Cody

Alfonso Nieto-Castanon

1
Rank
41
Badges
30825
Score

Activities

1 – 50 of 5,563

Alfonso Nieto-Castanon submitted a Comment to Solution 1355443

added a few more test cases to discourage look-up table solutions (by I liked your bin2dec idea to encode the board, I copied that in my testsuite now :)

on 23 Nov 2017 at 21:24

Alfonso Nieto-Castanon submitted a Comment to Solution 1354397

wow, I love everything about this solution, the equivalence with a 3x9 board, the BubbleSort strategy, even the 190-size :)

on 23 Nov 2017 at 15:37

Alfonso Nieto-Castanon liked Solution 1354397

on 23 Nov 2017 at 15:25

Alfonso Nieto-Castanon submitted a Comment to Problem 44340. Recaman Sequence - III

(otherwise a perfectly valid answer would be to use a seed = 1+n^2/2-n/2 which always results in a Recaman sequence with a 1 in the n-th position)

on 23 Nov 2017 at 15:10

Alfonso Nieto-Castanon submitted a Comment to Problem 44340. Recaman Sequence - III

that said, his general comment still stands that the solutions to this problem are often not unique (several different sequences will contain a 1 in the n-th position) so the problem statement perhaps should clarify that you are asking for the 'lowest integer' to start a Recaman sequence containing a 1 in the n-th position (or something along these lines)

on 23 Nov 2017 at 15:06

Alfonso Nieto-Castanon submitted a Comment to Problem 44340. Recaman Sequence - III

David sequence examples are incorrect, the series cannot have repeated 1's (by definition, it is possible to have repeated elements but not repeated 1's). For example, the sequence starting with 1 would be [1 2 4 7 3 8 14 21 13 22 12 23 11] instead of [1, 2, 4, 1, 5,10, 4,11, 3,12, 2,13,1] as David suggests (e.g. after [1 2 4], 4-3 appears already in the sequence so it jumps to 4+3 instead)

on 23 Nov 2017 at 10:37

Alfonso Nieto-Castanon submitted a Comment to Solution 1351541

if you are feeling overwhelmed by a particularly persistent cheater just know that you can ban a player from your own problems by adding to your testsuite something like: lines=textread('groupDist.m','%s'); id=str2num(regexp(lines{end},'\d+','match','once')); assert(~ismember(id,[123456789 234567890]),'banned player id'); (where the array in the ismember line contains the list of player ids that you want to ban from your problem -e.g. my player id is 1379371; please do not ban me :)

on 21 Nov 2017 at 23:16

Alfonso Nieto-Castanon submitted a Comment to Solution 1352215

scenario #1: you run into a problem that you really like but no matter how hard you try you do not seem to be able to figure out how to solve it. You are ready to "give up" but would *really really really* like to see other players' solutions and learn from them. Unfortunately Cody does not let you see other people's solutions until you provide one of your own. A freepass allows you to bypass this restriction (from time to time and with a little effort from your side so it does not break the logic/motivation of the game)

on 21 Nov 2017 at 19:02

Alfonso Nieto-Castanon submitted a Comment to Solution 1350166

and this?

on 21 Nov 2017 at 13:41

Alfonso Nieto-Castanon submitted a Comment to Solution 1348113

can you do better than this?

on 21 Nov 2017 at 13:39

Alfonso Nieto-Castanon submitted a Comment to Solution 1342154

added a few test cases to discourage this sort of solutions

on 20 Nov 2017 at 0:02

Alfonso Nieto-Castanon submitted a Comment to Solution 1348734

added a few test cases to discourage this sort of solutions

on 18 Nov 2017 at 15:54

Alfonso Nieto-Castanon submitted a Comment to Problem 44378. Five-dimensional maze

