Updates from June, 2026 Toggle Comment Threads | Keyboard Shortcuts

  • Unknown's avatar

    Jim Randell 8:20 am on 3 June 2026 Permalink | Reply
    Tags:   

    Teaser 2431: [Football league] 

    From The Sunday Times, 26th April 2009 [link]

    Teams A, B, C… (up to some letter) play in a football league. Last season, they finished in alphabetical order. This season, the league consisted of the same teams, but last season’s losers were the champions, and B finished below C, but not in the bottom three. One third of the teams finished higher this season than last; no two moved the same number of places. So, if one team went up four places, no other team will have gone up or down by four places. In particular, just one team had the same position in both seasons.

    Starting with the champions, list the order of the teams this season.

    This puzzle was originally published with no title.

    [teaser2431]

     
    • Jim Randell's avatar

      Jim Randell 8:21 am on 3 June 2026 Permalink | Reply

      If exactly one third of the teams moved up, then the number of teams must be divisible by 3, and if they are each allocated a letter of the alphabet there can be no more than 26.

      B finished below C, but not in the bottom three teams, so there must be at least 5 teams.

      So the possible number of teams is: 6, 9, 12, 15, 18, 21, 24.

      This Python program looks for the smallest possible number of teams.

      It labels the teams from 1 .. n, so last year’s position is the same as the label. It then tries all possible allocations for this year’s positions, and checks to see if all the conditions are met.

      It runs in 126ms. (Internal runtime is 55ms).

      from enigma import (
        Enumerator, irange, subsets, multiset, compare, update,
        diff, str_upper, item, join, map2str, printf
      )
      
      # check that pos: map team -> position satisfies the requirements
      def check(n, pos):
        # for each team collect the up/down and the gap
        (updown, gap) = (multiset(), multiset())
        for (k, v) in pos.items():
          d = k - v
          updown.add(compare(d, 0))
          gap.add(abs(d))
        # ups = n/3, non-movers = 1, no duplicate gaps
        return (updown.count(1) * 3 == n and updown.count(0) == 1 and not gap.is_duplicate())
      
      # solve for teams 1 .. n (= A .. ?)
      def solve(n):
        # choose values for teams B and C (1 < C < B < n - 2)
        for (C, B) in subsets(irange(2, n - 3), size=2):
          pos1 = { n: 1, 2: B, 3: C }
          # allocate the remaining positions
          ks = diff(irange(1, n), pos1.keys())
          for vs in subsets(diff(irange(1, n), pos1.values()), size=len, select='P'):
            pos = update(pos1, ks, vs)
            if check(n, pos):
              yield pos
      
      # consider multiples of 3
      for n in irange(6, 24, step=3):
        sols = Enumerator(solve(n))
        for pos in sols:
          # construct the order of the teams
          order = tuple(str_upper[k - 1] for (k, v) in sorted(pos.items(), key=item(1)))
          # output solution
          printf("{n} teams -> {order} [pos = {pos}]", order=join(order), pos=map2str(pos))
        # stop at first n with solutions
        if sols.count > 0: break
      

      Solution: The order of the teams (first to last) is: I, H, G, C, B, F, E, A, D.

      Fortunately the solution occurs with 9 teams, because this approach gets much slower as the number of teams increases. (I used a more sophisticated version of this approach to check teams up to 18, but it would take too long to check 21 and 24).

      Like

    • John Crabtree's avatar

      John Crabtree 1:49 pm on 4 June 2026 Permalink | Reply

      Let there be 3n teams.
      The maximum number of places gained is (3n – 1) + (3n – 3) + …(n + 1) with n terms.
      The minimum number of places lost is (3n – 2) + (3n – 4) + …(n) + sum[0 to (n -1)].
      And so n >= n(n-1)/2, ie n<=3
      If n = 1 or 2, the sum of the changes in places is odd. And so n = 3, with places gained = places lost = 18.
      The changes in the places are then forced: champions I up 8 to 1; H up 6 to 2; G up 4 to 3; A (not B) down 7 to 8; D (not B or C) down 5 to 9; B down 3 (max) to 5 and C down 1 to 4; E down 2 to 7; and F stays at 6.

      Like

  • Unknown's avatar

    Jim Randell 7:51 am on 31 May 2026 Permalink | Reply
    Tags:   

    Teaser 3323: Planet 9 from outer space 

    From The Sunday Times, 31st May 2026 [link] [link]

    T’zer was the ninth planet of the Sol-Vit system. Over time, astronomers on T’zer observed each of the eight innermost planets when the lines from the sun to the planet and the planet to T’zer were perpendicular. Radar showed the eight T’zer-to-planet distances at these positions were different whole numbers of light-chrons.

    Simultaneously, measuring the angle between the sun’s centre and the planet, they then calculated each planet’s distance from the sun (they were all whole numbers of light-chrons).

    T’zer’s calculated distance from the sun was always the same whole number of light-chrons due to its circular orbit.

    Astrologers grumbled about this latter distance being under the 80 light-chrons used, historically, when producing horoscopes.

    What is T’zer’s distance from its sun (in light-chrons)?

    [teaser3323]

     
    • Jim Randell's avatar

      Jim Randell 7:52 am on 31 May 2026 Permalink | Reply

      When the distances are measured, T’zer (T), the planet (P), and the sun (S) form a right-angled triangle, where all distances are whole numbers of light-chrons, and the distance ST forms the hypotenuse of the triangle.

      Using the [[ pythagorean_triples() ]] function from the enigma.py library allows us to get a very fast solution.

      The following Python program runs in 70ms. (Internal runtime is 95µs).

      from enigma import (defaultdict, pythagorean_triples, printf)
      
      # collect possible SP distances by ST distance
      d = defaultdict(set)
      for (x, y, z) in pythagorean_triples(79):
        d[z].update({x, y})
      
      # look for ST distances with (at least) 8 values
      for (k, vs) in d.items():
        if len(vs) >= 8:
          printf("{k} -> {vs}", vs=sorted(vs))
      

      Solution: [To Be Revealed]

      Like

    • Ruud's avatar

      Ruud 3:23 pm on 31 May 2026 Permalink | Reply

      import math
      import itertools
      
      for c in range(10, 80):
          if len(sol := [b for b in range(1, c) if (a := math.sqrt(c * c - b * b)) == int(a)]) >= 8:
              print("T-zer:", c, "; others:", *sol)
      

      Like

  • Unknown's avatar

    Jim Randell 8:15 am on 29 May 2026 Permalink | Reply
    Tags:   

    Teaser 2435: [Swimming across the river] 

    From The Sunday Times, 24th May 2009 [link]

    Philip, the fastest swimmer, was on one bank of the River Times; Beth and Sam were directly opposite him on the other bank. Simultaneously, each of them started to swim at their own steady speed to the opposite bank, then, without pause, back to where they had started. On each stretch, Philip crossed each of the others once, and he noted the distances to the nearer bank on each occasion. These four distances were all different whole numbers of metres between 10 and 15, inclusive.

    How wide was the river?

    This puzzle was originally published with no title.

    [teaser2435]

     
    • Jim Randell's avatar

      Jim Randell 8:16 am on 29 May 2026 Permalink | Reply

      Consider Philip (starting from the far bank) and one of the other swimmers (starting from the near bank). They set off at the same time, and as Philip is the stronger swimmer, he encounters the other swimmer a distance a from the near bank. They have been swimming for the same amount of time, so the ratio of the swimming speeds is:

      P/Q = (x − a) / a

      As the stronger swimmer, Philip makes the turn first, and then passes the other swimmer for a second time at a distance b from the far bank (b < a). This time we get:

      P/Q = (2x − b) / (x + b)

      Equating these ratios gives:

      (x − a) / a = (2x − b) / (x + b)
      (x − a)(x + b) = a(2x − b)
      (x² + bx − ax − ab = 2ax − ab
      x² = (3a − b)x
      x = 3a − b

      So, we can choose a pair of values (a, b) (given 10 ≤ a < b ≤ 15) and calculate the corresponding value x (which is the width of the river).

      And we then need to find two disjoint pairs that give the same value for the width of the river.

      This Python program runs in 68ms. (Internal runtime is 81µs).

      from enigma import (defaultdict, irange, subsets, union, printf)
      
      # record pairs by calculated width
      d = defaultdict(list)
      
      # consider possible pairs of distances
      for (b, a) in subsets(irange(10, 15), size=2):
      
        # calculate the width of the river = x
        # (x - a)(x + b) = (2x - b)a
        # => x = 3.a - b
        x = 3*a - b
        # check b < x / 2
        if x < 2 * b: continue
        d[x].append((a, b))
        printf("[({b}, {a}) -> x={x}]")
      
      # look for 2-different, disjoint pairs
      for (x, ps) in d.items():
        for (p1, p2) in subsets(ps, size=2):
          if len(set(p1 + p2)) != 4: continue
          # output solution
          printf("x={x} -> p1={p1} p2={p2}")
      

      Solution: The river is 32 m wide.

      The pairs of distances are (14, 10) and (15, 13).

      And the ratios of the speeds are 15/17 (≈ 0.882) and 7/9 (≈ 0.778).

      Other widths that give multiple pairs are:

      x=29 → (13, 10) and (14, 13)
      x=31 → (14, 11) and (15, 14)

      But these cannot give two disjoint pairs.

      Like

    • Frits's avatar

      Frits 12:11 pm on 2 June 2026 Permalink | Reply

      # we are looking for 2 pairs (a, b) and (c, d) with a > b and c > d
      # and we assume a < c.
      
      # we have river width = 3a - b = 3c - d or 3.(c - a) = d - b
      # as d - b <= 5 (c - a) must be 1 and (d - b) must be 3 which leads to
      # pairs (a, b) and (a + 1, b + 3)
      # also a + 1 > b + 3 or a > b + 2, as a may not be equal to b + 3 
      # we have a >= b + 4, as a <= 14 and b >= 10 we see that a = 14, b = 10
      # leading to pairs (14, 10) and (15, 13) with river width = 3.14 - 10 = 32
      

      Like

  • Unknown's avatar

    Jim Randell 10:35 am on 27 May 2026 Permalink | Reply
    Tags: ,   

    Teaser 2434: [Metal box] 

    From The Sunday Times, 17th May 2009 [link]

    In metalwork class, Pat was given a thin tin sheet, the shape of a regular polygon, each of whose sides was a two-digit number of centimetres. By sawing and discarding an identical piece from each corner, and doing some folding, he made an open-topped box. Its base was a similar-shaped polygon to the original piece, with vertical sides.

    Pat had chosen the size of the discarded pieces to maximise the volume of the box: a six-figure number of cubic centimetres. That number, written as dd/mm/yy, was his grandfather’s date of birth.

    What was the volume of the box?

    This puzzle was originally published with no title.

    [teaser2434]

     
    • Jim Randell's avatar

      Jim Randell 10:35 am on 27 May 2026 Permalink | Reply

      This Python program considers possible n-gons for n = 3 .. 10, and possible 2-digit side lengths. And looks for viable combinations where the maximum volume is a recognisable 6-digit date.

      It runs in 94ms. (Internal runtime is 25ms).

      from math import (tan, pi)
      from enigma import (irange, fdiv, polygon_area_regular, cproduct, find_max, snapf, nsplit, printf)
      
      # calculate volume of a box, starting with a regular n-gon
      # with side <x>, and then make a box by measuring <y> from
      # each corner and cutting off a quadrilateral from each corner
      def vol(n, x, y):
        # the area of the base = A
        A = polygon_area_regular(n, x - 2*y)
        # height of the sides = z
        t = pi * fdiv(n - 2, 2 * n)
        z = y * tan(t)
        # volume
        return A * z
      
      # consider possible n-gons and possible 2-digit sides (= x)
      for (n, x) in cproduct([irange(3, 10), irange(10, 99)]):
      
        # maximise the volume (= V)
        r = find_max((lambda y: vol(n, x, y)), 0, 0.5 * x)
        # volume is close to a 6-digit integer
        V = snapf(r.fv, t=1e-4, default=None)
        # V should be recognisable as a 6-digit date (dd/mm/yy)
        if V is None or V < 100000: continue
        (dd, mm, yy) = nsplit(V, k=3, base=100)
        if not (0 < dd < 32 and 0 < mm < 13): continue
        y = r.v
      
        # output solution
        printf("n={n} x={x} -> y={y:.4f} V={V} ({dd:02d}/{mm:02d}/{yy:02d})")
      

      Solution: The volume of the box is 140625 cm³ .

