and creative solution for The Most Numbers by veky - python coding challenges

Enable Javascript in your browser and then refresh this page, for a much enhanced experience.

and solution in Creative category for The Most Numbers by veky

checkio=lambda*a:len(a)and max(a)-min(a)

Feb. 20, 2014

Comments:

nickie on Feb. 20, 2014, 8:14 p.m.
One character shorter than my "obvious", if I counted right. Well done...

veky on Feb. 20, 2014, 9:14 p.m.
Yeah, and is powerful. And len is one char shorter than bool. ;-)

bryukh on Feb. 20, 2014, 11:07 p.m.
Clear?

veky on Feb. 20, 2014, 11:29 p.m.
Clearer than most of my code here. :-D You can recategorize it freely. But if I put it in puzzle, then I'll have to put everything in puzzle. :-]

bryukh on Feb. 20, 2014, 11:49 p.m.
Clear - Clear, Readable, Documented and Educational. Puzzle - Puzzle, Short, Obfuscated and Weird :-) It looks like "short" as for me.

veky on Feb. 21, 2014, 10:44 a.m.
As I said: ... then I'll have to put everything in puzzle. I told Oduvan, people consider my code short even if it has 5 lines. :-o :-D

Lingson on Feb. 27, 2014, 6:24 a.m.
I don't think any of your solutions would suit in the 'Clear' category :) It might be 'clear' for intermediate user, and would be very educational (in fact I learned quite a lot from your solutions) IF you would be kind to put more comments in your code for explanation (usually I can finally understand the code after reading the comments or your explanations to comments from other users). Everybody has unique coding style/strength. Maybe if your solutions match the style or area of someone's strength, it's easy to understand. But for example, myself, I never used lambda just because honestly I almost don't have any idea how to use it. I've read the documentations, can have a feel of why it works in your solutions, but using it myself? It's just not my style :)

veky on Feb. 27, 2014, 7:39 a.m.
<blockquote> I don't think any of your solutions would suit in the 'Clear' category :) </blockquote> I have a feeling you haven't seen all my solutions. :-] However, when I write clear code, usually somebody comes and points out it's not very fast. Well of course. :-P If by some luck I manage to write clear and fast code, they'd probably complain it was boring. :-D Comments are a frequent request. However, I don't think you've considered the implications. Most of my codes are so concise that real comments explaining all that's going on there, would be about 10 times longer than the code on the average. (I'm not exaggerating... see for example DOX on Xs and Os referee.:) And more important, there are various types of people, with various levels of Python skills, that have problems with various constructs in Python code. You yourself say you don't grok lambda, even if to me it seems like just a syntactic sugar in most of cases. To somebody else (not just me:-) - see solutions in Library 2.0), lambda is completely transparent. If I were to explain everything anybody might have a problem with, my two-liner would probably become a relatively thick book. :-P I do explain my code. Not in the code comments, but in the social comments below. You see, here on CheckiO, we have a unique benefit that people who invented code comments didn't have: we know when people struggle with something we've written, because they tell us. And then we can explain interactively, which is tremendously better than writing comments in the code, for all the reasons written above. Whenever you have a problem with something in a code I've written, just ask. I'll respond as soon as I can. As far as I can tell, nobody was disappointed by my explanations.

Lingson on Feb. 27, 2014, 8:03 a.m.
Of course I haven't see ALL of your solutions :) But quite a lot. How would I missed your solutions since it always shows up whenever we go to publications after solving a mission? LOL And it's quite fun to read your solution since it almost always puzzled me :p and would take sometime for me to understand it. Well, I said to understand it, doesn't mean I'm comfortable to use it myself. And like the saying 'Practice makes perfect', since I never practice the use of lambda, I would have difficulties to use it in my code (even last night, after not doing any python stuff and being absent from here for almost 2 months due to my works, I even forgot how to write the 'for', 'if' and other usual stuff correctly LOL). Used to be in programming field, but not currently, so I'm doing this just as a hobby and avoiding my mind becoming dull :)

Lingson on Feb. 27, 2014, 6:16 a.m.
Agreed, suits the Puzzle category better :)

veky on Feb. 27, 2014, 6:26 a.m.
Probably, but this whole thing is put upside down. Community should decide what category something fits into, not the author. For me, almost all of my solutions are Clear (I suppose for everybody else, too, their solutions are Clear, otherwise they wouldn't write them:). Also, a solution can be both speedy and clear, as it can be speedy and puzzle. Those categories are not well distinguished.