@Bryant, on second thought perhaps it would be nicer of us to wait until after the Cody5 deadline to implement the change to the testsuite suggested in my previous comment, if at all? (there is a relevant ongoing discussion in the 'Tautology' problem 44374 regarding the pros and cons of modifications to the testsuite, particularly for problems in the Cody5 groups; it is unclear whether some consensus will be reached but I thought I'd mention in case you were not aware of that)

on 4 Nov 2017 at 4:30

Alfonso Nieto-Castanon submitted a Comment to Solution 1325799

ha, will do :)

on 3 Nov 2017 at 20:42

Alfonso Nieto-Castanon submitted a Comment to Problem 44374. Tautology

@Daniel, and funny that you should mention, my apologies that this happened to you!! If I remember correctly you already had all cody5:hard problems solved on Oct 21st, despite your badge indicating a Oct 31st date now, so appropriate bragging rights to you :)

on 3 Nov 2017 at 20:28

Alfonso Nieto-Castanon submitted a Comment to Problem 44374. Tautology

@Daniel, that is true, but other approaches (e.g. @James random-sampling approach) would still fail on tests 15/16, or anything that looks like assert(isequal(tautology(strjoin(arrayfun(@(i)['~'*ones(rand>.5) char('A'+i-1)],1:26,'uni',0),'|')),false)) while still working just fine on less-taxing scenarios. The point is that the original testsuite did not include cases for this sort of extreme scenarios (to be precise the original testsuite only included two-variable A/B cases) so it feels a bit unfair to expect original solutions to be general enough to cover this sort of scenarios. For example, Jan could probably invalidate *all* of the current solutions if he considers "A" and "a" to be two different variables, and creates a few tests that include 52 mixed-case variables. I believe that is probably not a great idea in the context of Cody5, even though I think it would be a fantastic idea to create a new problem that asks you explicitly to solve this rather-difficult larger-number-of-variables case.

on 3 Nov 2017 at 20:23

Alfonso Nieto-Castanon submitted a Comment to Problem 44374. Tautology

I agree with Jean-Marie that Cody5 problems are somewhat special and perhaps we should be treating them a bit more carefully wrt rescoring. In my opinion it is somewhat unfortunate that Cody Badges can be lost after they are obtained (this happens when one of the problems gets rescored and all of your solutions to that problem fail). I believe this is sometimes fixed by Cody admins directly, but it still creates some lingering problems (e.g. badge events get removed from the timeline so they do not have a date/time). In addition, unless you come back to Cody often, it is simply very hard to notice when a problem has been rescored and one or all of your solutions now fail the new testsuite (e.g. you may come back to Cody and notice that your score has decreased, but it is very hard to know exactly why), which makes this a particularly sticky issue in a challenge like Cody5 with a definite deadline and a lot of new players. In any way, because of all of this, perhaps we, as problem creators, should be just a little bit more careful when dealing particularly with Cody5 problems just to make sure that players do not loose their achievements/solutions somewhat "unfairly". That does not mean, in my opinion, that we should not re-score at all (e.g. personally I still feel that it is perfectly fine to re-score as many times as you need in order to fix/improve your testsuite, as well as to rescore simply with the intention of removing cheat entries -e.g. look-up table solutions limited to testsuite cases only-), but it probably means that we should be careful (and perhaps a bit conservative) with testsuite changes just to make sure that we are not invalidating otherwise reasonable (even if not perfect) solutions. In this particular case, tests 15 and 16 probably lie right in this borderline scenario, where I can see how they are helping improve the testsuite, but they may also be removing somewhat-unfairly otherwise good solutions that simply did not consider this rather-extreme case of all 26 variables being simultaneously included in a logic statement, so perhaps these two tests could be removed or changed to some other case that is not so taxing?. Would love to hear your thoughts

on 3 Nov 2017 at 20:10

Alfonso Nieto-Castanon submitted a Comment to Solution 1324869

I believe this solution would fail on 'A|~B' (on the other hand, +1 for finding a trick solution in what I thought was a rather complete testsuite, at least for the A/B variables case!)

on 2 Nov 2017 at 16:45

1 – 50 of 5,563