22 votes

Happy Tau/2 day everyone!

15 comments

  1. [9]
    unkz
    Link
    I dunno, e^(tau/2)i +1 = 0 doesn't have the same cachet.

    I dunno,

    e^(tau/2)i +1 = 0

    doesn't have the same cachet.

    5 votes
    1. [5]
      RheingoldRiver
      Link Parent
      The tau version is e^(i * tau) = 1 or as the author points out, e^(i * tau) = 1 + 0i

      The tau version is e^(i * tau) = 1

      or as the author points out, e^(i * tau) = 1 + 0i

      6 votes
      1. [4]
        unkz
        Link Parent
        0i feels like cheating to me. I guess there’s an argument that the tau version without 0 at all is more succinct.

        0i feels like cheating to me. I guess there’s an argument that the tau version without 0 at all is more succinct.

        2 votes
        1. nosewings
          (edited )
          Link Parent
          e^(i * tau) = 1 is the more "honest" formula anyway, because it says exactly what it ought to: rotating by tau takes you all the way around a circle. 0i is also arguably not cheating there, since...

          e^(i * tau) = 1 is the more "honest" formula anyway, because it says exactly what it ought to: rotating by tau takes you all the way around a circle.

          0i is also arguably not cheating there, since the setting is complex analysis, so every number does have a real and an imaginary component.

          7 votes
        2. [2]
          ebonGavia
          Link Parent
          But then we don't have our additive identity 😭

          But then we don't have our additive identity 😭

          3 votes
          1. Minty
            Link Parent
            e^iτ - 1 = 0 close enough.

            e^iτ - 1 = 0

            close enough.

            1 vote
    2. [3]
      PuddleOfKittens
      Link Parent
      e^(tau/2)i = -1 makes perfect sense though - you're going halfway around the circle, so of course there's a "/2" in there.

      e^(tau/2)i = -1 makes perfect sense though - you're going halfway around the circle, so of course there's a "/2" in there.

      1 vote
      1. [2]
        unkz
        Link Parent
        It just feels less aesthetically pleasing.

        It just feels less aesthetically pleasing.

        1. RheingoldRiver
          Link Parent
          The "regular" version is saying, "a turn of the ratio of the circumference to the radius traces the circle halfway" The tau version says, "a turn of the ratio of the circumference to the diameter...

          The "regular" version is saying, "a turn of the ratio of the circumference to the radius traces the circle halfway"

          The tau version says, "a turn of the ratio of the circumference to the diameter traces the circle exactly once"

          To me the tau version is a lot nicer

          3 votes
  2. [3]
    em-dash
    Link
    My unpopular math opinion is that degrees and radians are both bad angle units. The common angle unit that normal people use should have been a full turn of a circle. A right angle is just a...

    My unpopular math opinion is that degrees and radians are both bad angle units. The common angle unit that normal people use should have been a full turn of a circle. A right angle is just a quarter of a turn.

    I like tau as a circle constant for the sole reason that you can kind of use it this way if you write radians as multiples of tau and pretend it's a unit.

    4 votes
    1. psi
      (edited )
      Link Parent
      The beauty of the radian is that it isn't really a unit at all. A radian is, by definition, just the angle subtended when the arc of a circle equals its radius. But since we know that we can...

      The beauty of the radian is that it isn't really a unit at all. A radian is, by definition, just the angle subtended when the arc of a circle equals its radius. But since we know that we can measure the arc length of a circle using s = θ r, that means that when s = r (i.e., when the angle is 1 radian), it is also the case that θ = 1 (per SI 2019, no units necessary!).

      This allows us to sensibly define Taylor series of functions that have angular arguments without needing a bunch of ad-hoc conversion factors, e.g. the tried and true

      sin(x [rad]) = x - 1/3! x^3 + ...
      

      instead of

      sin(x [deg]) = (x π / 180°) - 1/3! (x π / 180°)^3 + ...
      

      or

      sin(x [rev]) = (x / 2π)  - 1/3! (x / 2π)^3 + ...
      
      9 votes
    2. RheingoldRiver
      Link Parent
      Yep same, the closer an implementation can be to its underlying intuition, the better. Like understanding "measurement of an angle has to do with how far around the circle you go" is the...

      Yep same, the closer an implementation can be to its underlying intuition, the better. Like understanding "measurement of an angle has to do with how far around the circle you go" is the intuition, and so the implementation should be that angles are the same as fractions, with one whole turn being one. So one needs to be tau, not pi.

      2 votes
  3. aphoenix
    Link
    Tau vs Pi is such a silly argument. Obviously we should be using Pau as a compromise. Also if anyone is interested Pi won the debate 11 years ago because Matt is better at maths than Steve.

    Tau vs Pi is such a silly argument. Obviously we should be using Pau as a compromise.

    Also if anyone is interested Pi won the debate 11 years ago because Matt is better at maths than Steve.

    3 votes