Lingson on Feb. 27, 2014, 6:32 a.m.
That's true too, but from the category descriptions, I think Clear here doesn't refer to ourselves but to most people or the community. But no complains from me, was just commenting :), since I have difficulties too in deciding which category suits the solutions that I make. For example, there are easy missions that's so obviously can be solved in 1-3 lines. Does it then goes to Speed category? Or still Clear? or even Puzzle? Argghhh.. ok now I'm confusing myself LOL

veky on Feb. 27, 2014, 7:17 a.m.
<blockquote> I think Clear here doesn't refer to ourselves but to most people or the community. </blockquote> Like I said. It should not be expected from us to decide what would community think about our code. Community is here, why not just ask them?

Lingson on Feb. 27, 2014, 7:41 a.m.
Maybe all solutions should have an option where user can decide which category it suits, and when a solution have at least X (lets say 5) people who give their voices, the solution automatically moves to the category with most votes? (I know it might makes things more complicated, but it's just an idea). The question is: would it be worth the trouble to code that on this site? Because even though some of us might be a bit critical about which category a solution should be in, but I think most of the other don't really care about it.

veky on Feb. 27, 2014, 7:45 a.m.
I know they mostly wouldn't care. But that just proves my point. Why should we care? There are few people that care. They can fight with their votes. I might be amused by that, but probably wouldn't care myself. It's their decision. I don't feel undervalued when somebody puts my code in Puzzle. Bryukh practically admitted the category was made for me. :-]

Lingson on Feb. 27, 2014, 7:50 a.m.
There goes my 4 likes: "Why should we care?" You have a strong point LOL.. Totally agreed. So let's drop this whole category debate :D

oduvan on Feb. 28, 2014, 1:01 a.m.
I agree that community can say(suggest) which is the better category for this solution, but author should decide how he wants others to judge their work. Would you be happy to see this code in your production? Same as how the community can't edit your solution but they can judge how good your code is and suggest something. I agree that your solution can perfectly fit into both the Clear and Speed categories. Right now I would recommend that you publish two solutions, one in each category. This way the community will decide how good a fit it is for either category. In that case we will not mix different discussion subjects. Take a look at this discussion from the Speed category: <a href="http://www.checkio.org/mission/most-numbers/publications/pheromo/python-27/first/share/5f0566c4a6f78035cc94636c26319048/">http://www.checkio.org/mission/most-numbers/publications/pheromo/python-27/first/share/5f0566c4a6f78035cc94636c26319048/</a> This is a reason why reddit decided to do subreddits and not tags for posts. They might also have the same topic posted in different subreddits.

veky on Feb. 28, 2014, 5:12 a.m.
I would not be happy to see almost any of these codes in my production. I usually don't work that way, but code in 3 layers: basic mathematic metaclasses for various things Python misses, with proofs of correctness; then libraries of classes implementing a kind of domain specific language for a problem I'm solving, and then the third layer, which just "connects the dots". You can see glimpses of that in my pbs function (used for example in Feeding pigeons) and Node class (used for example in Anagrams by stacks). But CheckiO isn't really optimized for layered code (nor should it be, probably). What I wanted to say is that putting code in categories according to some vague criteria which is actually in the eye of beholder, is just unnecessary bureaucracy that helps noone. And people just don't care. Right now, in Ghost age, you have a fastest possible solution in Puzzle, and a slowest possible solution in Speedy. Some people consider lambdas completely transparent, for you it is enough to put whole code in Puzzle. A person complains to me that using str.endswith in a task where it explicitly asks you to find out if some string ends with another is somehow "unclear" (probably because of lambda:-P), while their code uses a monstrosity c.split(d)[-1]=="" (and they put it in Clear). It's a nightmare. And on top of that is that social component, where for example everything I write will puzzle people, simply because they expect to be puzzled, "because Veky". :-D But if categories must stay for some political reason, then we should discuss possibility of tags, and that completely depends on how many there should be. Number of subsets of {Speedy, Clear, Puzzle} (and yes, I maintain you can have a Clear Puzzle:-P) is about 3 orders of magnitude less than the number of subreddits in use. :-]

Lingson on Feb. 28, 2014, 5:37 a.m.
The Speedy category, does it actually relate to the speed of the process? Because at first, my understanding was for short codes like one liner codes.

veky on Feb. 28, 2014, 6:03 a.m.
Kinda, but not quite. My understanding is that it is for "algorihtmically fast" solutions, even though their implementation in Python is slow as hell (see nickie's solution of median for example:).