      So the grandfather’s DOB is 14/06/25 (14th June 1925).

      The original sheet of metal was a hexagon (6-gon) with a side length of 75 cm.

      Cuts perpendicular to the sides at a distance 12.5 cm from each corner are made to produce a hexagonal base with side length 50 cm (area = 3750 √3 cm²) and six rectangular sides with height 12.5 √3 cm.

      The volume of the box is then: (3750 √3) × (12.5 √3) = 3750 × 12.5 × 3 = 140625.

      Like

      • Jim Randell's avatar

        Jim Randell 3:30 pm on 28 May 2026 Permalink | Reply

        Using some analysis:

        The equation for the volume is:

        V(n, x, y) = (1/4) n y (x − 2y)² cot²(𝝅/n)

        And, for a fixed n and x this is at a maximum when:

        y = x/6

        hence:

        V = (1/54) n x³ cot²(𝝅 / n)

        And for an integer volume the cot²(𝝅/n) part (= c) must be rational, restricting the value of n to one of the following:

        n = 3 → c = 1/3
        n = 4 → c = 1
        n = 6 → c = 3

        And so we can simplify the program:

        from enigma import (irange, cproduct, div, cb, nsplit, printf)
        
        # rational (n, cot^2(pi/n)) pairs as (n, p/q)
        ncs = {
          3: (1, 3),
          4: (1, 1),
          6: (3, 1),
        }
        
        for ((n, (p, q)), x) in cproduct([ncs.items(), irange(10, 99)]):
        
          # calculate volume V
          V = div(n * cb(x) * p, 54 * q)
          # must be recognisable as a 6-digit date (dd/mm/yy)
          if V is None or V < 100000: continue
          (dd, mm, yy) = nsplit(V, k=3, base=100)
          if not (0 < dd < 32 and 0 < mm < 13): continue
        
          # output solution
          printf("n={n} x={x} -> V={V} ({dd:02d}/{mm:02d}/{yy:02d})")
        

        Like

  • Unknown's avatar

    Jim Randell 6:53 am on 24 May 2026 Permalink | Reply
    Tags:   

    Teaser 3322: Three houses in long lane 

    From The Sunday Times, 24th May 2026 [link] [link]

    My two best friends and I all live along Long Lane, a row of 60 houses numbered consecutively from 1 to 60. I live at number 10, while my two friends have higher house numbers. One day, I was walking from one friend’s house to the other, and I kept my mind busy by adding up all the house numbers that I passed (not including their two house numbers). Thinking about this sum for a while, I realised that it is precisely the product of their two house numbers.

    In which two houses do my friends live?

    [teaser3322]

     
    • Jim Randell's avatar

      Jim Randell 7:02 am on 24 May 2026 Permalink | Reply

      (See also: Enigma 914, Teaser 2994, Enigma 1).

      If the friends houses are numbers A and B:

      10 < A < B ≤ 60

      Then the product A × B is the same as the sum of the house numbers in the interval [A + 1, B − 1].

      A simple search suffices to find the answer.

      This Python program runs in 70ms. (Internal runtime is 871µs).

      from enigma import (irange, subsets, printf)
      
      for (A, B) in subsets(irange(11, 60), size=2):
        if A * B == sum(irange(A + 1, B - 1)):
          printf("A={A} B={B}")
      

      But we can convert the O(n²) search into an O(n) one with a bit of analysis:

      We have:

      AB = (B − 1)B / 2 − A(A + 1) / 2

      2AB = (B − 1)B − A(A + 1)

      So we can choose a value for A, and then look for roots of the following quadratic equation in B:

      B² − B(2A + 1) − A(A + 1) = 0

      The following Python program runs in 72ms. (Internal runtime is 2.7ms).

      from enigma import (irange, quadratic, printf)
      
      # consider the lower friends number
      for A in irange(11, 59):
        # look for possible B values
        for B in quadratic(1, -(2*A + 1), -A*(A + 1), domain='Z'):
          if A < B <= 60:
            printf("A={A} B={B}")
      

      For the small search space of this puzzle, it is no faster.

      Solution: The house numbers of the friends are 14 and 35.

      The product is: 14 × 35 = 490.

      And the sum of the house numbers in the interval [15, 34] is also 490.

      Like

      • Jim Randell's avatar

        Jim Randell 9:56 am on 24 May 2026 Permalink | Reply

        With a bit more analysis we can find a family of solutions where the numbers are not limited:

        If we suppose there are k houses between A and B (so B = A + k + 1).

        Then the sum of the numbers of these k houses is:

        S = kA + tri(k) = kA + k(k + 1)/2

        And the product of A and B is:

        P = A(A + k + 1)

        Equating these gives:

        k(k + 1)/2 = A(A + 1)

        tri(k) = 2 tri(A)

        And if we make the following substitution: X = 2k + 1, Y = 2A + 1 we get:

        X² − 2Y² = −1

        Which is a form of Pell’s equation [@wikipedia], and there is already code to solve such Diophantine equations in the pells.py library.

        The following program finds the first 10 non-negative candidate (k, A, B) values.

        from enigma import (div, arg, printf)
        import pells
        
        (n, i) = (arg(10, 0, int), 0)
        for (X, Y) in pells.diop_quad(1, -2, -1):
          k = div(X - 1, 2)
          A = div(Y - 1, 2)
          if k is None or A is None: continue
          B = A + k + 1
          # assert A * B == irange.sum(A + 1, B - 1)
          sol = (10 < A < B <= 60)
          printf("[X={X} Y={Y}] -> k={k} A={A} B={B}{s}", s=(" [*SOLUTION*]" if sol else ""))
          i += 1
          if i == n: break
        
        % python3 teaser3322pells.py 10
        [X=1 Y=1] -> k=0 A=0 B=1
        [X=7 Y=5] -> k=3 A=2 B=6
        [X=41 Y=29] -> k=20 A=14 B=35 [*SOLUTION*]
        [X=239 Y=169] -> k=119 A=84 B=204
        [X=1393 Y=985] -> k=696 A=492 B=1189
        [X=8119 Y=5741] -> k=4059 A=2870 B=6930
        [X=47321 Y=33461] -> k=23660 A=16730 B=40391
        [X=275807 Y=195025] -> k=137903 A=97512 B=235416
        [X=1607521 Y=1136689] -> k=803760 A=568344 B=1372105
        [X=9369319 Y=6625109] -> k=4684659 A=3312554 B=7997214
        

        Only one candidate is in the required range for the puzzle (and is found in 133µs).

        The (k, A) pairs form an increasing sequence such that tri(k) = 2 tri(A).

        Like

    • Ruud's avatar

      Ruud 7:22 am on 24 May 2026 Permalink | Reply

      for a in range(10, 60):
          for b in range(a + 1, 61):
              if (b - a - 1) * (a + b) / 2 == a * b:
                  print(a, b)
      

      Like

    • Frits's avatar

      Frits 9:12 am on 24 May 2026 Permalink | Reply

      from math import ceil
      
      # (f2 - f1)^2 = 2.f1^2 + f1 + f2
      for f1 in range(11, 60):
        # determine the minimum and maximum of (f2 - f1)
        mn = ceil((f1 * (2 * f1 + 1) + f1 + 1)**.5)
        mx = int((f1 * (2 * f1 + 1) + 60)**.5)
        for f2 in range(mn + f1, min(mx + f1 + 1, 61)):
          if (f2 - f1)**2 == f1 * (2 * f1 + 1) + f2:
            print(f"answer: {f1} and {f2}")
      

      Like

    • Ruud's avatar

      Ruud 10:43 am on 24 May 2026 Permalink | Reply

      As a one liner:

      import itertools
      
      print(*[f"{a} {b}" for a, b in itertools.combinations(range(10, 61), 2) if (b - a - 1) * (a + b) / 2 == a * b])
      

      Like

  • Unknown's avatar

    Jim Randell 8:21 am on 22 May 2026 Permalink | Reply
    Tags:   

    Teaser 2437: [Spending money] 

    From The Sunday Times, 7th June 2009 [link]

    My granddaughter Madeline has been shopping. When I asked her the prices of the three items she had bought, she said that each cost less than £10, and that the three prices between them used all the digits 1 to 9. Furthermore, the middle-priced item cost one penny less than three times the cheapest, and the total amount spent was a whole number of pounds.

    What were the three prices?

    This puzzle was incorrectly published as Teaser 2537.

    This puzzle was originally published with no title.

    [teaser2437]

     
    • Jim Randell's avatar

      Jim Randell 8:23 am on 22 May 2026 Permalink | Reply

      Here is a solution using the [[ SubstitutedExpression ]] solver from the enigma.py library.

      The following run-file executes in 78ms. (Internal runtime of the generated code is 211µs).

      #! python3 -m enigma -rr
      
      SubstitutedExpression
      
      # suppose the prices (in pence) are:
      #
      #  ABC = cheapest
      #  DEF
      #  GHI = most expensive
      #
      --digits="1-9"
      
      # the prices are in order
      "A < D < G"
      
      # the middle price is 1p less than 3 times the cheapest
      "3 * ABC - 1 = DEF"
      
      # the total came to a whole number of pounds (= multiple of 100p)
      "(BC + EF + HI) % 100 = 0"
      
      # answer is the 3 prices
      --answer="(ABC, DEF, GHI)"
      

      Solution: The prices were: 239p, 716p, 845p.

      The total amount spent being 1800p = £18.

      Like

    • ruudvanderham's avatar

      ruudvanderham 9:52 am on 22 May 2026 Permalink | Reply

      import peek
      import istr
      
      set123456789=set(istr.range(1, 10))
      for price2 in istr.range(1, 1000):
          price1 = 3 * price2 - 1
          if price2 < 1000 and (price1 | price2).all_distinct():
              for price0 in istr.range(price1 + 1, 1000):
                  if set(price0 | price1 | price2) == set123456789 and (price0 + price1 + price2)[-2:]=="00":
                      peek(price0, price1, price2, price0 + price1 + price2)
      

      Like

    • ruudvanderham's avatar

      ruudvanderham 1:37 pm on 22 May 2026 Permalink | Reply

      Here’s a different (significantly faster) version:

      import peek
      import istr
      
      set123456789 = set(istr.range(1, 10))
      for price2 in istr.range(1, 1000):
          price1 = 3 * price2 - 1
          if price2 < 1000:
              for total in istr.range((price2 + price1 * 2 + 1).ceil(100), 3000, 100):
                  price0 = total - price1 - price2
                  if price0 >= 1000:
                      break
                  if set(price0 | price1 | price2) == set123456789:
                      peek(price0, price1, price2, price0 + price1 + price2)
      

      Like

    • Frits's avatar

      Frits 5:22 pm on 2 June 2026 Permalink | Reply

      # prices p1 < p2 < p3
      for p2 in range(123 * 3 - 1, 876 + 1, 3):
        if len(s3 := set(str(p2))) != 3: continue
        if '0' in s3: continue
        # calculate and check p1
        if len(s6 := set(str(p1 := (p2 + 1) // 3)) | s3) != 6: continue
        if '0' in s6: continue
        # last 2 digits of p3 in order to get a whole number of pounds
        if not (11 < (l2_p3 := 100 - ((p1 + p2) % 100)) < 99): continue
        # different digits
        if not s6.isdisjoint(s2 := set(str(l2_p3))): continue
        # calculate first digit of p3
        if len(f_p3 := set("123456789") - s6 - s2) != 1: continue
        if (p3 := 100 * int(f_p3.pop()) + l2_p3) <= p2: continue
        print(f"answer: {p1}, {p2} and {p3}")
      

      Like

  • Unknown's avatar

    Jim Randell 9:01 am on 20 May 2026 Permalink | Reply
    Tags: ,   

    Teaser 2418: [Square park] 

    From The Sunday Times, 25th January 2009 [link]

    Our park is square, with sides of length 104 metres. In the centre is a circular garden with a diameter of 38 metres.

