5D Library › 黄历择吉布尔格、建除十二直同余与卢恩符文组合学 · English

第三章 联合周期:干支日、二十八宿与七曜

并行的几个周期

有三种日计数与太阳无关地贯穿通书:六十天的干支循环,二十八宿的值日循环,以及印有星期的通书里的七天一周。长度为 aa 与 bb 的两个周期,经过 lcm⁡(a,b)\operatorname{lcm}(a, b) 天才一起回到原点,而 abab 种组合中只有 ab/gcd⁡(a,b)ab/\gcd(a, b) 种会出现。对干支日与二十八宿,gcd⁡(60,28)=4\gcd(60, 28) = 4。

推导

# Chapter 3. p05: The day cycles of the almanac and their joint period.
# Stem-branch days repeat every 60 days, the 28 lodges every 28 and the
# week every 7; since 28 = 4 x 7 each lodge always falls on the same
# weekday. gcd(60, 28) = 4, so a (stem-branch, lodge) pair repeats every
# lcm(60, 28) = 420 days and only 420 of the 60 x 28 = 1,680 pairs ever
# occur. Checked by walking 1,680 days from 2027-01-01 (geng-chen, lodge
# Gui, a Friday); also the 7 weekday-planets of the lodges.
import datetime
import math

if __name__ == "__main__":
    d0 = datetime.date(2027, 1, 1)
    pairs, weekday_of = set(), {}
    first_repeat = None
    for k in range(1680):
        gz, lodge = (16 + k) % 60, (22 + k) % 28
        wd = (d0 + datetime.timedelta(days=k)).weekday()
        weekday_of.setdefault(lodge, set()).add(wd)
        if (gz, lodge) in pairs and first_repeat is None:
            first_repeat = k
        pairs.add((gz, lodge))
    print("   gcd(60, 28) = {}, lcm = {}".format(math.gcd(60, 28),
                                                 60 * 28 // math.gcd(60, 28)))
    print("   distinct (stem-branch, lodge) pairs in 1,680 days: {} of {:,}"
          .format(len(pairs), 60 * 28))
    print("   first repeated pair after {} days".format(first_repeat))
    print("   every lodge keeps one weekday: {}".format(
        all(len(v) == 1 for v in weekday_of.values())))
    print("   lodge 0 (Jiao) falls on weekday {} (0 = Monday)".format(
        sorted(weekday_of[0])[0]))
    print("   in 365 days: {} distinct pairs".format(
        len({((16 + k) % 60, (22 + k) % 28) for k in range(365)})))
    for n, m in ((60, 12), (60, 10), (60, 7), (12, 28)):
        print("   cycles {:2d} and {:2d}: gcd {:2d}, joint period {:3d} days"
              .format(n, m, math.gcd(n, m), n * m // math.gcd(n, m)))
#    gcd(60, 28) = 4, lcm = 420
#    distinct (stem-branch, lodge) pairs in 1,680 days: 420 of 1,680
#    first repeated pair after 420 days
#    every lodge keeps one weekday: True
#    lodge 0 (Jiao) falls on weekday 3 (0 = Monday)
#    in 365 days: 365 distinct pairs
#    cycles 60 and 12: gcd 12, joint period  60 days
#    cycles 60 and 10: gcd 10, joint period  60 days
#    cycles 60 and  7: gcd  1, joint period 420 days
#    cycles 12 and 28: gcd  4, joint period  84 days

从 2027 年 1 月 1 日(庚辰日、值鬼宿、星期五)起走 1,680 天,在 1,680 种可能的「干支-星宿」组合中只出现 420 种,第一个重复的组合出现在 420 天之后。因为 28=4×728 = 4 \times 7,每个星宿总落在同一个星期几;角宿总在星期四,与传统上二十八宿配七曜的对应一致。相反,干支日与值神完全没有联合周期:值神依附于节气月,而节气月的长度不是整数天。

今年的吉日明年还吉吗

因为一个太阳年不是 60 天的整数倍,同一个日期的干支每年前移 5 或 6 位,值神也随之移动。所以附着在公历日期上的吉凶判定,不能从一年带到下一年。

推导

# Chapter 3. p06: Is the same date good again next year? School A of p02
# (good officer and a yellow-road god) is computed for every date of 2027
# and 2028, with the jie times from the solar formula of p04 (within 5
# minutes of NAOJ in 2027). The two years are compared date by date
# (29 February left out). If the verdicts were unrelated, agreement would
# be pA pB + (1 - pA)(1 - pB). Jie within 10 minutes of a midnight are
# reported, since only those could move a repeat day.
import datetime
import math