Cjkjvfnby on Feb. 28, 2014, 7:59 a.m.
Cool. Do you have some public repos? What kind of code you wrote for production? I consider lambdas completely transparent. And according to Python concept assigning lambda to variable is bad style. For me clear code is code that will be clear to my colleagues. Example (based on real life experience) I wanted to replace Fizz on web page by Fuzz. I try to grep "Fizz" from code and find nothing. It takes me couple minutes to find "view" with code, instead of 5 seconds. Code was simple: function("BaFi#") + "zz"

veky on Feb. 28, 2014, 11:15 a.m.
<blockquote> Cool. Do you have some public repos? What kind of code you wrote for production? </blockquote> Unfortunately, no. Currently I'm writing software for a (kind of a) bank. Those things are pretty closed. :-/ <blockquote> I consider lambdas completely transparent. And according to Python concept assigning lambda to variable is bad style. </blockquote> First, PEP 8 is not "Python concept". It is a set of guidelines for writing Python libraries. But we've discussed this already. And second, there is a reason for that: debugging. Not much relevant here. :-) <blockquote> For me clear code is code that will be clear to my colleagues. </blockquote> Same here, but my colleagues are math and economic PhDs with LISP and ML background. Among them, I am the practical one. :-P <blockquote> Example (based on real life experience) I wanted to replace Fizz on web page by Fuzz. I try to grep "Fizz" from code and find nothing. It takes me couple minutes to find "view" with code, instead of 5 seconds. Code was simple: function("BaFi#") + "zz" </blockquote> My example is completely orthogonal. You go through your source and simply replace first "Fizz" you found with "Fuzz". Oops, now you have inconsistent code: "Fuzz" in one place, but "Fizz Buzz" in another. While I have spent a few minutes, you have a nasty bug you'll notice much later when it bites you.

Cjkjvfnby on March 1, 2014, 11:25 a.m.
Thanks for answer. You rest here from being an enterprise programmer :)

veky on March 1, 2014, 12:32 p.m.
Now you finally understood. :-D

oduvan on Feb. 21, 2014, 4:20 a.m.
the fact of using <pre class='brush: python'>checkio=lambda*a: </pre> instead of <pre class='brush: python'>def checkio(a): </pre> make it look like puzzle for me

veky on Feb. 21, 2014, 10:46 a.m.
First, * is not the difference: def also has to have checkio(*a). And look around Library 2.0, you'll find lambdas in Clear category (not from me:). If I'm guilty, they are too. :-P

oduvan on Feb. 21, 2014, 7:52 p.m.
<blockquote> First, * is not the difference: def also has to have checkio(*a). </blockquote> You are right. Typos

oduvan on Feb. 21, 2014, 7:54 p.m.
<blockquote> And look around Library 2.0, you'll find lambdas in Clear category (not from me:). If I'm guilty, they are too. :-P </blockquote> Guilty :) No... I just explained my vote here. I would rather see my version of definition of function

veky on Feb. 21, 2014, 10:15 p.m.
That's your right and I respect it. I'm just saying you're not consistent. :-P <a href="http://www.checkio.org/mission/binary-count/publications/ciel/python-3/first">http://www.checkio.org/mission/binary-count/publications/ciel/python-3/first</a> (for example)

oduvan on Feb. 22, 2014, 1:21 a.m.
Thank you for pointing me.