    When walking around the edge of the garden, I can reach a point that is a whole number of metres from the nearest corner of the park, and is also a whole number of metres from the furthest corner of the park.

    What are those two distances?

    This puzzle was originally published with no title.

    [teaser2418]

     
    • Jim Randell's avatar

      Jim Randell 9:02 am on 20 May 2026 Permalink | Reply

      We place the corners of the park at (±52, ±52), and the circle is centred on (0, 0) with a radius of 19.

      If the point on the circumference of the circle is (x, y), and the distance to the nearest corner (+52, +52) is a, and the distance to the furthest corner (−52, −52) is b, then we have the following equations:

      x² + y² = 19
      (52 − x)² + (52 − y)² = a²
      (52 + x)² + (52 + y)² = b²

      And we can eliminate x and y to get:

      a² + b² = 4 × 52² + 2 × 19² = 11538

      So we can look for two squares that sum to give the required value.

      The following Python program runs in 70ms. (Internal runtime is 51µs).

      from enigma import (
        sq, sum_of_squares, circle_intersect_circle, peek, point_dist, printf
      )
      
      S = 4 * sq(52) + 2 * sq(19)
      for (a, b) in sum_of_squares(S, 2):
        printf("a={a} b={b} [S={S}]")
      
        # find one of the intersection points
        p = peek(circle_intersect_circle(((52, 52), a), ((-52, -52), b)), default=None)
        if p is None: continue
        printf("-> p={p}")
      
        # determine distance to the corners
        for q in [(52, 52), (-52, -52), (-52, 52), (52, -52)]:
          d = point_dist(p, q)
          printf("-> dist to {q} = {d:.6f}")
        printf()
      

      Solution: The distances are 63 m and 87 m.

      However there is a problem with this answer.

      If we look at the distances to each of the corners, we find that the integer distances are not the distances to the nearest and furthest corners (at least not the nearest and furthest from (x, y)).

      (The integer distances are shown in red).

      The distances are:

      AX = 63 m
      BX = 88.92 m
      CX = 87 m
      DX = 60.26 m

      The shortest distance is DX and the longest distance is BX.

      But the puzzle can be saved if the radius of the circle is changed from 19 m to 18 m, then the integer distances are 58 m and 90 m, and these are the nearest and furthest distances:

      In this case the distances are:

      AX = 58 m
      BX = 83.16 m
      CX = 90 m
      DX = 67.44 m

      In fact for N = 52 there are solutions when R = 18, 27, 48.

      And there are solutions with integer distances that are not the nearest/furthest distances when R = 19, 26, 33, 39, 51.

      Like

  • Unknown's avatar

    Jim Randell 6:24 am on 17 May 2026 Permalink | Reply
    Tags:   

    Teaser 3321: The more you buy, the more you save 

    From The Sunday Times, 17th May 2026 [link] [link]

    … or so we are often encouraged to believe.

    I needed to buy exactly 42 coloured pens and I could buy them in packs of 1, 4, 7, 11 or 17. A pack of 1 cost £2, a pack of 17 cost £23, and the prices of the other packs were whole numbers of pounds. The average price per pen always decreased going to larger packs. Given the prices I was faced with, I found a suitable and unique combination of the packs that gave me the lowest possible total cost, and this involved buying at least one pack of 11.

    In ascending order of pack size, how many of each pack did I buy, and what were the pack prices (e.g., 5 at £2 each, 3 at £8 each, etc.)?

    [teaser3321]

     
    • Jim Randell's avatar

      Jim Randell 6:39 am on 17 May 2026 Permalink | Reply

      We can use the more sophisticated [[ Denominations() ]] class from the enigma.py library to solve this puzzle.

      The following Python program runs in 70ms. (Internal runtime is 1.6ms).

      from enigma import (
        Denominations, Accumulator,
        irange, choose, fdiv, dot, singleton, seq2str, fmts, printf
      )
      
      # denominations
      ds = [1, 4, 7, 11, 17]
      # find ways to express 42 using the given denominations
      qss = list(Denominations(ds).express(42))
      
      # find prices for 4, 7, 11 packs (cost per item decreases)
      (f1, f17) = (fdiv(2, 1), fdiv(23, 17))
      fns = [
        (lambda p4: f1 > fdiv(p4, 4) > f17),
        (lambda p4, p7: fdiv(p4, 4) > fdiv(p7, 7) > f17),
        (lambda p4, p7, p11: fdiv(p7, 7) > fdiv(p11, 11) > f17),
      ]
      for (p4, p7, p11) in choose(irange(3, 22), fns):
        ps = (2, p4, p7, p11, 23)
      
        # find ways to buy exactly 42 for the lowest total cost
        r = Accumulator(fn=min, collect=1)
        for qs in qss:
          t = dot(qs, ps)
          r.accumulate_data(t, qs)
        # there is a single way to achieve the lowest total cost
        # which involves buying at least 1 pack of 11 (index 3)
        (t, qs) = (r.value, singleton(r.data))
        if qs is None or qs[3] < 1: continue
      
        # output solution
        fmt = lambda vs: seq2str(vs, fn=fmts("2d"), enc="[]")
        printf("total cost = {t}")
        printf("-> packs      = {ds}", ds=fmt(ds))
        printf("-> prices     = {ps}", ps=fmt(ps))
        printf("-> quantities = {qs}", qs=fmt(qs))
        printf()
      

      Solution: 3 at £2 each; 2 at £15 each; 1 at £23.

      i.e. the pack prices were:

      1-pack = £2 (= £2.00/item)
      4-pack = £7 (= £1.75/item)
      7-pack = £12 (= £1.71/item)
      11-pack = £15 (= £1.36/item)
      17-pack = £23 (= £1.33/item)

      And the 42 pens were bought as:

      3× 1-pack (= 3 items, £6)
      2× 11-pack (= 22 items, £30)
      1× 17-pack (= 17 items, £23)
      total = 42 items, £59 (avg = £1.40/item)

      Like

    • Ruud's avatar

      Ruud 5:07 pm on 17 May 2026 Permalink | Reply

      import collections
      import itertools
      
      p17 = 23
      p1 = 2
      for p4 in range(23):
          if not (p17 / 17) < p4 / 4 < (p1 / 1):
              continue
          for p7 in range(23):
              if not (p17 / 17) < p7 / 7 < (p4 / 4):
                  continue
              for p11 in range(23):
                  if not (p17 / 17) < p11 / 11 < (p7 / 7):
                      continue
      
                  n = 42
                  c = collections.defaultdict(list)
                  for n17 in itertools.count():
                      a17 = n - n17 * 17
                      if a17 < 0:
                          break
                      for n11 in itertools.count():
                          a11 = a17 - n11 * 11
                          if a11 < 0:
                              break
                          for n7 in itertools.count():
                              a7 = a11 - n7 * 7
                              if a7 < 0:
                                  break
      
                              for n4 in itertools.count():
                                  a4 = a7 - n4 * 4
                                  if a4 < 0:
                                      break
                                  n1 = a4
      
                                  sum = n17 * p17 + n11 * p11 + n7 * p7 + n4 * p4 + n1 * p1
                                  c[sum].append((n1, n4, n7, n11, n17))
                  if len(lowest := c[min(c)]) == 1 and lowest[0][3]:
                      print(f"total={min(c)} quantities={lowest[0]} prices={p1, p4, p7, p11, p17}")
      

      Like

    • Hugo's avatar

      Hugo 8:02 am on 24 May 2026 Permalink | Reply

      I reckoned two 7-packs at £10 each, one 11-pack at £15, and one 17-pack at £23,
      making £58 in all. Where did I go wrong?

      Like

      • Jim Randell's avatar

        Jim Randell 9:04 am on 24 May 2026 Permalink | Reply

        For example, with prices of:

        1-pack = £2
        4-pack = £7
        7-pack = £10
        11-pack = £15
        17-pack = £23

        The lowest possible cost for 42 pens is £58 (as you say).

        But there are two ways this minimum price can be achieved:

        2× 7-packs (@ £10) = £20
        1× 11-pack (@ £15) = £15
        1× 17-pack (@ £23) = £23
        total (42 items) = £58

        1× 1-pack (@ £2) = £2
        1× 7-pack (@ £10) = £10
        2× 17-pack (@ £23) = £46
        total (42 items) = £58

        So this cannot be scenario, as there must be only one way to achieve the minimum price.

        In fact there are 9 possible sets of prices that allow the condition that the prices per item decreases as the pack price increases. Of these only 3 have a unique way to achieve the minimum price (which, in each case, is £59). And only one of these 3 candidates uses 11-packs in the unique arrangement that achieves the minimum price, and so this provides the answer to the puzzle.

        Like

  • Unknown's avatar

    Jim Randell 8:21 am on 15 May 2026 Permalink | Reply
    Tags: by: Rupert Segar   

    Brain teaser 975: Ambiguous birthdays 

    From The Sunday Times, 29th March 1981 [link]

    Recently I attended the Numerical Astrologers Conference, held annually in Brighton. At this meeting, members wore badges on which were written not only their names, but also their dates of birth. These dates were written numerically in the usual English manner of first putting the day, then the month, then the year (e.g. 7th March 1946 would be represented by 7/3/46, and the 31st December 1954 by 31/12/54). The year always uses two digits.

    I had just entered the meeting when I was accosted by an internationally acclaimed cabalist, who, I noticed, shared with me the month of his birthday, though he was born in a year eight years prior to the year of my birth. He had been looking at my badge, and now told me that the digits of my date of birth were, in sequence, in exactly the reverse order to his own.

    We were eagerly discussing this coincidence when an old acquaintance of mine introduced himself to the cabalist, who then made the same claim to my friend about their dates of birth. At first this puzzled me, as I knew my friend to be younger than myself, but, upon inspection, both the claims proved to be correct.

    Strangely, both of us had dates of birth whose digits, when taken in completely the reverse order, were exactly the same as those of the cabalist.

    Numerically, what was the cabalist’s date of birth?

    The puzzle is included in the book The Sunday Times Book of Brainteasers (1994).

    [teaser975]

     
    • Jim Randell's avatar

      Jim Randell 8:22 am on 15 May 2026 Permalink | Reply

      The cabalist’s DOB cannot be 6 digits or 4 digits, as there is only one way to reverse these:

      UV/WX/YZ → ZY/XW/VU
      W/X/YZ → Z/Y/XW

      But if it had 5 digits, there are 2 ways it can be reversed:

      V/WX/YZ or VW/X/YZ → Z/YX/VW or ZY/X/VW

      As the friend is younger than the setter they must be born later in the year, so:

      setter = ZY/X/VW
      friend = Z/YX/VW

      And the cabalist and the setter share their birth month, so:

      cabalist = VW/X/YZ
      setter = ZY/X/VW
      friend = Z/YX/VW

      Here is a solution using the [[ SubstitutedExpression ]] solver from the enigma.py library.

      It runs in 86ms. (Internal runtime of the generated code is 2.2ms).

      #! python3 -m enigma -rr
      
      SubstitutedExpression
      
      # convert a 2-digit year
      --code="year = lambda y: (1900 + y if y < 70 else 1800 + y)"
      # check for a valid date
      --code="from datetime import date"
      --code="valid = lambda d, m, y: catch(date, year(y), m, d) is not None"
      
      # the cabalist's DOB is VW/X/YZ
      # and can be reversed as setter = ZY/X/WV, friend = Z/YX/WV
      --distinct=""
      --invalid="0,VYZ" # dates do not have components with leading zeros
      
      # check the three dates are valid
      "valid(VW, X, YZ)"
      "valid(ZY, X, WV)"
      "valid(Z, YX, WV)"
      
      # the [2-digit] years are 8 apart
      "year(YZ) + 8 == year(WV)"
      
      --template="VW/X/YZ -> ZY/X/WV and Z/YX/WV"
      --solution=""
      

      Solution: The cabalist’s date of birth was 12/1/13 (12th January 1913).

      And so the setter’s date of birth is 31/1/21 (31st January 1921) and the friend’s date of birth is 3/11/21 (3rd November 1921).

