🌐 Mailman 3 December 2008 - Python-list - python.org

Re: Implementaion of random.shuffle
by Hrvoje Niksic Feb. 20, 2009

Feb. 20, 2009

Steve Holden <steve(a)holdenweb.com> writes: > So it would appear that the developers chose the Knuth algorithm > (with a slight variation) for *their* implementation. Now you have > to ask yourself whether your surmise is genuinely correct (in which > case the documentation may contain a bug) or whether the > documentation is indeed correct and you are in error. That is a good question. The random module uses the Mersenne twister, which has a repetition period of 2**19937. The number of n-sized permutations of a list with n elements is n!, while each shuffle requires n calls to the PRNG. This means that to be able to generate all permutations, the PRNG must have a period of at least n! * n. In the case of MT, it means that, regarding the period, you are safe for lists with around 2079 elements. shuffle's documentation may have been written before the random module was converted to use the MT. 2**19937 being a really huge number, it's impossible to exhaust the Mersenne twister by running it in sequence. However, there is also the question of the spread of the first shuffle. Ideally we'd want any shuffle, including the first one, to be able to produce any of the n! permutations. To achieve that, the initial state of the PRNG must be able to support at least n! different outcomes, which means that the PRNG must be seeded by at least log2(n!) bits of randomness from an outside source. For reference, Linux's /dev/random stops blocking when 64 bits of randomness are available from the entropy pool, which means that, in the worst case, shuffling more than 20 elements cannot represent all permutations in the first shuffle!

8 7

is it possible to add a property to an instance?
by Darren Dale Feb. 20, 2009

Feb. 20, 2009

Does anyone know if it is possible to add a property to an instance at runtime? I didn't see anything about it in the standard library's new module, google hasn't turned up much either. Thanks, Darren

4 4

Re: Python is slow?
by Lawrence D'Oliveiro Feb. 15, 2009

Feb. 15, 2009

In message <mailman.1764.1222798473.3487.python-list(a)python.org>, José Matos wrote: > The gnuplot license is a free software according to FSF ... Not listed as one <http://www.fsf.org/licensing/licenses/index_html/view?searchterm=license>.

2 1

Re: Mathematica 7 compares to other languages
by Xah Lee Jan. 31, 2009

Jan. 31, 2009

On Dec 4, 6:09 pm, jason-s...(a)creativetrax.com wrote: > For the interested, with MMA 6, on a Pentium 4 3.8Ghz: > > The code that Jon posted: > > Timing[Export["image-jon.pgm", Graphics@Raster@Main[2, 100, 4]]] > {80.565, "image-jon.pgm"} > > The code that Xah posted: > > Timing[Export["image-xah.pgm", Graphics@Raster@Main[2, 100, 4.]]] > {42.3186, "image-xah.pgm"} > > So Xah's code is about twice as fast as Jon's code, on my computer. > > The resulting files were identical (and both looked like pure white > images; I thought they'd be interesting!). The result is not pure white images. They are ray traced spheres stacked in some recursive way. Here's the output in both my and jon's version: http://xahlee.org/xx/image.pgm also, note that Mathematica 6 has the function Normalize builtin, which is used in Jon's code deeply in the core. Normalize is not in Mathematica 4, so i had to code it myself, in this line: “norm=Function [#/Sqrt@(Plus@@(#^2))];”. This possibly slow down my result a lot. You might want to replace any call of “norm” in my program by the builtin Normalize. Also, each version of Mathematica has more optimizations. So, that might explain why on v4 the speed factor is ~0.2 on my machine while in v6 you see ~0.5. My machine is OS X 10.4.x, PPC G5 1.9 Ghz. ------------------------- let me take the opportunity to explain some high powered construct of Mathematica. Let's say for example, we want to write a function that takes a vector (of linear algebra), and return a vector in the same direction but with length 1. In linear algebar terminology, the new vector is called the “normalized” vector of the original. For those of you who don't know linear algebra but knows coding, this means, we want a function whose input is a list of 3 elements say {x,y,z}, and output is also a list of 3 elements, say {a,b,c}, with the condition that a = x/Sqrt[x^2+y^2+z^2] b = y/Sqrt[x^2+y^2+z^2] c = z/Sqrt[x^2+y^2+z^2] For much of the history of Mathematica, normalize is not a builtin function. It was introduced in v6, released sometimes in 2007. See bottom of: http://reference.wolfram.com/mathematica/ref/Normalize.html Now, suppose our task is to write this function. In my code, you see it is: norm=Function[#/Sqrt@(Plus@@(#^2))]; let me explain how it is so succinct. Mathematica's syntax support what's called FullForm, which is basically a fully nested notation like lisp's. In fact, the Mathematica compiler works with FullForm. The FullForm is not something internal. A programer can type his code that way if he so pleases. in FullForm, the above expression is this: Set[ norm, Function[ Times[Slot[1], Power[ Sqrt[ Apply[ Plus, Power [ Slot[1], 2 ] ] ], -1 ] ] ] Now, in this norm=Function[#/Sqrt@(Plus@@(#^2))] The “Function” is your lisper's “lambda”. The “#” is the formal parameter. So, in the outset we set “norm” to be a pure function. Now, note that the “#” is not just a number, but can be any argument, including vector of the form {x,y,z}. So, we see here that math operations are applied to list entities directly. For example, in Mathematica, {3,4,5}/2 returns {3/2,2,5/2} and {3,4,5}^2 returns {9,16,25}. In typical lang such as python, including lisp, you would have to map the operation into each lisp elements instead. The “Sqrt@...” is a syntax shortcut for “Sqrt[...]”, and the “Plus@@...” is a syntax shortcut for “Apply[Plus, ...]”, which is lisp's “funcall”. So, taking the above all together, the code for “norm” given above is _syntactically equivalent_ to this: norm=Function[ #/Sqrt[ Apply[Plus, #^2] ]] this means, square the vector, add them together, take the square root, then have the original vector divide it. The “#” is in fact a syntax shortcut for “Slot[1]”, meaning the first formal parameter. The “=” is in fact a syntax shortcut for “Set[]”. The “^” is a shortcut for “Power[]”, and the “/” is a shortcut for “Power[..., -1]”. Putting all these today, you can see how the code is syntactically equivalent to the above nested FullFolm. Note, that the “norm” as defined above works for any dimentional vectors, i.e. list of any length. In lisp, python, perl, etc, you'll have 10 or so lines. In C or Java, you'll have 50 or hundreds lines. For more detail on syntax, see: • The Concepts and Confusions of Prefix, Infix, Postfix and Fully Nested Notations http://xahlee.org/UnixResource_dir/writ/notations.html Xah ∑ http://xahlee.org/ ☄