Cjkjvfnby on Feb. 26, 2014, 11:15 p.m.
Using lambdas instead of def in Clear category is bad practice.

veky on Feb. 26, 2014, 11:32 p.m.
Happy now? :-P

Cjkjvfnby on Feb. 26, 2014, 11:48 p.m.
What answer you expect from me?

veky on Feb. 27, 2014, 12:19 a.m.
I learned not to expect anything from you. We are simply too different. :-)

metabuddy on Feb. 26, 2014, 2:20 p.m.
can someone explain to me like I am 5, why is there len(a)?

veky on Feb. 26, 2014, 3:54 p.m.
So that min and max don't raise ValueError when called on an empty sequence. In newer Pythons you can use default keyword argument, but not here. Here you can use <pre class='brush: python'>max(*a,0)-min(*a,0) </pre> but that's a bigger hack than the above IMO. :-) This way, if a is empty, len(a) is evaluated and returned as 0. If not, "and" evaulates its second operand, which is the true result max-min. HTH.

metabuddy on Feb. 26, 2014, 3:57 p.m.
thank you

Lingson on Feb. 27, 2014, 7:46 a.m.
Aside from my other comments here about which category this solution would fit better, I have totally no clue about this solution. Putting the len, max, min aside, there is an 'and' in the code, so shouldn't it then return just True or False instead of the max(a)-min(a)?

veky on Feb. 27, 2014, 8:06 a.m.
No. That is a frequent misunderstanding of "and" and "or", that they return bool values only. They don't. I suppose you know that many things can be true or false without being them True or False: for example, 5, None, (3, 4), "",... So they can be input to and (and or, but everything I write dually holds for both of them). But let's see how would somebody implement "and" in a dynamic language like Python. What we know is that we must support short-circuiting. You know what's that? Very briefly, somebody should be able to write: <pre class='brush: python'>if 0 <= i < len(a) and a[i] < 6: ... </pre> without it raising an IndexError if i is invalid index for a. In essence, if first condition is false, we don't evaluate second one. (And that's why "and" is a keyword, it can't be a function: function must evaluate all its arguments at the time of call.) So, to evaluate a and b, we first evaluate a to see if it's true, and decide what to do next. We have an if statement that does precisely that. So, the first version looks like this: <pre class='brush: python'>if a: if b: return True else: return False else: return False </pre> But that's obviously too complicated, and not general enough at the same time. Look at the lines 2~5. They can obviously be written as just "return b". That not only is shorter, but doesn't forget information: for example, if b is some special kind of false (None, for example), it returns the same kind of false, not a universal False literal. So, now we have <pre class='brush: python'>if a: return b else: return False </pre> But that has the same problem as above, only in another half. If a is false, we would like to keep it instead of converting it to False. So the "real" code "and" is equivalent to is: <pre class='brush: python'>if a: return b else: return a </pre> Nice, symmetric, and in fact strictly powerful than what we had above. Now, with that semantics, do you see how the code works? (Hint: it uses short-circuiting in an essential way.)

Lingson on Feb. 27, 2014, 8:17 a.m.
Honestly, after reading the long explanations (which I understand, but just didn't see the relationship), I was more confused than before. But you said that and/or doesn't always return True/False, and I just tested it in interactive python and BOOM, now I get it :) AND always return the second value, and OR always return the first value, which obviously I didn't know before (or maybe forgot). Know everything make sense. Thanks.

veky on Feb. 27, 2014, 8:45 a.m.
You don't really understand, but if you're fine with it, I am too. To see why you're wrong: try <pre class='brush: python'>"" and () </pre> What do you think it will be? And then ask Python.

masero on Aug. 11, 2014, 8:33 p.m.
"and" : the first wrong value encountered is returned (0, [], "", False, ...) or, if both values are true, the last one is returned. "or" : the first true value encountered is returned or, if both values are false, the last one is returned. With several "and" or several "or", or with a mix of both, same pattern is applied and operations are made from left to right : a and b and c or d is equivalent to (((a and b) and c) or d). Is that correct ? I am missing something ?

veky on Aug. 11, 2014, 10:30 p.m.
Not much, but there is different precedence. For example, a or b and c is a or (b and c).

masero on Aug. 11, 2014, 10:43 p.m.
Oh, so "and" takes precedence over "or", I didn't know it, thanks :-)

veky on Aug. 11, 2014, 10:56 p.m.
It makes sense, once you see that "and" is boolean multiplication, while "or" is boolean addition. ;-)

masero on Aug. 11, 2014, 11:01 p.m.
Yes completely! Thanks for the clarification

Lingson on Feb. 27, 2014, 8:57 a.m.
Sorry, there was a mistake in my previous post. What I meant was assuming both values are right/true. But yes, it's good that you point it out because I think I got the wrong understanding before. After your question, I think I get it now.