      Like

    • ruudvanderham's avatar

      ruudvanderham 2:01 pm on 15 May 2026 Permalink | Reply

      import datetime
      
      date0 = datetime.datetime(1880, 1, 1)
      date1 = datetime.datetime(1980, 1, 1)
      
      day = {}
      month = {}
      date = date0
      while date < date1:
          date = date + datetime.timedelta(1)
          as_str = f"{date.day}{date.month}{date.year % 100:02d}"[::-1]
          if len(as_str) == 5 and "0" not in as_str[:3]:
              year = int(as_str[3:])
              year += 1800 if year > 80 else 1900
              if year - date.year == 8:
                  for i in range(2):
                      month[i] = int(as_str[i + 1 : 3])
                      day[i] = int(as_str[: i + 1])
                      if month == date.month:
                          i_equal = i
                      try:
                          datetime.datetime(year, month[i], day[i])
                      except ValueError:
                          break
                  else:
                      for i in range(2):
                          if month[i] == date.month:
                              if datetime.datetime(year, month[i], day[i]) < datetime.datetime(year, month[1 - i], day[1 - i]):
                                  print(
                                      f"Cabalist: {date:%Y-%m-%d} / Me: {datetime.datetime(year, month[i], day[i]):%Y-%m-%d} / Friend:{datetime.datetime(year, month[1 - i], day[1 - i]):%Y-%m-%d}"
                                  )
      

      Like

  • Unknown's avatar

    Jim Randell 8:05 am on 13 May 2026 Permalink | Reply
    Tags:   

    Teaser 2417: [Flower bed] 

    From The Sunday Times, 18th January 2009 [link]

    Old Mellors, the estate gardener, has designed a large flower bed. He has marked out two overlapping circles whose radii differ by one metre: the bed consists of the area within one or both of the circles. He wants to plant various straight lines of seeds in the flower bed. The longest such possible is 25 metres long; if he wants the line to touch the perimeter of the bed in three places, then the longest possible is 22 metres.

    What are the radii of the circles?

    This puzzle was originally published with no title.

    [teaser2417]

     
    • Jim Randell's avatar

      Jim Randell 8:08 am on 13 May 2026 Permalink | Reply

      If we suppose the circle have radii R and r, where Rr, then:

      For the circles to overlap the distance d between their centres must be:

      R − r < d < R + r

      In this case we have R = r + 1, so:

      1 < d < 2r + 1

      The longest possible line is 25, so:

      R + d + r = 25

      d = 24 − 2r

      Hence:

      d > 1

      r < 11.5

      and:

      d < 2r + 1

      r > 5.75

      So we have bounds on possible values for r, we can look for situations where the maximum line through an intersection is 22 m.


      First I defined some additional 2D geometry functions to help with circles:

      Circle = namedtuple('Circle', 'centre radius')
      
      # find intersection points of circle <c>
      # with the line defined by points <p1>, <p2>
      # may return 0, 1, 2 points
      def circle_intersect_line(c, p1, p2):
        (((x0, y0), r), (x1, y1), (x2, y2)) = (c, p1, p2)
        (xd, yd) = (x2 - x1, y2 - y1)
        # construct a polynomial for the intersection
        f = sq(Polynomial([x1 - x0, xd])) + sq(Polynomial([y1 - y0, yd]))
        ts = f.roots(sq(r), domain='F')
        return list(P2(x1 + t * xd, y1 + t * yd) for t in ts)
      
      # parameterised circle t = -1 to +1 for a full circle
      # 0 = 3 o'clock; 0 -> 1 = anticlockwise to 9 o'clock; 0 -> -1 = clockwise to 9 o'clock
      def circle_param(c, t=None):
        ((x, y), r) = c
        f = lambda t, x=x, y=y, r=r: P2(x + r * math.cos(t * pi), y + r * math.sin(t * pi))
        return (f if t is None else f(t))
      
      # find <t> parameter for point <p> on circle centre <c> radius <r>
      def circle_param_t(c, p, t=1e-9):
        (o, r) = c
        if not (abs(point_dist(o, p) - r) < t): return None
        ((x0, y0), (x, y)) = (o, p)
        t = math.acos(fdiv(x - x0, r)) / pi
        return min(t, -t, key=(lambda t: abs(point_dist(circle_param(c, t), p))))
      

      (These functions are in the latest version of enigma.py).

      The following Python program solves the problem constructively. Given a value for r (the radius of the smaller circle) it first derives the values of R (the radius of the larger circle) and d (the separation between the centres), and then finds P (one of the intersection points of the circle). It then extends the tangent to the smaller circle at P to intersect with the larger circle at Q.

      We then look for a point A on the circumference of the larger circle between P and Q, and extend AP to a point B on the smaller circle. And we choose A so to maximise the length of AB (using the [[ find_max() ]] function from the enigma.py library.

      We can then use the [[ find_values() ]] function from the enigma.py library to determine what value of r gives a maximum AB of 22.

      It runs in 220ms. (Internal runtime is 151ms).

      from enigma import (
        Circle, circle_intersect_line, circle_param, circle_param_t,
        triangle_point, point_dist, line_param, line_bisect,
        find_max, find_values, call, item, printf
      )
      
      # extract leftmost and rightmost points from a sequence
      leftmost = lambda pts: (min(pts, key=item(0)) if pts else None)
      rightmost = lambda pts: (max(pts, key=item(0)) if pts else None)
      
      # find maximum length line through intersection point P, given <r>
      def solve(r):
        # larger circle radius is 1 m larger
        R = r + 1
        # distance between circle centres
        d = 24 - 2*r
      
        # specify the two circles (centre, radius)
        (CR, Cr) = (Circle((0, 0), R), Circle((d, 0), r))
      
        # find the upper intersection point = P
        P = triangle_point(d, R, r)
        # and the corresponding start <t> parameter on the larger circle
        tP = circle_param_t(CR, P)
      
        # calculate the tangent to the smaller circle at P
        f = line_param(Cr.centre, P)
        (p1, p2) = line_bisect(f(0), f(2))
        # and where it intersects the larger circle at Q
        Q = leftmost(circle_intersect_line(CR, p1, p2))
        # and the finish <t> parameter on the larger circle
        tQ = circle_param_t(CR, Q)
        if tQ < tP: tQ += 2
      
        # points on the larger circle parameterised by <t>
        fR = circle_param(CR)
        # find the A, B intersection points for A at parameter <t>
        def fAB(t):
          # A is on the larger circle
          A = fR(t)
          # the line from A extends through P and intersects the smaller circle at B
          B = rightmost(circle_intersect_line(Cr, A, P))
          return (A, B)
        # find the maximum length AB line for <t> parameters between P and Q
        m = find_max((lambda t: call(point_dist, fAB(t))), tP, tQ)
        # return the maximal value
        return m.fv
      
      # find <r> when the max line through an intersection point is 22
      for s in find_values(solve, 22.0, 5.75, 11.5):
        # determine parameters
        r = s.v
        R = r + 1
        d = 24 - 2*r
        printf("r = {r:.2f}; R = {R:.2f}; d = {d:.2f}")
      

      Solution: The radii of the circles are 6.5 m and 7.5 m.

      And the distance between the centres of the circles is 11 m.


      Manually:

      Suppose the centres of the circles are separated by a distance d.

      Then we can plot the circles as follows:

      centre of the larger circle (radius R) = (0, 0)
      centre of the smaller circle (radius r) = (d, 0)
      upper intersection point of the circles, P = (x, y)

      We can draw a straight line L through P at an angle of 𝛉.

      At a distance t (negative to the left and positive to the right) it has the following parametric equation:

      L(t) = (x + t cos(𝛉), y + t sin(𝛉))

      The intersection with the larger circle (A) is given when t ≠ 0 and:

      (x + t cos(𝛉))² + (y + t sin(𝛉))² = R²

      t = −2(x cos(𝛉) + y sin(𝛉))

      So the distance AP is given by:

      AP = 2(x cos(𝛉) + y sin(𝛉))

      Similarly the intersection with the small circle (B) is given by:

      (x + t cos(t) − d)² + (y + t sin(t))² = r²

      t = 2(d cos(𝛉) − x cos(𝛉) − y sin(𝛉))

      So the distance BP is given by:

      BP = 2(d cos(𝛉) − x cos(𝛉) − y sin(𝛉))

      And so the total distance AB is:

      AB = 2(x cos(𝛉) + y sin(𝛉)) + 2(d cos(𝛉) − x cos(𝛉) − y sin(𝛉))
      AB = 2d cos(𝛉)

      From which we see the distance AB achieves a maximum when 𝛉 = 0 (i.e. when the line is horizontal [*]), and the maximum value is twice the distance between the centres of the circles.

      We can now apply this finding to the puzzle.

      The maximum length line through the intersection is 22 m, hence the distance between the circles is 11 m.

      And the maximum length line between two points on the perimeter of the bed is 25 m.

      Hence:

      R + d + r = 25
      (r + 1) + 11 + r = 25

      2r = 13
      r = 6.5

      And the solution follows.


      [*] However, not all configurations permit a horizontal line to be drawn, in which case the maximum length line is given by the tangent to the smaller circle (PQ), with the angle increased infinitesimally to permit the line to intercept the perimeter of the combined shape at 3 distinct points.

      For example:

      Like

  • Unknown's avatar

    Jim Randell 12:07 am on 10 May 2026 Permalink | Reply
    Tags:   

    Teaser 3320: Penblwydd Hapus 

    From The Sunday Times, 10th May 2026 [link] [link]

    I’m now in my thirties and for many years I’ve treated myself to some bottles of beer on my birthday. To compensate for my increasing age, I buy the same number of bottles each year as my age. On my recent birthday, after storing my bottles, I noticed that, on the beer receipt from the supermarket, all of the nine digits 1 to 9 appeared once, and only once, in the number of bottles, the price per bottle in pence and the total cost in pence.

    What was the total cost of the beer in pence?

    [teaser3320]

     
    • Jim Randell's avatar

      Jim Randell 12:20 am on 10 May 2026 Permalink | Reply

      I think the wording on this puzzle could be tightened up a little. (Specifically to clear up: “what constitutes a recent birthday?”, and “how many times can the digit 0 appear?”).

      I assumed we are looking for a sum consisting of 9 digits, using each of the digits 1-9 exactly once, but I find multiple solutions. (And if we allow the digit 0 to appear then there are even more solutions).

      Using birthdays in the last 5 years, but disallowing 0 digits, I get 3 candidate solutions. If we take “recent birthday” to mean “most recent birthday”, then we can eliminate two of these, so maybe that is what the setter intended.

      This Python program runs in 88ms. (Internal runtime is 19ms).

      from enigma import (irange, inf, nsplit, flatten, printf)
      
      # consider possible ages for a "recent" birthday
      for n in irange(26, 39):
        # consider price per bottle
        for p in irange(100, inf):
          # calculate the total cost
          t = n * p
          # find the digits used in the sum
          ds = flatten(nsplit(x) for x in (n, p, t))
          # look for 9 different non-zero digits
          if len(ds) > 9: break
          if 0 not in ds and len(set(ds)) == 9:
            # output solution
            printf("{n} * {p} = {t}")
      

      Solution: The total cost of the beer was 7254p.

      If we suppose the “recent birthday” is the setter’s most recent birthday (that perhaps happened within the last few weeks), then the birthday is 30 .. 39, and the only solution using just the 9 digits 1-9 exactly once is:

      39 × 186 = 7254

      i.e. the beer cost 186p per bottle.

      However, if the recent birthday is one of the last (say) 5 (the puzzle text does say the practice has been going on “for many years”), then it could be between 26 .. 39, and this permits the following additional solutions:

      27 × 198 = 5346
      28 × 157 = 4396

      Also, the puzzle text does not place restrictions on how many occurrences of the digit 0 there may be, so we can get the following reasonable solutions:

      27 × 594 = 16038
      36 × 495 = 17820
      39 × 402 = 15678

      And if very expensive bottles of beer were purchased there are further solutions with more than one 0 digit, e.g.:

      39 × 1284 = 50076

      In fact any solution can have the price per bottle and the total price multiplied by 10 to give a further solution if no restrictions are placed on the digit 0.


