Shop OBEX P1 Docs P2 Docs Learn Events
The Collatz Conjecture Is a Simple Problem That Mathematicians Can't Solve — Parallax Forums

The Collatz Conjecture Is a Simple Problem That Mathematicians Can't Solve

Ron CzapalaRon Czapala Posts: 2,418
edited 2016-08-11 16:24 in General Discussion
http://www.popularmechanics.com/technology/news/a22246/simple-problem-that-still-stumps-mathematicians/
Part of the beauty of mathematics is that seemingly simple patterns lead to much more complicated questions and theories. The Collatz conjecture, which is the subject of a new video from YouTube channel Numberphile, is the perfect example of a simple problem that even the greatest mathematical minds in the world haven't been able to solve.

The essence of the question is simple. We have an interesting pattern: If you take a positive integer, and then divide by 2 if it is even, or multiply by 3 and add 1 if it is odd, then repeat this process with the resulting number, eventually you will end up with the number 1. Professor David Eisenbud of the University of California, Berkeley, demonstrates this process, also known as the hailstone sequence, using the number 7 in the video.



«134567

Comments

  • Heater.Heater. Posts: 21,230
    I watched that yesterday. Love numberphile.

  • And the obligatory: https://xkcd.com/710/

  • It's tricks like this that get kids interested in math.
  • ercoerco Posts: 20,256
    edited 2016-08-11 01:12
    This math wizard conjures his baffling manner of sorcery on Youtube.

    Pick a number. Now subtract your number. You have....................... ZERO! :)

  • I'm sure Inter-Universal Teichmuller Theory will make short work of this.
  • localroger wrote: »
    I'm sure Inter-Universal Teichmuller Theory will make short work of this.

    It's been around since 1937 with no solution yet...
    The longest progression for any initial starting number less than 100 million is 63,728,127, which has 949 steps. For starting numbers less than 1 billion it is 670,617,279, with 986 steps, and for numbers less than 10 billion it is 9,780,657,631, with 1132 steps
    https://en.wikipedia.org/wiki/Collatz_conjecture

    http://io9.gizmodo.com/5813673/the-easiest-math-conjecture-it-took-74-years-to-prove (NOT)
  • MikeDYurMikeDYur Posts: 2,176
    edited 2016-08-11 15:27
    It would be fun to watch the Propeller do the math on the PST, slowed down to see what's happening.
  • Ron CzapalaRon Czapala Posts: 2,418
    edited 2016-08-11 15:53
    I wrote a little vbscript to do the job, but it overflows with large numbers...

    Save it as a .vbs file and double-click it to run.
    Option Explicit
    
    Dim val1, cntr, result, idata
    
      val1 = inputbox("Enter starting number",  "The Collatz Conjecture")
      val1 = CLng(val1)
      cntr = 0
      idata = val1	
      do
        if val1 = 1 then
          exit do
        end if
        result = clng(val1 Mod 2)
        if result = 0 then
          val1 = val1 / 2
        else
          val1 = (val1 * 3) + 1   
        end if
        cntr = cntr + 1
        idata = idata & "," & val1    
      loop
      msgbox idata,,cntr & " Loops"
    
  • Great start, Thanks Ron.

    I may have been able to come up with something like this but, it might have taken awhile. I don't have any problem setting up pst, I want to make it kid friendly, and they can check their progress.

    MikeY
  • Duane DegnDuane Degn Posts: 10,588
    edited 2016-08-11 16:57
    Edit: I made a mistake. I'll post it again in a couple minutes.
  • Duane DegnDuane Degn Posts: 10,588
    edited 2016-08-11 18:21
    MikeDYur wrote: »
    It would be fun to watch the Propeller do the math on the PST, slowed down to see what's happening.

    Try this program.
    CON
      _clkmode = xtal1 + pll16x
      _xinfreq = 5_000_000
    
      BAUD = 115200
      QUOTE = 34
      UPPERLIMIT = posx / 3 - 1  ' I'm not sure if this is a low enough value.
      
    OBJ
    
       Pst: "Parallax Serial Terminal"
       
    PUB Main | stepCount
    
      Pst.Start(BAUD)
    
      repeat
        waitcnt(clkfreq / 2 + cnt)
        result := Pst.RxCount
        Pst.Str(string(11, 13, "Press any key to begin."))
      until result
    
      Pst.RxFlush
      
      repeat
        stepCount := 0
        Pst.Str(string(11, 13, 11, 13, "Collatz Conjecture"))
        Pst.Str(string(11, 13, 11, 13, "Enter a positive interger between ", QUOTE, "1", QUOTE, " and ", QUOTE))
        Pst.Dec(UPPERLIMIT)
        Pst.Str(string(QUOTE, "."))   
        result := Pst.DecIn
        Pst.Str(string(11, 13, "You entered ", QUOTE))
        Pst.Dec(result)
        Pst.Str(string(QUOTE, "."))
        if CheckForInputError(result)
          next ' start loop from beginning
        repeat while result <> 1
          Pst.Str(string(11, 13, "Step # "))
          Pst.Dec(++stepCount)
          Pst.Str(string(": The value ", QUOTE))
          Pst.Dec(result)
          Pst.Str(string(QUOTE, " is "))
          result := CheckType(result)
          waitcnt(clkfreq / 4 + cnt) 
         
        Pst.Str(string(13, 11, "The value 1 was reached in "))
        Pst.Dec(stepCount)
        Pst.Str(string(" steps."))
     