20 40

Is this pythonic?
by ipytest＠gmail.com Jan. 27, 2009

Jan. 27, 2009

x.validate_output(x.find_text(x.match_filename (x.determine_filename_pattern(datetime.datetime.now())))) Is it even good programming form?

8 7

Mathematica 7 compares to other languages
by Xah Lee Jan. 24, 2009

Jan. 24, 2009

Wolfram Research's Mathematica Version 7 has just been released. See: http://www.wolfram.com/products/mathematica/index.html Among it's marketing material, it has a section on how mathematica compares to competitors. http://www.wolfram.com/products/mathematica/analysis/ And on this page, there are sections where Mathematica is compared to programing langs, such as C, C++, Java, and research langs Lisp, ML, ..., and scripting langs Python, Perl, Ruby... See: http://www.wolfram.com/products/mathematica/analysis/content/ProgrammingLan… http://www.wolfram.com/products/mathematica/analysis/content/ResearchLangua… http://www.wolfram.com/products/mathematica/analysis/content/ScriptingLangu… Note: I'm not affliated with Wolfram Research Inc. Xah ∑ http://xahlee.org/ ☄

30 69

PyQt, Cannot send events to objects owned by a different thread?
by Alexander Tuchacek Jan. 18, 2009

Jan. 18, 2009

hallo, i try to adress an qt object self.statusbar.showMessage("rtt %s...." % (n.rtt)) in an callback function, comming from a shared lib importet by ctypes, on osx this works wonderfull when i run the same code on linux (ubuntu gutsy), i get this core dump, ok, i understand that the problem is, that i cant speak to the qt thread, but why does it work on osx? shall i recompile python? pyqt or sip? without threads? could somebody give me a hint what to do best? how can i call a qt object in an c-lib callback? thanks for any help, alex ASSERT failure in QCoreApplication::sendEvent: "Cannot send events to objects owned by a different thread. Current thread 82c31e8. Receiver 'MainWindow' (of type 'MainWindow') was created in thread 81f5060", file kernel/qcoreapplication.cpp, line 269 Aborted (core dumped)