      So, I think the intended puzzle is:

      I am now in my thirties. On my most recent birthday I bought a number of identical bottles of beer. That number was the same as my age in years.

      The number of bottles, the price per bottle in pence, and the total cost in pence, when taken together comprised 9 different non-zero digits, with each of the digits 1–9 occurring exactly once.

      What (in pence) was the total cost of the beer?

      This revised puzzle is easily solved directly from the command line using the [[ SubstitutedExpression ]] solver from the enigma.py library:

      % python3 enigma.py SubstitutedExpression "AB * CDE = FGHI" --digits="1-9" --assign="A,3"
      (AB * CDE = FGHI)
      (39 * 186 = 7254) / A=3 B=9 C=1 D=8 E=6 F=7 G=2 H=5 I=4
      [1 solution]
      

      (The internal runtime of the generated code is 550µs).

      Like

    • Ruud's avatar

      Ruud 6:48 am on 10 May 2026 Permalink | Reply

      It is easy to see that the price should have 3 digits and the total cost 4 digits.

      import peek
      import istr
      
      for age in istr.range(31, 40):
          for rest in istr.permutations({*istr.range(1, 10)} - {*age}):
              if (price := istr.join(rest[:3])) * age == (total := istr.join(rest[3:])):
                  peek(age, price, total)
      

      Like

    • Frits's avatar

      Frits 11:04 am on 10 May 2026 Permalink | Reply

      # forbidden unit digits (3 is reserved for the tens digit of the age) 
      fb = {0, 3}
      
      # dictionary of price unit digits per n_bottles unit digit
      d = {b: {p for p in range(10) if p not in fb and p != b and 
                                      (u := (p * b) % 10) not in fb | {b, p}} 
              for b in range(10) if b not in fb}
      
      # the number of bottles (equal to age)
      for b in [n for n in range(30, 40) if n % 10 in d]:
        # unit digit
        u = b % 10
        # minimum/maximum price, price/cost should have resp. 3/4 digits
        mn, mx = 124, 9876 // b
        #pr(b, (mn, mx), (b * mn, b * mx), d[u])
        # the price per bottle in pence
        for p in [n for n in range(mn, mx + 1) if n % 10 in d[u]]:
          if '0' in (dgts := set(str(b) + str(p) + str((c := b * p)))): continue
          if len(dgts) == 9:
            print(f"answer: {c}p")
      

      Like

    • Hugo's avatar

      Hugo 12:29 pm on 12 May 2026 Permalink | Reply

      Keeping the pattern AB × CDE = FGHI but without the condition that A = 3, I found
      18 × 297 = 5346
      27 × 198 = 5346
      28 × 157 = 4396
      42 × 138 = 5796
      48 × 159 = 7632

      I didn’t try any other patterns, with or without the inclusion of 0.
      No doubt Jim can fill that gap for us.

      Like

  • Unknown's avatar

    Jim Randell 8:35 am on 8 May 2026 Permalink | Reply
    Tags:   

    Teaser 2441: [Letter grid] 

    From The Sunday Times, 5th July 2009 [link]

    Your task today is to place nine different letters of the alphabet in a 3-by-3 grid.

    The letters must include S, A, M, E. You must do this in such a way that, when the letters are given their numerical value (A = 1, B = 2, C = 3, etc.) then each row, column and diagonal of the grid has the same total.

    You will find that the other five letters used can swiftly be rearranged into a common word.

    What is that word?

    This puzzle was originally published with no title.

    [teaser2441]

     
    • Jim Randell's avatar

      Jim Randell 8:36 am on 8 May 2026 Permalink | Reply

      Here is a solution using the [[ SubstitutedExpression ]] solver from the enigma.py library.

      The following run file executes in 92ms. (Internal runtime of the generated code is 10.2ms).

      #! python3 -m enigma -rr
      
      SubstitutedExpression
      
      # assign values from 1 - 26 to the grid:
      #
      #   A B C
      #   D E F
      #   G H I
      #
      # to make a magic square with sum 3E
      --base="27"
      --digits="1-26"
      
      # rows sum to 3E
      "A + B + C == 3 * E"
      "D + F == 2 * E"
      "G + H + I == 3 * E"
      
      # cols sum to 3E
      "A + D + G == 3 * E"
      "B + H == 2 * E"
      "C + F + I == 3 * E"
      
      # diags sum to 3E
      "A + I == 2 * E"
      "C + G == 2 * E"
      
      # required values (S, A, M, E)
      "{1, 5, 13, 19}.issubset({A, B, C, D, E, F, G, H, I})"
      
      # remove symmetric arrangements
      "A < C" "C < G" "A < I"
      
      --template=""
      --answer="ordered(A, B, C, D, E, F, G, H, I)"
      

      The completed grid is:

      [ C  (3) | U (21) | I  (9) ]      [ S (19) | A  (1) | M (13) ]
      [ Q (17) | K (11) | E  (5) ]  ->  [ E  (5) | K (11) | Q (17) ]
      [ M (13) | A  (1) | S (19) ]      [ I  (9) | U (21) | C  (3) ]

      Each row, column and diagonal has a sum of 33.

      And the corresponding letters are: SAME (= 19, 1, 13, 5) + KQIUC (= 11, 17, 9, 21, 3).

      Solution: The word is: QUICK.

      Like

  • Unknown's avatar

    Jim Randell 8:44 am on 5 May 2026 Permalink | Reply
    Tags:   

    Teaser 2433: [Date of birth] 

    From The Sunday Times, 10th May 2009 [link]

    I was looking at the dates of birth of three of my friends. When written in the format DDMMYY, each date of birth uses six consecutive digits in some order.

    Today (just taking each of their ages on their last birthday) the difference between Alex’s age and Bernard’s age is 40 or more, and the difference between Alex’s age and Charlie’s age is 14.

    What, in the format DDMMYY, is Charlie’s date of birth?

    This puzzle was originally published with no title.

    [teaser2433]

     
    • Jim Randell's avatar

      Jim Randell 8:45 am on 5 May 2026 Permalink | Reply

      Note that the puzzle was originally set on 10th May 2009.

      The first digit of the month can only be 0 (for Jan – Sep) or 1 (for Oct – Dec).

      So the consecutive digits can only be 0-5 or 1-6. And if they are 1-6 then the month can only be 12 and then it is not possible for there to be a date in the range 01-31 using digits 3-6, so the digits used must be 0-5.

      This Python program runs in 64ms. (Internal runtime is 1.5ms).

      from datetime import date
      from enigma import (irange, subsets, catch, printf)
      
      # date the puzzle was set
      dS = date(2009, 5, 10)
      
      # calculate age
      def age(dX, dY):
        return (dY.year - dX.year) - ((dY.month, dY.day) < (dX.month, dX.day))
      
      # collect possible dates and ages
      ds = set()
      for (u, v, w, x, y, z) in subsets(irange(0, 5), size=6, select='P'):
        (d, m, y) = (10*u + v, 10*w + x, 1900 + 10*y + z)
        t = catch(date, y, m, d)
        if t is not None:
          ds.add((t, age(t, dS)))
      
      ss = set()
      # consider DOB for A
      for (dA, aA) in ds:
        # choose DOB for C
        for (dC, aC) in ds:
          if not (abs(aA - aC) == 14): continue
          # chose DOB for B
          for (dB, aB) in ds:
            if not (abs(aA - aB) >= 40): continue
            printf("[A={dA} ({aA}); B={dB} ({aB}); C={dC} ({aC})]")
            ss.add(dC)
      
      # output solutions
      for dC in ss:
        printf("C = {dC:%d%m%y} [{dC}]")
      

      Solution: Charlie’s date of birth is 250341 (25th March 1941).

      So Charlie was 68 when the puzzle was set.

      Alex was born in 1954, and his date of birth (and age) is one of the following:

      23/10/54 (54)
      03/12/54 (54)
      30/12/54 (54)

      Whichever it is his age is 54 and he is 14 years younger than Charlie.

      Bernard’s date of birth (and age) is one of the following:

      25/04/13 (96)
      24/05/13 (95)
      25/03/14 (95)
      23/05/14 (94)
      24/03/15 (94)
      23/04/15 (94)

      And he is 94, 95, or 96.

      But whichever it is he is at least 40 years older than Alex.


      Note: The puzzle only works when set between 25th March 2009 and 22nd May 2009.

      Like

  • Unknown's avatar

    Jim Randell 7:57 am on 3 May 2026 Permalink | Reply
    Tags:   

    Teaser 3319: Watch out! 

    From The Sunday Times, 3rd May 2026 [link] [link]

    One January a few years ago I was given a new watch that also displayed the date. It did this by turning two wheels, one with 0-3 on it and the other with 0-9. So, if left running, it displayed in turn 00, 01, 02, … 29, 30, 31, … 39, 00, 01, … . Therefore it needed resetting each month, but I never bothered. The date was correctly displayed when I received the watch but then was incorrect for every other month of that year. However, on my birthday that year the display showed an odd number that was in fact the correct date but with the digits reversed! I had to console myself with the fact that, before the watch was 30 years old, on one occasion my birthday would be correctly displayed!

    When is my birthday (day and month)?

    [teaser3319]

     
    • Jim Randell's avatar

      Jim Randell 8:13 am on 3 May 2026 Permalink | Reply

      We only need to consider 4 candidate start years (a leap year, +1, +2, +3).

      This Python program runs in 63ms. (Internal runtime is 755µs).

      from datetime import (date, timedelta)
      from enigma import (irange, repeat, inc, nrev, printf)
      
      # start in year y
      def solve(y):
        # look for possible birthdays in the first year
        bds = set()
        # consider odd displays
        n = 1
        for t in repeat(inc(timedelta(days=2)), date(y, 1, 1)):
          if t.year > y: break
          # display must be incorrect for months other than January
          if n == t.day:
            if t.month > 1: return
          elif n == nrev(t.day, 2):
            # display is the reverse of the day
            bds.add((t, n))
          # advance the display
          n = (n + 2) % 40
      
        # consider birthdays up to 29 years later
        for (t, n) in bds:
          for i in irange(1, 29):
            t1 = t.replace(year=y + i)
            n1 = (n + (t1 - t).days) % 40
            if n1 == t1.day:
              printf("{t} ({n:02d}) -> {t1} ({n1:02d}) [+{i} years]")
      
      # consider 4 candidate years ("a few years ago")
      for y in [2020, 2021, 2022, 2023]:
        solve(y)
      

      Solution: Your birthday is 30th August.

      The setter was given the watch in a leap year (2020 for example), when it showed the correct date for January (but not in any other months).

      On 2020-08-30 the watch displays “03” (instead of “30”). And 28 years later on 2048-08-30 the watch will display “30”.

      There are other birthdays in the first year when the watch will display a date that is incorrect, but the actual date reversed:

      10th February (displays “01” in any year; displays “10” after 55 years)
      10th June (displays “01” in a non-leap year)
      31st July (displays “13” in a leap year; displays “31” after 34 years)

      Like

    • Ruud's avatar

      Ruud 12:57 pm on 3 May 2026 Permalink | Reply

      import datetime
      import istr
      import calendar
      
      istr.int_format("02")
      
      
      def display_time(start_date, year, month, day):
          diff = (datetime.datetime(year, month, day) - start_date).days
          return istr((diff + 1) % 40)
      
      
      for year in (2021, 2022, 2023, 2024):
          start_date = datetime.datetime(year, 1, 1)
          for month in range(2, 13):
              if display_time(start_date, year, month, 1) == 1:
                  break
          else:
              for month in range(1, 13):
                  for day in range(1, calendar.monthrange(year, month)[1]):
                      displayed = display_time(start_date, year, month, day)
                      if displayed.is_odd() and day == displayed.reversed() and displayed.reversed() != displayed:
                          correct_years = [year for year in range(year, year + 30) if display_time(start_date, year, month, day) == day]
                          if len(correct_years) == 1:
                              print(year, correct_years[0], month, day)
      

      Like

      • Jim Randell's avatar

        Jim Randell 4:38 pm on 3 May 2026 Permalink | Reply

        @Ruud: Python’s [[ range(a, b) ]] builtin excludes the final value (b in this case). So your code at line 20 will miss checking the final day of the month.

