关于「先到先得的代价:Littlewood 两档规则与被挤掉的整天订单」,《空间租赁收益管理与档期装箱》给出的定量结论是什么?

结论:第一章在一个 20 个棚日的示意月份里,用精确期望算出保护 10 个棚日可使月收入从 47323.9 元升到 57432.1 元,多 21.36 %,但模拟显示保护在 0.1900 的月份里少收钱。

数据口径:书中的棚日、价格、需求与到达概率均为示意;全部计算由书中程序生成;本书是非个人化经营建议,所有价格明码标价,稀缺提示只来自真实档期,小工具给出的价格表须经营者人工复核后发布。 第一章在一个 20 个棚日的示意月份里,用精确期望算出保护 10 个棚日可使月收入从 47323.9 元升到 57432.1 元,多 21.36 %,但模拟显示保护在 0.1900 的月份里少收钱。

English version

对照数据表

下表数字由本章 Python 程序原样打印,可重新运行复现。 程序只用 Python 标准库与结论吻合的打印值: 10, 20, 47323.9, 57432.1

studio month: C20 days; full-day 4800 yuan (Poisson 9),
short booking blocks a day1800 yuan (Poisson 16), books first
y P(DH>=y)4800*P accept low at x=y? E[revenue]
01.0000 4800.0 no 47323.9
40.9788 4698.1 no 50928.1
60.8843 4244.7 no 53933.4
80.6761 3245.3 no 56555.9
90.5443 2612.9 no 57265.6
100.4126 1980.4 no 57432.1
110.2940 1411.3 yes 57060.1
120.1970 945.6 yes 56224.5
140.0739 354.5 yes 53593.1
Littlewood y*10 (brute-force best y = 10)
first-come-first-served47323.9 yuan
protect 10 days57432.1 yuan (+10108.2, 0.2136)
perfect hindsight bound62331.0 yuan
share of the FCFS-to-hindsight gap closed0.674
查看完整验算程序
# Chapter 1. p11: Littlewood's two-class rule for one studio month.
# Model: capacity C studio-days; low-fare short bookings (each blocks a
# whole day) arrive first, full-day shoots arrive later. Accept a low
# booking while remaining capacity x satisfies f_L >= f_H * P(D_H >= x);
# equivalently protect y* days for the high class (Littlewood 1972).
# Expected revenue is computed exactly by enumerating Poisson demand.
# All fares and demand means are illustrative (示意), not market data.
import math

C = 20                 # 本月可售棚日(示意)
F_H, F_L = 4800, 1800  # 整天商拍价、短订占日价(元,示意)
M_H, M_L = 9.0, 16.0   # 两档需求的泊松均值(示意)
N = 80                 # 截断上限,尾部概率小于 1e-20


def pois(mean):
    p = [math.exp(-mean)]
    for k in range(1, N + 1):
        p.append(p[-1] * mean / k)
    return p


PH, PL = pois(M_H), pois(M_L)
# tail[k] = P(D_H >= k)
TAIL = [math.fsum(PH[k:]) for k in range(N + 1)]
# emin[r] = E[min(D_H, r)] = sum_{k=1..r} P(D_H >= k)
EMIN = [math.fsum(TAIL[1:r + 1]) for r in range(C + 1)]


def revenue(y):
    # 低价预订上限 b = C - y;低价先到,高价后到
    b = C - y
    terms = []
    for d, p in enumerate(PL):
        s = min(d, b)
        terms.append(p * (F_L * s + F_H * EMIN[C - s]))
    return math.fsum(terms)


def hindsight():
    terms = []
    for dh, ph in enumerate(PH):
        sh = min(dh, C)
        for dl, pl in enumerate(PL):
            sl = min(dl, C - sh)
            terms.append(ph * pl * (F_H * sh + F_L * sl))
    return math.fsum(terms)


print("studio month: C=%d days; full-day %d yuan (Poisson %.0f),"
      % (C, F_H, M_H))
print("short booking blocks a day: %d yuan (Poisson %.0f), books first"
      % (F_L, M_L))
print(" y  P(DH>=y)  %d*P   accept low at x=y?  E[revenue]" % F_H)
ystar = max(y for y in range(C + 1) if F_H * TAIL[y] >= F_L)
for y in (0, 4, 6, 8, 9, 10, 11, 12, 14):
    mv = F_H * TAIL[y]
    print("%2d   %.4f   %7.1f   %-18s  %8.1f"
          % (y, TAIL[y], mv, "yes" if F_L >= mv else "no", revenue(y)))
best = max(range(C + 1), key=revenue)
r0, rs, rh = revenue(0), revenue(ystar), hindsight()
print("Littlewood y* = %d (brute-force best y = %d)" % (ystar, best))
print("first-come-first-served: %.1f yuan" % r0)
print("protect %d days:          %.1f yuan  (+%.1f, %.4f)"
      % (ystar, rs, rs - r0, rs / r0 - 1))
print("perfect hindsight bound: %.1f yuan" % rh)
print("share of the FCFS-to-hindsight gap closed: %.3f"
      % ((rs - r0) / (rh - r0)))

唯一数据源与引用

范雅福(Fan, Y.). (2026). 《空间租赁动态收益管理与档期装箱、舱位控制嵌套保护水平与碎片空置消除母典》第 1 章「先到先得的代价:Littlewood 两档规则与被挤掉的整天订单」(版本 1.0.0). 5D Library. DOI 待发布

DOI 待发布 · 专著页正式发布后上线(本站暂无专著页) · 5DL-B-282 · v1.0.0 · CC-BY-NC-ND-4.0

BibTeX
@book{fan2026282,
  author    = {Fan, Yafu},
  title     = {Dynamic Revenue Management and Slot Packing for Space Rentals: Nested Protection Levels and the Elimination of Fragmented Idle Time},
  year      = {2026},
  publisher = {5D Library},
  edition   = {1.0.0},
  note      = {DOI pending}
}

本书其他章节结论