veky on Feb. 27, 2014, 9:05 a.m.
And now read just the last code block in my previous comment. <pre class='brush: python'>if a: return b else: return a </pre> That's "and". Plain and simple. It can even be written as an inline conditional: <pre class='brush: python'>b if a else a </pre> Now write "or" as a homework. ;-)

Lingson on Feb. 27, 2014, 9:16 a.m.
<pre class='brush: python'>if a: return a else: return b </pre> :)

veky on Feb. 27, 2014, 11:56 a.m.
Bravo. Now you understand. :-) Next level: what is the difference between <pre class='brush: python'>b if a else c </pre> and <pre class='brush: python'>a and b or c </pre> ?

Lingson on Feb. 27, 2014, 4:28 p.m.
I know that: <pre class='brush: python'>b if a else c </pre> equals: <pre class='brush: python'>(a or c) and b </pre> but the second one: <pre class='brush: python'>a and b or c </pre> I'm not totally sure how to write the short way. Is it correct that it equals: <pre class='brush: python'>b if b if a else c </pre> since the translation that I know how to write is: <pre class='brush: python'>if a: if b: return b else: return c else: return c </pre>

veky on Feb. 27, 2014, 7:36 p.m.
I would very much like to know from where do you "know" that, since it's wrong. For example, <pre class='brush: python'>>>> a, b, c = False, False, True >>> b if a else c, (a or c) and b (True, False) </pre> In fact, it has the same kind of problem (on the other side) that my code "a and b or c" has, only it has parentheses extra. And no, "b if b if a else c" is syntactically incorrect, since number of ifs must be equal to number of elses in an expression. What you write in long form would in short form be <pre class='brush: python'>(b if b else c) if a else c </pre> and yes, that would be kinda equal to "a and b or c" (assuming they don't change their values during computation). But the problem remains: what is operational difference between that one and "b if a else c"? And why it is usually not a problem? (Specifically, why it is not a problem on my solution "Shortest" for "All in a row"?:) And when you answer that, make the same thing for (a or c) and b: long and short form, find operational difference (above you have concrete values of a, b, c for which they are different), and compare all three: <pre class='brush: python'>b if a else c a and b or c (a or c) and b </pre> :-)

Lingson on Feb. 28, 2014, 5:32 a.m.
Ok, I was careless. I said I 'know' because the long version of <pre class='brush: python'>b if a else c </pre> would be <pre class='brush: python'>if a: return b else: return c </pre> And my logic for (a or c) and b was that if a is true, it doesn't check c and goes directly to b (returning b), but if a is false, it checks c. If c is also false, then it doesn't need to check b since a and c are both false. But I did forget to consider if c is true, it would still continue to b. At first glance I had a thought that <pre class='brush: python'>b if a else c </pre> would be equal to <pre class='brush: python'>a and b or c </pre> but then, the first code would return b whether b is true or false. As for the second one, if both a and b are false, it would still goes to c. That's cause my confusion to change it with and/or statement. So I think it should be: <pre class='brush: python'>not a and c or b </pre> :)

veky on Feb. 28, 2014, 6 a.m.
If you say "not a", you have the same problem, only on different branch*. :-P Ok, here is the solution: <pre class='brush: python'>P: b if a else c Q: a and b or c R: (a or c) and b </pre> Q is equivalent to P, except in the case where b is false - same as you saw that R is equivalent to P, except in the case where c is true. But since there are much more true than false things in Python, people usually used Q instead of P where they wanted conditional expression, and P wasn't available (it wasn't until Python 2.5). Moreover, since they usually used it to provide some "interesting" values for variables, those values were often impossible to be false (especially b, since it is in the "a is true" branch), so code was working. Someone saw it, said "look, a nice way to code conditional expression", and used it in the code where it didn't necessarily work. That finally compelled Guido to add P construct to Python. It's ugly since "no new keywords" was a priority, but it works every time, as opposed to alternatives. :-D Now another question: we know Q and R are different in semantics. Find values of a, b, c where they really differ in value. :-) *: you could write "(a and b) or (not a and c)", and that, despite being bool-equivalent to P, has its own share of problems. Can you spot some of them? :-)