        The enigma.py library has the [[ irange(a, b) ]] function, if you want to include both endpoints.

        Like

        • Ruud's avatar

          Ruud 4:47 pm on 3 May 2026 Permalink | Reply

          Well spotted. It should read

          import datetime
          import istr
          import calendar
          
          istr.int_format("02")
          
          
          def display_time(start_date, year, month, day):
              diff = (datetime.datetime(year, month, day) - start_date).days
              return istr((diff + 1) % 40)
          
          
          for year in (2021, 2022, 2023, 2024):
              start_date = datetime.datetime(year, 1, 1)
              for month in range(2, 13):
                  if display_time(start_date, year, month, 1) == 1:
                      break
              else:
                  for month in range(1, 13):
                      for day in range(1, calendar.monthrange(year, month)[1] + 1):
                          displayed = display_time(start_date, year, month, day)
                          if displayed.is_odd() and day == displayed.reversed() and displayed.reversed() != displayed:
                              correct_years = [year for year in range(year, year + 30) if display_time(start_date, year, month, day) == day]
                              if len(correct_years) == 1:
                                  print(year, correct_years[0], month, day)
          

          Like

  • Unknown's avatar

    Jim Randell 9:21 am on 1 May 2026 Permalink | Reply
    Tags:   

    Teaser 2445: [Divisible by 9] 

    From The Sunday Times, 2nd August 2009 [link]

    I started with a list of six numbers (with no leading zeros, of course). Five of the six numbers were divisible by 9.

    Then I coded the numbers by consistently using letters for digits, with different letters representing different digits.

    In this way, the list became:

    SETTER
    SENT
    PRETTY
    EASY
    SUNDAY
    TEASER

    Which of those words represented the number that was not divisible by 9?

    This puzzle was originally published with no title.

    [teaser2445]

     
    • Jim Randell's avatar

      Jim Randell 9:22 am on 1 May 2026 Permalink | Reply

      Here is a straightforward solution using the [[ SubstitutedExpression ]] solver from the enigma.py library.

      It runs in 106ms. (Internal runtime of the generated code is 21ms).

      #! python3 -m enigma -rr
      
      SubstitutedExpression
      
      --distinct="ADENPRSTUY"
      --invalid="0,EPST"
      
      # exactly five of the six numbers are divisible by 9
      # so just 1 of them isn't, identify it as X = 1..6
      --invalid="0|7-9,X"
      
      "(X == 1) == (SETTER % 9 != 0)"
      "(X == 2) == (SENT % 9 != 0)"
      "(X == 3) == (PRETTY % 9 != 0)"
      "(X == 4) == (EASY % 9 != 0)"
      "(X == 5) == (SUNDAY % 9 != 0)"
      "(X == 6) == (TEASER % 9 != 0)"
      
      --answer="X"
      
      --template="(SETTER) (SENT) (PRETTY) (EASY) (SUNDAY) (TEASER) (X)"
      

      Solution: TEASER is not divisible by 9.

      There are 36 ways to assign the digits to the letters, but in all cases TEASER is the odd number out.


      If the puzzle had been set so that just one of the words is not divisibly by 29, then there is a single way to assign digits to the letters.

      Like

      • Ruud's avatar

        Ruud 11:07 am on 1 May 2026 Permalink | Reply

        Brute force reveals that there are 36 different solutions, all resulting in TEASER to be the one that’s not divisible by 9.

        import peek
        import istr
        
        count = 0
        for s, e, t, r, n, p, y, a, u, d in istr.permutations(range(10)):
            if s * p * e * t: # may not start with 0
                not_divisible_by_9 = ""
                for name in "setter sent pretty easy sunday teaser".split():
                    if not istr(f":={name}").is_divisible_by(9):
                        if not_divisible_by_9:
                            break
                        not_divisible_by_9 = name
                else:
                    if not_divisible_by_9:
                        count += 1
                        peek(count, not_divisible_by_9, setter, sent, pretty, easy, sunday, teaser)
        

        Like

  • Unknown's avatar

    Jim Randell 9:04 am on 28 April 2026 Permalink | Reply
    Tags:   

    Brainteaser 1049: A question of colour 

    From The Sunday Times, 5th September 1982 [link]

    A frame at snooker, as every player — and every watcher of “Pot Black” — will know, consists of fifteen red balls, worth one point each, and six “colours”: yellow = 2, green = 3, brown = 4, blue = 5, pink = 6 and black = 7.

    The reds must all be potted first: after each red the player chooses a single colour to pot (immediately replaced on the table) before addressing the next red. Any miss (or potting of the wrong colour) causes the other player to have his turn. After the last red has been potted, with an attempt at a colour, all the colours are on the table, and they must be potted in ascending order of value.

    The game ends when all the colours have been potted, or earlier if the current player has reached a score that cannot be equalled or exceeded by his opponent. For our present purposes, there are no penalty points or snookers.

    Unlike the champions we see on television, Arthur and Bertie recently played a frame which was as  undistinguished in quality (though neither actually incurred any penalty points) as it was remarkable in other ways. Each of them potted the same number of balls; the scores were level immediately before Bertie potted the last of the reds; and the winner — Arthur by a single point —was not decided until the table was finally cleared. They then discovered that Arthur’s score was the lowest possible for such a close win.

    What was Arthur’s score? Which player potted each of the final six “colours” (e.g. B, B, B, B, B, A).

    This puzzle is included in the book The Sunday Times Book of Brainteasers (1994).

    [teaser1049]

     
    • Jim Randell's avatar

      Jim Randell 9:04 am on 28 April 2026 Permalink | Reply

      We are looking for the lowest possible scores.

      The scores are level with 1 red remaining. So 14 of the reds have been potted. For the lowest possible scores the associated colours must have been missed. So each player has potted 7 reds (and no colours) for a total score of 7.

      The remaining balls are: the final red, a possible additional colour, and the 6 colours in order.

      So, if the table is cleared and each player pots the same number of balls: B pots the final red, and a colour, to put him on 8 + x points (and has now potted 9 balls). So of the 6 remaining colours B pots 2 and A pots 4.

      This Python program examines possible ends to the match.

      It runs in 70ms. (Internal runtime is 143µs).

      from enigma import (Enumerator, printf)
      
      colours = [2, 3, 4, 5, 6, 7]
      
      # play the remaining colours
      def play(A, B, cols, As=[], Bs=[]):
        # are we done?
        if not cols:
          yield (A, B, cols, As, Bs)
        else:
          rest = sum(cols)
          # has someone won?
          if A > B + rest or B > A + rest:
            yield (A, B, cols, As, Bs)
          else:
            # someone pots the next colour
            x = cols[0]
            yield from play(A + x, B, cols[1:], As + [x], Bs)
            yield from play(A, B + x, cols[1:], As, Bs + [x])
      
      # solve the puzzle, where B pots the final red with colour <x>
      def solve(x):
        for (A, B, cols, As, Bs) in play(7, 8 + x, colours):
          # A wins by 1 point
          if not (A == B + 1): continue
          # the table is cleared
          if cols: continue
          # A and B potted the same number of balls
          if not (len(As) == 4 and len(Bs) == 2): continue
          # return the final scores and colours potted
          yield (A, B, As, Bs)
      
      # choose a colour for A to pot with the final red
      for x in colours:
        ss = Enumerator(solve(x))
        for (A, B, As, Bs) in ss:
          # output solution
          printf("A pots colours {As}; total = {A} pts")
          printf("B pots final red + {x}; pots colours {Bs}; total = {B} pts")
          printf()
        # we only need the smallest solution
        if ss.count > 0: break
      

      Solution: Arthur’s score was 23. The final six colours were potted as: A, A, A, B, B, A.

      The match started with A and B each potting 7 reds and no colours, and then:

      → A=7, B=7, [1 red remaining]
      B pots the final red and the green (3)
      → A=7, B=11, [27 points remaining]
      A pots the yellow (2)
      → A=9, B=11, [25 points remaining]
      A pots the green (3)
      → A=12, B=11, [22 points remaining]
      A pots the brown (4)
      → A=16, B=11, [18 points remaining]
      B pots the blue (5)
      → A=16, B=16, [13 points remaining]
      B pots the pink (6)
      → A=16, B=22, [7 points remaining]
      A pots the black (7)
      → A=23, B=22 [final scores]

      B has potted 7 + 2 + 2 = 11 balls. And A has potted 7 + 4 = 11 balls.

      Like

  • Unknown's avatar

    Jim Randell 6:24 am on 26 April 2026 Permalink | Reply
    Tags:   

    Teaser 3318: Squaring the circle 

    From The Sunday Times, 26th April 2026 [link] [link]

    Liam is designing a logo as part of a school project. He has drawn a circle that touches all the sides of an octagon. The side lengths of the octagon are all different two-digit numbers of mm; in fact, they are all triangular numbers (i.e. numbers of the type 1+2+3+4+…). Taken together these numbers use all the digits 0 to 9 (at least once each). If I told you the value of one of the lengths that is an odd number, then you would be able to answer the question below.

    What (in mm) is the perimeter of the octagon?

    [teaser3318]

     
    • Jim Randell's avatar

      Jim Randell 7:01 am on 26 April 2026 Permalink | Reply

      I used the characterisation of a tangential polygon given here [@wikipedia].

      Namely:

      If the sides are (a0, …, a7), then a tangential octagon exists if and only if the following equation set has positive real solutions for (x0, …, x7):

      x0 + x1 = a0
      x1 + x2 = a1

      x7 + x0 = a7

      For positive solutions to exist this means:

      x0 > 0
      x0 < a0
      x0 > a0 − a1
      x0 < a0 − a1 + a2

      x0 < a0 − a1 + a2 − a3 + a4 − a5 + a6
      x0 > a0 − a1 + a2 − a3 + a4 − a5 + a6 − a7 = 0

      So we can bracket possible x0 values by constructing the alternating sum of the candidate side lengths, and then look for non-empty intervals at the end of the process. Any x0 value in this interval will give a viable tangential octagon.

      This Python program runs in 68ms. (Internal runtime is 861µs).

      from enigma import (
        defaultdict, irange, inf, tri, first, lt, div,
        diff, subsets, cproduct, union, nsplit, printf
      )
      
      # 2-digit triangular numbers
      ns = first((tri(n) for n in irange(1, inf)), count=lt(100), skip=lt(10))
      
      # is there a tangential octagon with sides <ss>?
      def is_tangential(ss):
        xs = None
        # for a valid arrangement, the sum of the even numbered sides
        # is the same as the sum of the odd numbered sides
        s = div(sum(ss), 2)
        if s is None: return
        # choose the even numbered sides
        a0 = min(ss)
        ssx = diff(ss, {a0})
        for ss0 in subsets(ssx, size=3, fn=list):
          if a0 + sum(ss0) != s: continue
          # arrange the sides
          ss1 = diff(ssx, ss0)
          for ((a2, a4, a6), (a1, a3, a5, a7)) in cproduct(subsets(xs, size=len, select='P') for xs in [ss0, ss1]):
            if not (a1 < a7): continue
            # bracket possible values for x0 by constructing the alternating sum of the sides
            lo = max(0, a0 - a1, a0 - a1 + a2 - a3, a0 - a1 + a2 - a3 + a4 - a5)
            hi = min(a0, a0 - a1 + a2, a0 - a1 + a2 - a3 + a4, a0 - a1 + a2 - a3 + a4 - a5 + a6)
            # is the bracket non-empty?
            if lo < hi:
              return (a0, a1, a2, a3, a4, a5, a6, a7)
      
      # map odd side lengths to perimeter
      d = defaultdict(set)
      
      # choose 8 of the numbers
      for ss in subsets(ns, size=8):
        # all 10 digits should be used
        ds = union(nsplit(n) for n in ss)
        if len(ds) < 10: continue
        # is there a tangential octagon?
        rs = is_tangential(ss)
        if rs is None: continue
        # record perimeter by side length
        p = sum(ss)
        printf("[{rs} -> {p}]")
        for n in ss:
          d[n].add(p)
      
      # look for odd side lengths with only one possible perimeter
      for n in sorted(d.keys()):
        vs = d[n]
        if n % 2 == 1 and len(vs) == 1:
          printf("{n} -> {vs}")
      

      Solution: The perimeter of the octagon is 358 mm.