LON = (285, 315, 345, 15, 45, 75, 105, 135, 165, 195, 225, 255)
GOOD_OFFICER = {1, 4, 5, 7, 8, 10}
QINGLONG = {2: 0, 8: 0, 3: 2, 9: 2, 4: 4, 10: 4, 5: 6, 11: 6,
            0: 8, 6: 8, 1: 10, 7: 10}
YELLOW = {0, 1, 4, 5, 7, 10}
EPOCH = datetime.datetime(2000, 1, 1, 12)      # JD 2451545.0

def sun_lon(jde):
    T = (jde - 2451545.0) / 36525
    L0 = 280.46646 + 36000.76983 * T + 0.0003032 * T * T
    M = math.radians(357.52911 + 35999.05029 * T - 0.0001537 * T * T)
    C = ((1.914602 - 0.004817 * T - 0.000014 * T * T) * math.sin(M) +
         (0.019993 - 0.000101 * T) * math.sin(2 * M) +
         0.000289 * math.sin(3 * M))
    om = math.radians(125.04 - 1934.136 * T)
    return (L0 + C - 0.00569 - 0.00478 * math.sin(om)) % 360

def jie_times(year):
    out = []
    for k, lon in enumerate(LON):
        t = datetime.datetime(year, k + 1, 6) - datetime.timedelta(hours=8)
        jd = 2451545.0 + (t - EPOCH).total_seconds() / 86400
        for _ in range(40):
            jd += ((lon - sun_lon(jd + 69 / 86400) + 180) % 360 - 180) \
                / 0.9856
        out.append(EPOCH + datetime.timedelta(days=jd - 2451545.0, hours=8))
    return out

def verdicts(year):
    jie = jie_times(year)
    d0 = datetime.date(year, 1, 1)
    gz0 = (d0.toordinal() + 1721425 + 49) % 60
    out = {}
    day = d0
    while day.year == year:
        n = sum(1 for j in jie if j.date() <= day)
        b, M = (gz0 + (day - d0).days) % 12, n % 12
        good = ((b - M) % 12 in GOOD_OFFICER and
                (b - QINGLONG[M]) % 12 in YELLOW)
        out[(day.month, day.day)] = good
        day += datetime.timedelta(days=1)
    near = [j for j in jie if min(j.hour * 60 + j.minute,
                                  1440 - j.hour * 60 - j.minute) <= 10]
    return out, near

if __name__ == "__main__":
    a, na = verdicts(2027)
    b, nb = verdicts(2028)
    keys = [k for k in a if k in b]
    agree = sum(1 for k in keys if a[k] == b[k]) / len(keys)
    pa = sum(a[k] for k in keys) / len(keys)
    pb = sum(b[k] for k in keys) / len(keys)
    print("   good share 2027 {:.3f}, 2028 {:.3f}".format(pa, pb))
    print("   same date, same verdict: {:.3f}; if unrelated {:.3f}".format(
        agree, pa * pb + (1 - pa) * (1 - pb)))
    both = sum(1 for k in keys if a[k] and b[k])
    print("   dates good in both years: {} of {}".format(both, len(keys)))
    print("   jie within 10 minutes of midnight: {}".format(len(na + nb)))
#    good share 2027 0.258, 2028 0.255
#    same date, same verdict: 0.685; if unrelated 0.619
#    dates good in both years: 36 of 365
#    jie within 10 minutes of midnight: 0

第一章的流派 A 判定 2027 年 0.258 的日子为吉,2028 年为 0.255。同一个日期在两年得到相同判定的比例是 0.685,而若两年的判定毫无关系,这个比例是 0.619;两年都是吉日的日期只有 36 个。略高于偶然的部分来自月份:同一日期每年都落在同一个节气月,月建因而固定,而日支在移动。2028 年没有任何节气落在离午夜 10 分钟之内,所以公式的低精度在这里不会移动任何重复日。

注

「吉日」不是日期。 某一年的嫁娶吉日表,对第二年同样的日期毫无意义。出售「今年吉日」的生意,出售的是一项每年都必须根据干支与节气重新计算的东西。

小结

六十干支与二十八宿有公因数 4,所以 1,680 种组合中只出现 420 种,每 420 天重复一次,每个星宿固定在一个星期几。一个日期在今年的判定与明年的判定一致的比例,只比偶然略高(0.685 对 0.619)。

Fan, Y. (2026). Almanac Auspicious Boolean Lattices, Jianchu Twelve-Officer Congruences and Runic Combinatorics: Constraint Satisfaction Paradoxes, Solar-Term Phase Jumps and Elder Futhark Inversion Distributions [黄历择吉布尔格、建除十二直同余与卢恩符文组合学] (Version 1.0.0). 5D Library. https://doi.org/10.5281/zenodo.24090239 · Licensed under Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0).