[HN Gopher] Squares in Squares
       ___________________________________________________________________
        
       Squares in Squares
        
       Author : yowzadave
       Score  : 132 points
       Date   : 2023-02-15 19:13 UTC (3 hours ago)
        
 (HTM) web link (erich-friedman.github.io)
 (TXT) w3m dump (erich-friedman.github.io)
        
       | Recursing wrote:
       | Other fun packings: https://erich-friedman.github.io/packing/
       | 
       | I particularly liked "squares in circles", found 8 pretty
       | surprising
        
       | fabledAble wrote:
       | It reminds me of a real life "problem" I like to solve around the
       | holidays: what is the smallest cut of wrapping paper required to
       | wrap a given box. With a restriction in one dimension. I find
       | that, most of the time for cube-ish boxes, using a somewhat
       | diagonal orientation can result in a shorter cut of wrapping
       | paper used, and reduces the amount of tape needed to keep it on
       | the shape as well!
        
         | bee_rider wrote:
         | My mom did upholstery as a hobby; it only occurred to me in
         | adulthood that wrapping paper isn't designed so that the
         | patterns would just magically line up somehow.
        
         | layer8 wrote:
         | I'd like to read a systematic analysis of that.
        
       | sitkack wrote:
       | If you are fond of these geometric constructions you might like
       | the artist https://www.artsy.net/artist/victor-vasarely
        
         | mzs wrote:
         | I can recommend the museum as well:
         | https://en.vasarely.hu/vasarelys-periods/
        
       | gerikson wrote:
       | Really fun and counterintuitive arrangements.
       | 
       | The linked paper has more details
       | 
       | https://erich-friedman.github.io/papers/squares/squares.html
        
       | BeefySwain wrote:
       | I'd like to see the percentage improvement over the "naive"
       | approach (just stacking them edge-to-edge) for each of them. I
       | can't tell if the improvements are significant enough for this to
       | have obviously practical benefits.
        
         | layer8 wrote:
         | Obviously(?) the improvement decreases with increasing n. From
         | the table in [0], regarding the length (not area) of the
         | square, it is about 10% for n=5 and 7.5% for n=10, and 4.6% for
         | n=82.
         | 
         | [0] https://erich-
         | friedman.github.io/papers/squares/squares.html...
        
       | yowzadave wrote:
       | Originally saw this in a twitter thread by Daniel Piker:
       | https://twitter.com/kangaroophysics/status/16254239511563755...
       | 
       | As a former architect, it's tempting to see similarities in the
       | way architects do space planning: continually re-arranging the
       | various rooms/circulation elements to find the most efficient use
       | of space. Seeing how counter-intuitive these solutions are (and
       | the fact that most of them are not proved, only "found") makes me
       | think that space-planning may not be "solvable" computationally.
        
         | Swizec wrote:
         | > space-planning may not be "solvable" computationally
         | 
         | It isn't. The knapsack problem is NP-complete and roughly
         | translates to "Given a space of size X and a set of items of
         | sizes A, B, C, how do you pack X optimally?". Designing a room
         | layout seems like it would fall squarely into that, with
         | additional restrictions added on top to make it even harder.
         | 
         | https://en.wikipedia.org/wiki/Knapsack_problem
         | 
         | According to wikipedia we do have algorithms that are useful in
         | practice, but that's not quite the same as "definitely optimal"
        
           | Sharlin wrote:
           | I mean, NP-complete problems are perfectly solvable
           | computationally, and given the relatively small number of
           | rooms per floor that architects are concerned with, finding
           | the optimal solution with a computer should still take much
           | less time than it takes a human to come up with some local
           | optimum in the design space. The _real_ problem is actually
           | encoding all the constraints that an architect is really
           | working under, including such inherently difficult-to-
           | quantify aspects as aesthetics, ergonomics, and familiarity!
           | A solution that maximizes the available m^2 is useless if its
           | human users find it too weird.
        
           | layer8 wrote:
           | The knapsack solution space is countable and effectively
           | finite, and thus is solvable, even if not efficiently due to
           | NP-completeness. Space-planning has to work with a
           | potentially uncountable number of positions/orientations in
           | continuous space, and thus might be different.
        
             | turtledragonfly wrote:
             | > effectively finite
             | 
             | This made me scratch my head a bit. I'm aware of there
             | being different types of infinities, but to call a
             | "smaller" infinity "effectively finite" seems a bit odd.
             | I'm not up-to-date on math terminology, though.
             | 
             | In truth, I'm not quite sure what you mean by "solvable" vs
             | "not solvable" here, either. In the above discussion I
             | thought it meant "not tractable in a reasonable amount of
             | time" (i.e. NP-complete => "not solvable"). But you seem to
             | have a different definition you're working from.
             | 
             | Could you elaborate on what you mean by "effectively finite
             | => solvable" ?
        
         | NKosmatos wrote:
         | I saw it as well and immediately I thought of posting it here,
         | but as always I search before I post and it was reposted a year
         | ago, so I decided not to.
         | 
         | Glad you did post it though because it's a really nice
         | visualization and a very good page and deserves to be seen by
         | more people who like such things!
        
       ___________________________________________________________________
       (page generated 2023-02-15 23:00 UTC)