wu :: forums - Print Page


    
      
        wu :: forums
        (http://www.ocf.berkeley.edu/~wwu/cgi-bin/yabb/YaBB.cgi)
      

        riddles >> easy >> new york times problem solving test
        
(Message started by: towr on Aug 25^th, 2015, 8:39am)

Title: new york times problem solving test
Post by towr on Aug 25^th, 2015, 8:39am

Might be a bit too easy, but it's cute.

http://www.nytimes.com/interactive/2015/07/03/upshot/a-quick-puzzle-to-test-your-problem-solving.html

Title: Re: new york times problem solving test
Post by rloginunix on Aug 25^th, 2015, 9:45am

The perils of finding a formula for a finite series ...

Being a C programmer guess what "the rule" I came up with when is saw 2¹, 2², 2³.

Title: Re: new york times problem solving test
Post by rmsgrey on Aug 25^th, 2015, 11:25am

The real trick is not so much figuring out a suitable rule, as figuring out ways to test potential rules...

Title: Re: new york times problem solving test
Post by JiNbOtAk on Sep 1^st, 2015, 6:56pm

So, how many of us intentionally tried getting a "No" ? How many "No"s did you try to get before you were satisfied with the 'rule' ?

Title: Re: new york times problem solving test
Post by towr on Sep 1^st, 2015, 10:15pm

I can't remember how many "no"s I tried for. But after the initial "successive powers of two" phase I did test for "geometric sequence", "anything goes", "strictly monotonically increasing", "weakly monotonically increasing", "strictly monotonic", "weakly monotonic".

Title: Re: new york times problem solving test
Post by rmsgrey on Sep 2^nd, 2015, 9:02am

on 09/01/15 at 18:56:10, JiNbOtAk wrote:

So, how many of us intentionally tried getting a "No" ? How many "No"s did you try to get before you were satisfied with the 'rule' ?

It's been a while since I first came across it, but I got something like 4 "no"s before settling on "strictly increasing numbers" with a general approach of trying to find boundaries between "yes" and "no" rather than trying to get easy "yes"es

Title: Re: new york times problem solving test
Post by rloginunix on Sep 2^nd, 2015, 10:20am

I've got at least 4 "no"s: tried reversing the sequence as 8, 4, 2. Then 3, 9, 27. Then 27, 9, 3. Then rearranging the numbers: 4, 8, 2 and 9, 27, 3 etc.

I think that on this forum you are dealing with, in general, a rather specific bunch: as a C(Java) Solaris/CentOS programmer I can say that codewise - especially the more you move away from the raw academic compsci code - before you get to a "yes" and return you step through a fair amount of "no"s and after a while it's in our blood.

My family sometimes calls me "doom and gloom". Well. It's not doom and gloom. If it can go wrong - it will. So it's error/edge/boundary conditions/special cases handling ...

Title: Re: new york times problem solving test
Post by Grimbal on Sep 3^rd, 2015, 7:19pm

I did not start with what I should have, i.e. a global approach of searching yesses and nos.

Instead, I saw the pattern 2^k and tried to generalize it. I tried a*2^k and a*b^k. Only then I tried sequences that don't translate to a closed formula.

I quickly accepted "strictly increasing" because I realized the point of the exercise was best made with the simplest rule and monotonic was the simplest I could think of. I checked whether it is strictly increasing or not.

One explanation of what happens would be that we are usually asked to recognize a sequence. Once the idea of "powers of two" came to us, we have to first kill that idea before searching more globally.