple code editor, whose input can be executed by Python; * a special "doctest" mode for the code editor; * a graphics canvas which can be drawn upon using a simple api. * simple sounds can be generated and played with all of the above. A comprehensive set of examples are included in the package, as well as two "standard Python tutorials" (mini-sorting HowTo, and regular expression HowTo) which have been adapted for use with Crunchy Frog. Requirements: Python 2.4+ (it might work, but has not been tested with earlier versions) Elementtree; (a link is provided in the included docs) Firefox 1.5+ The website address is http://crunchy.sourceforge.net The files can be downloaded from http://sourceforge.net/project/showfiles.php?group_id=169458 André and Johannes

1 0

Daniel Sayre is out of the office.
by DSayre＠wiley.com July 21, 2006

July 21, 2006

I will be out of the office starting 07/19/2006 and will not return until 07/24/2006. I will have limited access to email while I am away. If you require immediate assistance, please contact Lindsay Murdock (lmurdock(a)wiley.com).

1 0

Re: [Edu-sig] Alan Kay - another one of his ideas
by Yoshiki Ohshima July 19, 2006

July 19, 2006

Hello, I'm Yoshiki Ohshima, working with Alan for a few years. I happened to be the only person who enjoyed the full show of Alan's keynote *in person*, so I guess I can clarify a few questions^^; Guido wrote: > Well, for a start I'd like to see what Alan Kay used in his keynote. > I'm guessing that's Logowiki. What Alan used in the talk was a slight variation of Squeakland Squeak. Virtually all code he used in the demo was available in the Squeak image downloadable at http://squeakland.org. Most of the content, including graphics, paintings, and "scripts" written in the tile-script are not there (only in Alan's image), but some eToys examples are available at the site, too. The dynamic interaction, full control of the graphics rendering, object manipulation with decent performance you saw in the demo is way beyond what today's JavaScript in a browser could deal with. The demo ain't Logowiki. The remote collaboration tool Alan used is called "Nebraska"; a low-tech screen sharing system in Squeak. Squeak running on a computer at CERN was sending the encoded drawing command to another computer in LA, and the user-events by Alan was again encoded and sent to the computer at CERN. (Nebraska is in the Squeakland image, too.) BTW, we are working on an eToys-like environment that is capable of server-based content authoring (asynchronous collaboration like wiki) as well as real-time collaboration based on a "better theory of collaboration". In regards to the "is the web good?" discussion. Sure, the web is good, but it could have been much better. For an historical account, the idea of "objects that talk to each other bi-directionally over the network" weren't new by the time HTTP was created. In fact, when the word "hypertext" was invented, Ted Nelson surely wanted the things on the net connected bi-directionally. (And ACM Hypertext conference predates HTTP, etc., etc.) Today's JavaScript in a browser is not exactly a great way of doing stuff, but the fast computers now let us do reasonably interesting stuff. However, if we would have started with a "language for the Web", which were as dynamic as JavaScript, and its execution engine properly incoporated known implementation techniques by the time, we would have been a few steps further. (HotJava in JavaScript on a well written execution engine?) Of course, hindsight is 20/20. I couldn't have done it if I were there, but to think of a possible better and fun world that we might have missed, I feel that learning from the history helps a lot. And, it gives good insights on the directions from "here". -- Yoshiki Microsoft Windows helped to create this world where reasonably cheap computers are abundant, and "everybody" gets the access to the net. But there could be a better world, right?

2 2