3 2

change only the nth occurrence of a pattern in a string
by TP Jan. 14, 2009

Jan. 14, 2009

Hi everybody, I would like to change only the nth occurence of a pattern in a string. The problem with "replace" method of strings, and "re.sub" is that we can only define the number of occurrences to change from the first one. >>> v="coucou" >>> v.replace("o","i",2) 'ciuciu' >>> import re >>> re.sub( "o", "i", v,2) 'ciuciu' >>> re.sub( "o", "i", v,1) 'ciucou' What is the best way to change only the nth occurence (occurrence number n)? Why this default behavior? For the user, it would be easier to put re.sub or replace in a loop to change the first n occurences. Thanks Julien -- python -c "print ''.join([chr(154 - ord(c)) for c in '*9(9&(18%.\ 9&1+,\'Z4(55l4('])" "When a distinguished but elderly scientist states that something is possible, he is almost certainly right. When he states that something is impossible, he is very probably wrong." (first law of AC Clarke)

6 5

Understanding search queries, semantics, and "Meaning" ...aren't we all looking for meaning?
by 5lvqbwl02＠sneakemail.com Jan. 14, 2009

Jan. 14, 2009

I have Section 4.4.1 of SICP rattling around in my head (database queries), and I'm trying to come up with a simple dictionary-based database in Python to represent circuit diagrams. My main confusion isn't one of implementation, but a matter of "big thinking", fundamentally, about the problem. Please don't suggest using a SQL library, as I'm looking to learn to fish, so to speak, and to learn a bit about the biology of fish. I've subclassed dict to hdict ("hashable dict") and rewritten the __hash__ function so I can include a dict into a set. Thanks to a previous poster here for providing that suggestion. A circuit component looks like this for example: comp1 = hdict(value=1e-6, footprint='0402', vendor='digikey') comp2 = hdict(value=10e3, footprint='0603', vendor='mouser') etc, etc. The database holds the values like this: db = dict() # normal dict db['value'] = ([1e-6, 10e3], [comp1, comp2]) #ordered set for fast lookup/insertion db['footprint'] = {'0402':set[comp1], '0603':comp2} # unordered uses normal dict for fast lookup db['vendor'] = {'digikey':[comp1], 'mouser':[comp2]} So basically the keys are the component parameters, and the values is the list of components with that value. Stuff that is comparable is ordered; stuff that is discrete is not ordered, using either 2-tuples or dicts, respectively. This allows extremely fast lookup of components based on their properties, with O(1) performance for non-ordered stuff, and O(log n) performance for ordered stuff (using bisect methods). The set operations are extremely fast, so I can do AND and OR operations on compound queries of this nature without worry. I need this speed not so much for selecting components when the user types in a query, but for when the mouse is hovering over the schematic and I need things to light up underneath, if the GUI is generating hundreds of mouse messages per second, and for inspector windows to appear quickly when you click, etc. If you have ever used Altium, you can see how effective this is in terms of creating a good interactive user experience. My question is what happens when I choose to search for NOT footprint='0402'. Should this return a blank list? This happens by default., and is in fact true: a blank list satisfies "not anything" actually. Should this return everything that is NOT footprint 0402 (ie returns 0603 components)? This *strongly implies* a pre-selection of *all* objects before performing the negating function, or of looking at the ordering of other queries in the overall search term, and then applying NOT to the results of another query. But I'm suspicious of a brute force preselection of all objects whenever I see a NOT, and anyway it messes up the clean query/combiner method of search I'm doing, and it requires an implied sequence of search, where I'm pretty sure it should not rely on sequencing. Even though this is single threaded, etc., the semantics of the query should not rely on ordering of the search term: footprint='0402' and NOT vendor='mouser' should return the same as NOT vendor='mouser' and footprint='0402'. So this is my philosophical quandary. I'm not sure what the correct thing is. In SICP they are using nondeterministic stuff which I don't quite get, so it's hard to follow. Also they are not using dictionaries and hashes, so I'm not sure if their generate-and-test method would work here anyway. Generate-and-test seems extremely inefficient. Can a wise guru please enlighten me? thanks Michael

3 4

sys.stdout.write()'s bug or doc bug?
by Qiangning Hong Jan. 14, 2009

Jan. 14, 2009

>>> u = u'\u554a' >>> print u 啊 >>> sys.stdout.write(u) Traceback (most recent call last): File "<stdin>", line 1, in <module> UnicodeEncodeError: 'ascii' codec can't encode character u'\u554a' in position 0: ordinal not in range(128) >>> type(sys.stdout) <type 'file'> >>> sys.stdout.encoding 'UTF-8' Quote from file object's documentation: encoding The encoding that this file uses. When Unicode strings are written to a file, they will be converted to byte strings using this encoding. ..... So, my question is, as sys.stdout IS a file object, why it does not use its encoding attribute to convert the given unicode? An implementation bug? A documenation bug?

4 3