    PUB CheckType(value)
    
      
      if value & 1 ' odd
        result := value * 3 + 1
        Pst.Str(string("odd. Mulitplying by 3 and adding 1"))     
      else
        result := value / 2
        Pst.Str(string("even. Dividing by 2"))
    
      Pst.Str(string(" which equals "))
      Pst.Dec(result)
      Pst.Char(".")
    
    PUB CheckForInputError(value)
    
      if value < 1 or value > UPPERLIMIT
        Pst.Str(string(11, 13, "The value ", QUOTE))
        Pst.Dec(value)
        Pst.Str(string(QUOTE, " is not a valid entry."))
        result := true     ' "true" is also -1 or $FFFFFFFF or %11111111111111111111111111111111 (32 bits of all ones)
    

    Here's the output when 7 is used as the input.
    Press any key to begin.
    
    Collatz Conjecture
    
    Enter a positive interger between "1" and "715827881".
    You entered "7".
    Step # 1: The value "7" is odd. Mulitplying by 3 and adding 1 which equals 22.
    Step # 2: The value "22" is even. Dividing by 2 which equals 11.
    Step # 3: The value "11" is odd. Mulitplying by 3 and adding 1 which equals 34.
    Step # 4: The value "34" is even. Dividing by 2 which equals 17.
    Step # 5: The value "17" is odd. Mulitplying by 3 and adding 1 which equals 52.
    Step # 6: The value "52" is even. Dividing by 2 which equals 26.
    Step # 7: The value "26" is even. Dividing by 2 which equals 13.
    Step # 8: The value "13" is odd. Mulitplying by 3 and adding 1 which equals 40.
    Step # 9: The value "40" is even. Dividing by 2 which equals 20.
    Step # 10: The value "20" is even. Dividing by 2 which equals 10.
    Step # 11: The value "10" is even. Dividing by 2 which equals 5.
    Step # 12: The value "5" is odd. Mulitplying by 3 and adding 1 which equals 16.
    Step # 13: The value "16" is even. Dividing by 2 which equals 8.
    Step # 14: The value "8" is even. Dividing by 2 which equals 4.
    Step # 15: The value "4" is even. Dividing by 2 which equals 2.
    Step # 16: The value "2" is even. Dividing by 2 which equals 1.
    The value 1 was reached in 16 steps.
    
    Collatz Conjecture
    
    Enter a positive interger between "1" and "715827881".
    

    Edit: I didn't set "stepCount" to zero with each new number being tested. The embedded code is fixed but I need to attach the fixed code to a different post. I can't change which files are attached in an edit. See post below for fixed program.
  • Heater.Heater. Posts: 21,230
    Clearly we need to do this in binary. Just hold your number as an array of longs.

    Divide by 2 is only a shift down by one of all the bits in the array.

    Mull by 3 and is only a shift up by one of all the bits then add the original number. Long by long, rippling the carry up.

    Add 1 is easy.

    Testing that you have hit 1 is easy.

    This enables us to work with numbers of hundreds or thousands of bits. Depends how long you want to wait for it to run.

    The intermediate steps could be easily displayed in hexadecimal.

    Converting those big numbers to decimal display is a bit more tricky.
  • Ron CzapalaRon Czapala Posts: 2,418
    edited 2016-08-12 23:03
    Duane Degn wrote: »
    MikeDYur wrote: »
    It would be fun to watch the Propeller do the math on the PST, slowed down to see what's happening.

    Try this program.

    UPPERLIMIT = posx / 3 - 1 ' I'm not sure if this is a low enough value.

    I doubt that this limit will work because the intermediate calculation results can go way higher (and lower) than the starting value before coming back down to 1.

    e.g.


    63728124
    31864062
    15932031
    47796094
    23898047
    71694142
    35847071
    107541214
    53770607
    161311822
    80655911
    241967734
    120983867
    362951602
    181475801
    544427404
    272213702
    136106851
    408320554
    204160277
    612480832
    306240416
    153120208
    76560104
    38280052
    19140026
    9570013
    28710040
    14355020
    7177510
    3588755
    10766266
    5383133
    16149400
    8074700
    4037350
    2018675
    6056026
    3028013
    9084040
    4542020
    2271010
    1135505
    3406516
    1703258
    851629
    2554888
    1277444
    638722
    319361
    958084
    479042
    239521
    718564
    359282
    179641
    538924
    269462
    134731
    404194
    202097
    606292
    303146
    151573
    454720
    227360
    113680
    56840
    28420
    14210
    7105
    21316
    10658
    5329
    15988
    7994
    3997
    11992
    5996
    2998
    1499
    4498
    2249
    6748
    3374
    1687
    5062
    2531
    7594
    3797
    11392
    5696
    2848
    1424
    712
    356
    178
    89
    268
    134
    67
    202
    101
    304
    152
    76
    38
    19
    58
    29
    88
    44
    22
    11
    34
    17
    52
    26
    13
    40
    20
    10
    5
    16
    8
    4
    2
    1
  • Duane DegnDuane Degn Posts: 10,588
    edited 2016-08-11 18:20
    I agree, the "UPPERLIMIT" value I used isn't a safe number. I'm writing a program to find the lowest start value which will cause a 32-bit overflow (a signed 32-bit number like those used in the Propeller).