      There are three candidate sets of sides that permit a tangential octagon to be constructed. Example arrangements for these sets are:

      (10, 15, 36, 28, 55, 91, 78, 45) → perimeter = 358
      (10, 21, 36, 28, 45, 55, 91, 78) → perimeter = 364
      (10, 21, 45, 36, 55, 66, 91, 78) → perimeter = 402

      (Note that these are not the only arrangements of the sides that permit a tangential octagon to be constructed. For the first and second set there are 20 possible arrangements, and for the third set there are 24. So not all candidate arrangements are viable).

      15 only appears in the first of these, and is the only odd valued side that uniquely identifies a candidate (the one with perimeter 358).

      There is also one even valued side which only appears in one of the candidates, namely 66, which identifies the candidate with perimeter 402.

      Like

      • Jim Randell's avatar

        Jim Randell 11:13 pm on 26 April 2026 Permalink | Reply

        Here is my program modified to use the [[ find_max() ]] solver from the enigma.py library to ensure the equations have a positive solution, and to find the corresponding tangent lengths, which allows candidate tangential octagons to be constructed.

        This Python program runs in 69ms. (Internal runtime is 924µs).

        from enigma import (
          defaultdict, irange, inf, tri, first, lt, div, diff,
          subsets, cproduct, find_max, union, nsplit, printf
        )
        
        # 2-digit triangular numbers
        ns = first((tri(n) for n in irange(1, inf)), count=lt(100), skip=lt(10))
        
        # is there a tangential octagon with sides <ss>?
        def is_tangential(ss):
          # for a valid arrangement, the sum of the even numbered sides
          # is the same as the sum of the odd numbered sides
          s = div(sum(ss), 2)
          if s is None: return
          # choose the even numbered sides
          a0 = min(ss)
          ssx = diff(ss, {a0})
          for ss0 in subsets(ssx, size=3, fn=list):
            if a0 + sum(ss0) != s: continue
            # arrange the sides
            ss1 = diff(ssx, ss0)
            for ((a2, a4, a6), (a1, a3, a5, a7)) in cproduct(subsets(xs, size=len, select='P') for xs in [ss0, ss1]):
              if not (a1 < a7): continue
        
              # can we solve the equations for positive <x> values?
              def solve(x0):
                x1 = a0 - x0
                x2 = a1 - x1
                x3 = a2 - x2
                x4 = a3 - x3
                x5 = a4 - x4
                x6 = a5 - x5
                x7 = a6 - x6
                return min(x0, x1, x2, x3, x4, x5, x6, x7)
        
              # look for the maximum smallest <x> value
              r = find_max(solve, 0, a0)
              if r.fv > 0:
                # return viable solution
                return (a0, a1, a2, a3, a4, a5, a6, a7)
        
        # map odd side lengths to perimeter
        d = defaultdict(set)
        
        # choose 8 of the numbers
        for ss in subsets(ns, size=8):
          # all 10 digits should be used
          ds = union(nsplit(n) for n in ss)
          if len(ds) < 10: continue
          # is there a tangential octagon?
          rs = is_tangential(ss)
          if rs is None: continue
          # record perimeter by side length
          p = sum(ss)
          printf("[{rs} -> {p}]")
          for n in ss:
            d[n].add(p)
        
        # look for odd side lengths with only one possible perimeter
        for n in sorted(d.keys()):
          ps = d[n]
          if n % 2 == 1 and len(ps) == 1:
            printf("{n} -> {ps}")
        

        If the equations admit a positive solution, then there are infinitely many, so the program finds the polygon where the closest tangent point to a vertex is as far as possible.

        We can then use the data to plot viable tangential octagons for candidate arrangements.

        Here are the three viable candidate arrangements, each side of the octagon is shown with its tangent point and side length. (Some of the angles at the vertices are very close to 180°).

        Liked by 2 people

    • Frits's avatar

      Frits 10:59 am on 26 April 2026 Permalink | Reply

      from itertools import combinations
      
      # collect possible triangular side lengths
      n, t, sides = 0, 0, []
      while True:
        t = n * (n + 1) // 2
        if t > 99: break
        if t > 9:
          sides.append(t)
        n += 1
      
      # claim from Google AI:
      # a convex octagon with the sum of its odd-numbered sides' lengths equal to  
      # the sum of its even-numbered sides' lengths has an inscribed circle.
      # thus s1 + s3 + s5 + s7 must be equal to s2 + s4 + s6 + s8
      
      sols = []
      # select eight sides of the octagon
      for c8 in combinations(sides, 8):
        h, r = divmod(sum(c8), 2)
        if r: continue
        # all digits should be present
        if len(set("".join(str(n) for n in c8))) != 10: continue
        # pick four sides s(i) from <c8> with sum <h>
        c1 = c8[0]
        for c3 in combinations(c8[1:], 3):
          if sum(c3) + c1 != h: continue
          sols.append(c8)
          
      # collect possible solutions per odd number
      cands = [[s for s in sols if n in s] for n in range(11, 100, 2)]
      # select odd numbers that only occur in one solution
      sols = [str(sum(vs[0])) for vs in cands if len(vs) == 1]
      print("answer:", ' or '.join(sols))
      

      Liked by 1 person

      • Frits's avatar

        Frits 12:02 pm on 26 April 2026 Permalink | Reply

        Wikipedia mentions that in a tangential polygon with an even number of sides, the sum of the odd numbered sides’ lengths is equal to the sum of the even numbered sides’ lengths. Google AI claims that the reverse is also true (at least for a convex octagonal).

        Like

        • Jim Randell's avatar

          Jim Randell 5:10 pm on 26 April 2026 Permalink | Reply

          @Frits: Yes. I can use that condition to reduce the number of side arrangements that are considered by my program. (Although Wikipedia does not claim it is an “if and only if” condition, and I would require a more trustworthy source than Google AI to verify the sufficiency condition).

          Like

          • Jim Randell's avatar

            Jim Randell 5:33 pm on 26 April 2026 Permalink | Reply

            This paper [link] seems to suggest that the condition is not sufficient for 2n-gons larger than quadrilaterals.

            I think a (2, 2, 2, 2, 100, 3, 3, 100) octagon would be a counterexample.

            Like

            • Frits's avatar

              Frits 7:28 pm on 26 April 2026 Permalink | Reply

              @Jim, Thanks for debunking the claim.
              I haven’t looked at your program yet (beside the use of Matrix.linear).

              Probably you can use your method to check if a sequence of 8 octagon side lengths (with property sum(odd) = sum(even)) has an inscribed circle.

              Liked by 1 person

              • Jim Randell's avatar

                Jim Randell 4:01 pm on 4 May 2026 Permalink | Reply

                Instead of attempting to solve the equations (and looking for non-inconsistent sets) I’ve rejigged my program to determine the interval for x0 where the equations give positive x_i values, and if this interval is non-empty then it is possible to construct a tangential octagon using the given side arrangement.

                [I am yet to see another solution that checks if a candidate arrangement of sides can actually form a tangential octagon].

                Like

    • Brian Gladman's avatar

      Brian Gladman 10:28 pm on 27 April 2026 Permalink | Reply

      @Jim Nice work on the three candidates shapes. The radius of the incircle is 2.A / sum(sides) where A is the polygon’s area but is there a direct route to this radius from the a and x values?

      Like

      • Jim Randell's avatar

        Jim Randell 7:58 am on 28 April 2026 Permalink | Reply

        I took the x values determined by my second program, and for a given radius r you can calculate the angles at each vertex of the octagon.

        You can then vary r until the sum of the vertex angles is 1080° (or 6𝝅 radians).

        I already had a program (written for Teaser 2438) to plot a cyclic polygon (given the angles subtended at the centre by the sides), and then construct a tangential polygon from that.

        So I just needed to calculate the angles at the centre of the octagon, these are supplementary to the angles at the vertices.

        This is the program I used to plot the diagrams:

        from math import (atan2, fsum, pi, degrees)
        from enigma import (find_value, run, arg, printf)
        
        # possible tangents
        xss = {
          1: [6.5, 3.5, 11.5, 24.5, 3.5, 51.5, 39.5, 38.5],  # arrangement 1
          2: [5, 5, 16, 20, 8, 37, 18, 73],  # arrangement 2
          3: [5, 5, 16, 29, 7, 48, 18, 73],  # arrangement 3
        }
        xs = xss[arg(1, 0, int)]
        printf("xs = {xs}")
        
        # calculate vertex angles for radius r
        def angles(r): return list(2 * atan2(r, x) for x in xs)
        
        # find radius that gives an angle sum of 6.pi
        r = find_value((lambda r: fsum(angles(r))), 6 * pi, 0, 1000).v
        printf("inradius = {r}")
        
        # angles at the centre are supplementary to the vertex angles
        args = list(degrees(pi - a) for a in angles(r))
        run("cyclic-polygon.py", *args)
        

        Liked by 1 person

        • Brian Gladman's avatar

          Brian Gladman 8:40 am on 28 April 2026 Permalink | Reply

          Thanks Jim, I hoped that you might have found a direct method to avoid the search for the inradius.

          Like

        • Frits's avatar

          Frits 9:44 am on 28 April 2026 Permalink | Reply

          @Jim, where can we find the program “cyclic-polygon.py”?

          Like

    • Ruud's avatar

      Ruud 7:55 am on 28 April 2026 Permalink | Reply

      Based on @Frits’s solution:

      import istr
      
      sides = [i for i in istr.range(10, 100) if i.is_triangular()]
      
      from itertools import combinations
      
      occurs = {istr(i): [] for i in range(10, 100)}
      sides = [i for i in istr.range(10, 100) if i.is_triangular()]
      for sel_sides in istr.combinations(sides, 8):
          if len(set(istr.join(sel_sides))) == 10:
              for sides3 in istr.combinations(sel_sides[1:], 3):
                  if (sel_sides[0] + sum(sides3)) * 2 == sum(sel_sides):
                      for i in sel_sides:
                          occurs[i].append(sum(sel_sides))
      
      for i, sum_sides in occurs.items():
          if i.is_odd() and len(sum_sides) == 1:
              print(*sum_sides)
      

      Like

  • Unknown's avatar

    Jim Randell 7:14 am on 24 April 2026 Permalink | Reply
    Tags: ,   

    Teaser 2444: [Pat’s lawn] 

    From The Sunday Times, 26th July 2009 [link]

    Pat’s lawn is a 600 sq metre rectangle with a tree at each corner. Working around the perimeter, these are apple, beech, cherry and damson. Pat has marked four straight lines on the grass: one from the apple to a point 75% of the way from the beech to the cherry; the second from the beech to 75% of the way from the cherry to the damson; the third from the cherry to 75% of the way from the damson to the apple; the fourth from the damson to 75% of the way from the apple to the beech. He plans to turn the area enclosed by the four lines into a rose garden.

    What will the rose garden’s area be?

    This puzzle was originally published with no title.

    [teaser2444]

     
    • Jim Randell's avatar

      Jim Randell 7:15 am on 24 April 2026 Permalink | Reply

      This is similar to Routh’s Theorem but in a rectangle instead of a triangle. (See: Teaser 3021).

      If the lines are drawn to a fraction f along the appropriate side, then the area of the central quadrilateral is (as a fraction of the overall parallelogram):

      A = (1 − f)² / (1 + f²)

      Specifically if f is represented as a fraction a/b we get:

      A = (b − a)² / (a² + b²)

      And in our case a = 3, b = 4, hence:

      A = (4 − 3)² / (3² + 4²)
      A = 1/25

      from enigma import (Rational, sq, printf)
      
      Q = Rational()
      
      # fraction along the sides
      f = Q(75, 100)
      printf("[f = {f}]")
      
      # calculate the area of the central quadrilateral
      A = Q(sq(1 - f), 1 + sq(f))
      printf("[A = {A}]")
      
      # output solution (area of central region)
      R = A * 600
      printf("area = {R} sq m")
      

      Solution: The area of the rose garden is 24 sq m.