Lingson on Feb. 28, 2014, 6:23 a.m.
Just using the value from your example: <pre class='brush: python'>a, b, c = False, False, True </pre> makes Q and R differ in value. I don't understand how <pre class='brush: python'>not a and c or b </pre> would be the same as <pre class='brush: python'>(a and b) or (not a and c) </pre> ? If you try a, b, c = False, True, False then it will result differently. According to mathematic rules,shouldn't it (not a and c or b) be equivalent with: <pre class='brush: python'>(not a or b) and (c or b) </pre>

veky on Feb. 28, 2014, 11:07 a.m.
... despite being bool-equivalent to P ... not to what you've written. I never mentioned "not a and c or b"

Lingson on Feb. 28, 2014, 11:14 a.m.
Ok.. Now I'm lost again.. Somehow in my thought it was related to what I wrote, since you put it in the *comment. I guess I'm lost in translation (English is not my mother language, although I can say that my english should be above average).

xiongbiao on May 24, 2014, 11:31 a.m.
Good

marlonxteban on April 2, 2017, 5:26 a.m.
There is your like xD

veky on April 2, 2017, 9:51 p.m.
Thanks. :-)

___413 on May 23, 2017, 2:39 a.m.
Thanks for answer!

veky on May 23, 2017, 10:06 a.m.
You're welcome. :-)

DahliaSR on June 4, 2017, 6:38 p.m.
Very Cool! A nice example for an effective way to use lazy evaluation

veky on June 6, 2017, 2:10 a.m.
It's funny how many things people call lazy evaluation. Would you call "if" lazy evaluation, too? :-) "Lazy evaluation" is usually mentioned when calling a function, when you don't evaluate its arguments until they are "needed" (in some way). Python has no such thing (unless you count `"blah"` or `lambda:blah` as passing `blah` lazily:). `and` is not a function: it is a keyword, controlling flow in the same way other keywords like `if` or `while` do. The word you're probably looking for is "<a href="https://en.wikipedia.org/wiki/Short-circuit_evaluation">shortcircuiting</a>".

DahliaSR on June 6, 2017, 3:44 p.m.
Yes, short-circuit evalutation is what I meant, learned something new again. :)

StefanPochmann on March 17, 2023, 11:12 p.m.
One char shorter: ``` checkio=lambdaa:len(a)and max(a)-min(a) checkio=lambdaa:max(a:=a or[0])-min(a) ```

nickie on March 18, 2023, 12:28 a.m.
You're right. Two comments: 1. `:=` was introduced in Python 3.8. I guess you can rewrite many golfing solutions from older times this way. 2. I'm too old to play this game anymore... 😎

StefanPochmann on March 18, 2023, 12:43 a.m.
Excuses!!!

veky on March 18, 2023, 12:30 p.m.
Ok, then this <pre class='brush: python'>checkio(*a):=-operator.sub(*minmax(a)) </pre> is one char shorter, and works in Python 5.8, since Python got single-pass minmax, autoimports of stdlib, toplevel walrus, and sane lambdaless function definitions. :-D

StefanPochmann on March 18, 2023, 2:16 p.m.
Ok but how does that not give you an error when `a` is empty? Surely `minmax` raises an error then, and none of the rest are catching it?

veky on March 18, 2023, 7:14 p.m.
Hm, interesting question. But a lot has happened on the way to 5.8. :-D My current explanation is that `minmax([])` returns `UnspecifiedNumericValue` on both min and max slots, and `sub` is intelligent enough to know that same (in the sense of `is`) `UnspecifiedNumericValue` cancels to zero.

StefanPochmann on March 18, 2023, 7:53 p.m.
Ugh, that `sub` behavior seems awful. I think it's more likely that CheckiO will have changed the expected outcome from `0` to `UnspecifiedNumericValue` by then. It might be more difficult to automtically update all solutions than what CheckiO did from Python 2 to Python 3, but we're talking about the future here. Presumably by then our AI overlords will have no trouble doing their pets this little favor.

Ice Base

0 %

You ClassRoom Demo is Ready With the new tool CheckiO ClassRoom, you will be able to analyze your students' progress, have your own leaderboard, change missions order on the map, and so on.