    I've attached the fixed code which is embedded in an earlier post. I forgot to clear the "stepCount" variable.
  • Heater.Heater. Posts: 21,230
    Perhaps if you could determine the UPPERLIMT prior to running the code you might get a Fields medal.

    (Only for those under 40 years of age)

  • Ron,
    I haven't got past checking for odd or even numbers, It will take a bit to catch up with the thread. I take it back, I can read VB and understand what's going on in your script. Don't think I could write one though.


    EDIT: That cool Duane, I wish I could put something together that quick. I need examples most of the time.

    We are going to be watching four of the grandkids for four days in a couple of weeks, three of them can do this math, should keep them busy for an hour or two.

    Thanks guys.
  • Heater.Heater. Posts: 21,230
    Duane Degn,

    I think one should check for overflow on every "n*3" and then "+ 1" operation.

    After all the user could input $10000000 which would go to 1 very fast but is much bigger than your UPPERLIMIT.

  • Heater.Heater. Posts: 21,230
    MikeDYur ,
    We are going to be watching four of the grandkids for four days in a couple of weeks, three of them can do this math, should keep them busy for an hour or two.
    That is cruel and unusual punishment :)
  • You only need to compare odd results to the upper limit. Even results always diminish.

    -Phil
  • Heater. wrote: »
    That is cruel and unusual punishment :)

    You are so right, they might put the prop though it for an hour maybe.
  • Ron CzapalaRon Czapala Posts: 2,418
    edited 2016-08-11 19:42
    Here is an updated vbscript - it creates a text file called Collatz.txt on your desktop with the results...
    Option Explicit
    
    Dim fso, WshShell, ts, outFile, desktop
    Dim val1, cntr, result, idata
    
      Set fso = CreateObject("Scripting.FileSystemObject")
      set WshShell = WScript.CreateObject("Wscript.Shell")
    
      desktop = WshShell.SpecialFolders("Desktop")
      outFile = fso.BuildPath(desktop, "Collatz.txt")
    
      val1 = inputbox("Enter starting number",  "The Collatz Conjecture")
      if val1 = "" then
        wscript.quit  
      end if
    
      set ts = fso.CreateTextFile(outFile, True, False)
      val1 = CLng(val1)
      if val1 < 1 then
        msgbox "Value must be > 0", vbcritical, "Error" 
        wscript.quit
      end if
      ts.writeline "-> " & val1
      cntr = 0
      idata = val1	
      do
        if val1 = 1 then
          exit do
        end if
        result = clng(val1 Mod 2)
        if result = 0 then
          val1 = val1 / 2
        else
          val1 = (val1 * 3) + 1   
        end if
        cntr = cntr + 1
        ts.writeline cntr & ". " & val1
        idata = idata & "," & val1    
      loop
      msgbox idata,vbinformation,cntr & " Loops " & outFile
      ts.close  
      set fso = nothing 
    
  • This could definitely use some paper, eight digits got me 183 steps. A lot more fun to watch the Propeller do the math.
  • Heater.Heater. Posts: 21,230
    Phil,
    You only need to compare odd results to the upper limit. Even results always diminish.
    That's what I said Phil.

    Except one can optimize that by checking that by checking "n*3" by UPPERLIMIT - 1 such that the "+1" will never fail.

  • MikeDYurMikeDYur Posts: 2,176
    edited 2016-08-11 19:57
    Sombody have an explanation why 600 produces 17 steps, and 601 is done in 56. Just one number different, It is amazing.


    EDIT: 603 took 69 steps, 604 took 17 again, it seems odd numbers take three times the steps.
  • Heater.Heater. Posts: 21,230
    MikeDYur,

    Like I said, if one could answer such questions one would probably have proved or disproved the conjecture. Thus becoming a Fields medal winner.
  • Ron CzapalaRon Czapala Posts: 2,418
    edited 2016-08-11 20:11
    deleted
  • Heater.Heater. Posts: 21,230
    I don't see anything related to the current topic on that link Ron.
  • Heater.Heater. Posts: 21,230
    Ah, sorry Ron, that link is in your sig. The thin dotted line separating post content from the sig did not show up on this crappy laptop screen.
  • MikeDYur wrote: »

    We are going to be watching four of the grandkids for four days in a couple of weeks, three of them can do this math, should keep them busy for an hour or two.

    Thanks guys.

    Count your blessings, MikeDYur. Sixteen GrandKidDays is a gift from the Gods.

  • MikeDYurMikeDYur Posts: 2,176
    edited 2016-08-11 20:24
    24684 came up with 263,


    Heater, I missed your comment above,

    Mr. Lothar Collatz is a brain to come up with it, and it's just simple math.
Sign In or Register to comment.