      Like

  • Unknown's avatar

    Jim Randell 8:13 am on 22 April 2026 Permalink | Reply
    Tags: by: Lachlan MacKinnon,   

    Teaser 2447: [Number grid] 

    From The Sunday Times, 16th August 2009 [link]

    In this variation on a sudoku theme, your task is to find a particular 5-by-5 array of digits in which each row and each column use the same five consecutive digits. When completed, you can read five 5-digit numbers across the rows and a further five down the columns. These 10 numbers should all be different, and their product should be divisible by the fourth powers of 11, 10 and 9. The product should also be divisible by the fourth (but not the fifth) power of 8.

    What are the lowest and highest of your 10 numbers?

    This puzzle was originally published with no title.

    [teaser2447]

     
    • Jim Randell's avatar

      Jim Randell 8:14 am on 22 April 2026 Permalink | Reply

      The overall product is divisible by 9^4 (= 3^8), so at least one of the numbers is divisible by 3, and as the numbers are rearrangements of each other, they must all be divisible by 3. (And so the final product will have a divisor of at least 3^10).

      We can fill out a grid using the digits 0-4 (allowing leading zeros), and than add the same value (between 0 and 5) to each digit to form a candidate grid.

      The sum of the digits 0-4 is 10, and if this combines with 5 copies of the additional digit to give a multiple of 3 then the additional digit can only be 1 or 4.

      This is the approach taken with the following run file, using the [[ SubstitutedExpression ]] solver from the enigma.py library.

      It executes in 661ms. (Internal runtime of the generated code is 448ms (using PyPy 7.3.21)).

      #! python3 -m enigma -rr
      
      SubstitutedExpression
      
      #  A B C D E
      #  F G H I J
      #  K L M N O
      #  P Q R S T
      #  U V W X Y
      #
      # and each number is increased by: ZZZZZ
      
      --macro="@rows = ABCDE, FGHIJ, KLMNO, PQRST, UVWXY"
      --macro="@cols = AFKPU, BGLQV, CHMRW, DINSX, EJOTY"
      --macro="@nums = [@rows, @cols]"
      
      # rows and cols are rearrangements of the same 5 digits
      --distinct="ABCDE,FGHIJ,KLMNO,PQRST,UVWXY,AFKPU,BGLQV,CHMRW,DINSX,EJOTY"
      
      # numbers are formed from 5 consecutive digits
      # so the numbers in the grid are formed from 0-4
      # and the base (Z) is 1 or 4
      --digits="0-4"
      --invalid="0|2|3,Z"
      
      # the numbers are all different
      "seq_all_different(@nums)"
      
      # check divisibility
      --code="""
      D = mlcm(8**4, 9**4, 10**4, 11**4)  # product is divisible by D
      X = 8**5  # but not X
      def check(ns, b=0):
        p = multiply(n + b for n in ns)
        return (p % D == 0) and (p % X != 0)
      """
      "check(@nums, ZZZZZ)"
      
      # remove duplication
      "ABCDE < AFKPU"
      
      --template=""
      --denest=-30  # CPython workaround
      

      Solution: The lowest of the numbers is: 45768. And the highest is: 86547.

      The prime factors are:

      74856 = (2^3)(3)(3119)
      68475 = (3)(5^2)(11)(83)
      86547 = (3)(17)(1697)
      57684 = (2^2)(3)(11)(19)(23)
      45768 = (2^3)(3)(1907)

      76854 = (2)(3)(12809)
      48675 = (3)(5^2)(11)(59)
      84567 = (3)(7)(4027)
      57486 = (2)(3)(11)(13)(67)
      65748 = (2^2)(3)(5479)

      These combine to give an overall product with the following repeated prime factors:

      (2^12)(3^10)(5^4)(11^4)


      In the published solution the lowest number was given as 45678, but this is likely to have been a typo.

      Like

  • Unknown's avatar

    Jim Randell 6:32 am on 19 April 2026 Permalink | Reply
    Tags:   

    Teaser 3317: Hitting all the spots 

    From The Sunday Times, 19th April 2026 [link] [link]

    A board game uses three coloured tetrahedral dice (red, green and blue) to determine which numbers are hit when the three dice are rolled. The 12 faces of the three dice each have different numbers, comprising a 1 and the primes from 2 to 31. A hit is made when it is the number rolled on any of the dice, the sum of any two numbers rolled or the sum of all three numbers rolled.

    The arrangement of the numbers on each die made it possible to hit every number in a range and this range was as large as it could be. The green die has numbers 2, 5, 7, 11 and the blue die also has three consecutive primes.

    In ascending order, what are the numbers on the red die?

    [teaser3317]

     
    • Jim Randell's avatar

      Jim Randell 6:32 am on 19 April 2026 Permalink | Reply

      The most interesting part was writing the [[ maxcss() ]] function to find the maximum continuous subsequence of consecutive integers contained in the parent sequence.

      This Python program runs in 70ms. (Internal runtime is 3.3ms).

      from enigma import (
        Accumulator, primes, irange, tuples, trim,
        diff, cproduct, union, subsets, printf
      )
      
      # calculate the hits for a set of dice
      def hits(ds):
        # sum of any (non-empty) subset of a throw
        hs = union(subsets(vs, min_size=1, fn=sum) for vs in cproduct(ds))
        # return as a sorted list
        return sorted(hs)
      
      # generate possible sets of dice
      def generate():
        # numbers used on the dice
        ns = list(primes.between(2, 31))
        ns.insert(0, 1)
      
        # green die
        G = {2, 5, 7, 11}
      
        # choose 3 consecutive primes for the blue die
        for b3 in tuples(trim(ns, head=6), 3, fn=set):
          rs = diff(ns, G, b3)
          # and one of the remaining numbers
          for bx in subsets(rs, size=1, fn=set):
            B = b3.union(bx)
            # red die is what's left
            R = diff(rs, bx, fn=set)
            yield (G, B, R)
      
      # find the largest continuous subsequence (length >= m)
      # in the list of integers <vs>, the list should be strictly increasing
      # return (<max-length>, (<lowest-value>, <highest-value>))
      def maxcss(vs, m=0):
        r = Accumulator(fn=max, value=m)
        # consider possible starts
        n = len(vs)
        for (i, v) in enumerate(vs):
          if i + r.value > n: break
          for j in irange(n - 1, i + r.value, step=-1):
            if vs[j] - v == j - i:
              r.accumulate_data(j - i + 1, i)
              break
        if r.data is None: return
        (k, i) = (r.value, r.data)
        return (k, (vs[i], vs[i + k - 1]))
      
      # find maximal length range of hits
      r = Accumulator(fn=max, value=1, collect=1)
      for ds in generate():
        hs = hits(ds)
        z = maxcss(hs, r.value)
        if z is None: continue
        (n, (a, b)) = z
        r.accumulate_data(n, ds)
      
      # output solution
      printf("max range = {r.value}")
      for (G, B, R) in r.data:
        printf("-> G={G} B={B} R={R}", G=sorted(G), B=sorted(B), R=sorted(R))
      

      Solution: The numbers on the red die are: 1, 13, 17, 31.

      The dice are:

      Green = (2, 5, 7, 11)
      Blue = (3, 19, 23, 29)
      Red = (1, 13, 17, 31)

      All hits from 1 – 57 can be made. (58 is the first number that cannot be hit using these dice).

      In fact, using these dice we can make: [1 – 57; 59 – 62; 65; 67; 71].

      Like

      • Jim Randell's avatar

        Jim Randell 4:51 pm on 19 April 2026 Permalink | Reply

        Here is a simpler implementation of [[ maxcss() ]], that just returns the length of a maximal subsequence:

        def maxcss(vs):
          if not vs: return 0
          return max(map(len, clump(v - i for (i, v) in enumerate(vs))))
        

        We calculate a derived sequence by subtracting the index of each element of the original sequence from the value of that element. We then look for a maximal length run of equal values in the derived sequence, and this corresponds to a maximal length continuous subsequence of consecutive integers in the original sequence.

        [I am posting this code because although I have seen several other solutions posted for this puzzle, none of them have managed to implement this logic correctly].

        Like

      • Ruud's avatar

        Ruud 5:10 am on 20 April 2026 Permalink | Reply

        @Frits
        Now, I see what you mean. i had interpreted the problem wrongly.
        Here’s a revised version:

        import itertools
        
        max_max_length = 0
        faces = [n for n in range(1, 32) if all(n % i for i in range(2, int(n**0.5) + 1))]  # 1, 2, 3, ...,31
        green = {2, 5, 7, 11}
        for primes3 in zip(faces[0:], faces[1:], faces[2:]):
            if not any(x in green for x in primes3):
                for prime4 in set(faces) - green - {primes3}:
                    blue = {*primes3, prime4}
                    red = set(faces) - green - blue
                    hits = sorted({sum(x) for a, b, c in itertools.product(green, blue, red) for x in itertools.combinations((0, 0, a, b, c), 3)})
                    max_length = 0
                    length = 1
                    for d0, d1 in zip(hits, hits[1:] + [0]):
                        if d1 - d0 == 1:
                            length += 1
                        else:
                            max_length = max(max_length, length)
                            length = 1
                    if max_length >= max_max_length:
                        if max_length > max_max_length:
                            solutions = []
                            max_max_length = max_length
                        solutions.append((green, blue, red))
        print("maximum range = ", max_max_length)
        for solution in solutions:
            for i, color in enumerate("green blue red".split()):
                print(color, sorted(solution[i]))
        

        Like

    • Ruud's avatar

      Ruud 10:14 am on 19 April 2026 Permalink | Reply

      import itertools
      
      max_max_hit = 0
      faces = [n for n in range(1, 32) if all(n % i for i in range(2, int(n**0.5) + 1))]  # 1, 2, 3, ...,31
      green = {2, 5, 7, 11}
      for primes3 in zip(faces[0:], faces[1:], faces[2:]):
          if not any(x in green for x in primes3):
              for prime4 in set(faces) - green - {primes3}:
                  blue = {*primes3, prime4}
                  red = set(faces) - green - blue
                  hits = {sum(x) for a, b, c in itertools.product(green, blue, red) for x in itertools.combinations((0, 0, a, b, c), 3)}
                  for max_hit in itertools.count():
                      if max_hit + 1 not in hits:
                          break
      
                  if max_hit >= max_max_hit:
                      if max_hit > max_max_hit:
                          solutions = []
                          max_max_hit = max_hit
                      solutions.append((green, blue, red))
      print("maximum range = ", max_max_hit)
      for solution in solutions:
          for i, color in enumerate("green blue red".split()):
              print(color, sorted(solution[i]))
      

      Like

      • Frits's avatar

        Frits 11:26 am on 19 April 2026 Permalink | Reply

        @Ruud, why do you assume the range has to start with 1?

        Like

        • Ruud's avatar

          Ruud 3:18 pm on 19 April 2026 Permalink | Reply

          Because 1 is also one of the numbers on the dice. So faces is really 1, 2, 3, …, 31 .

          Like

          • Frits's avatar

            Frits 9:32 pm on 19 April 2026 Permalink | Reply

            @Ruud,

            The collection of hits might be (if 1 and 3 belong to the same colour):

              
            [1, 2, 3, 5, 6, 7, 8, 10, 11, 12, 13, 14, 15, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 55, 57, 59, 61]
            

            In this case the biggest range isn’t 3 but 37 (17-53) and number 1 isn’t included in this range.

            Like

    • Ellen Napier's avatar

      Ellen Napier 1:21 pm on 20 April 2026 Permalink | Reply

      I found two different answers to this puzzle.

      Like

      • Jim Randell's avatar

        Jim Randell 6:53 am on 21 April 2026 Permalink | Reply

        @Ellen: I found there were 20 possible sets of dice. And one of them gave the single longest continuous subsequence of consecutive hits.

        However there are two sets which give the second longest continuous subsequence of consecutive hits.

        Like

    • Ellen Napier's avatar

      Ellen Napier 11:13 pm on 21 April 2026 Permalink | Reply

      Got it, thanks. Had a sorting error.

      Like

c
Compose new post
j
Next post/Next comment
k
Previous post/Previous comment
r
Reply
e
Edit
o
Show/Hide comments
t
Go to top
l
Go to login
h
Show/Hide help
shift + esc
Cancel
Design a site like this with WordPress.com
Get started