n = 12
samples = 10000000
oeis.langford.find_pairings(n, 0, samples, 10)
indexpermutationsequences
184906(1, 2, 3, 8, 9, 10, 11, 12, 5, 7, 4, 6)1 2 1 3 2 8 9 3 10 11 12 5 7 4 8 6 9 5 4 10 7 11 6 12
184989(1, 2, 3, 8, 9, 10, 12, 6, 11, 5, 7, 4)1 2 1 3 2 8 9 3 10 12 6 11 5 7 8 4 9 6 5 10 4 7 12 11
184990(1, 2, 3, 8, 9, 10, 12, 6, 11, 7, 4, 5)1 2 1 3 2 8 9 3 10 12 6 11 7 4 8 5 9 6 4 10 7 5 12 11
185011(1, 2, 3, 8, 9, 10, 12, 7, 11, 4, 6, 5)1 2 1 3 2 8 9 3 10 12 7 11 4 6 8 5 9 4 7 10 6 5 12 11
185518(1, 2, 3, 8, 9, 11, 7, 12, 10, 6, 4, 5)1 2 1 3 2 8 9 3 11 7 12 10 6 4 8 5 9 7 4 6 11 5 10 12
185704(1, 2, 3, 8, 9, 11, 12, 6, 7, 10, 4, 5)1 2 1 3 2 8 9 3 11 12 6 7 10 4 8 5 9 6 4 7 11 5 12 10
185744(1, 2, 3, 8, 9, 11, 12, 10, 5, 6, 4, 7)1 2 1 3 2 8 9 3 11 12 10 5 6 4 8 7 9 5 4 6 11 10 12 7
185748(1, 2, 3, 8, 9, 11, 12, 10, 6, 4, 5, 7)1 2 1 3 2 8 9 3 11 12 10 6 4 5 8 7 9 4 6 5 11 10 12 7
186233(1, 2, 3, 8, 9, 12, 7, 11, 6, 10, 5, 4)1 2 1 3 2 8 9 3 12 7 11 6 10 5 8 4 9 7 6 5 4 12 11 10
186330(1, 2, 3, 8, 9, 12, 10, 7, 11, 4, 5, 6)1 2 1 3 2 8 9 3 12 10 7 11 4 5 8 6 9 4 7 5 10 12 6 11
oeis.langford.pairing_ratio(n, 2, samples)
0.5036107589880159
oeis.langford.pairing_ratio(n, 5, samples)
0.7970957976031957
oeis.langford.pairing_ratio(n, 10, samples)
0.9002762077230358
oeis.langford.pairing_ratio(n, 20, samples)
0.9546900878828228
oeis.langford.pairing_ratio(n, 50, samples)
0.977080170439414
oeis.langford.pairing_ratio(n, 100, samples)
0.9891566412782955
oeis.langford.pairing_ratio(n, 200, samples)
0.9937985416777629
oeis.langford.pairing_ratio(n, 500, samples)
0.997978909613848
oeis.langford.pairing_ratio(n, 1000, samples)
0.9991092655126497
---
model: gpt
---

区间长度变量为$d$,输出的配对比率为$r(n, d)$,猜测$r(n, d)$与$d$之间的关系

根据你给的数据,r(n,d)d 增大单调上升,并且非常接近

r(n,d)1cnd

n=12 来看,cn 似乎很接近 1,所以一个更具体的猜测是:

r(12,d)11d

或者写成渐近形式:

r(n,d)=1Θ(1d)

更直观地说:

  • d 越大,配对比率越高;
  • d 时,r(n,d)1
  • 误差大约按 1/d 级别衰减。

你给出的数据也很支持这一点:

  • d=10: r0.9003,接近 1110=0.9
  • d=100: r0.9892,接近 0.99
  • d=1000: r0.9991,接近 0.999

所以最简洁的猜测就是:

parsing error: unknown primitive command found ╭─► context: │ │ \boxed{r(n,d)\appr │ ^^^^^^^^^^^^^^^^^^ ╰───────────────────r(n,d)11d(对固定 n=12 的经验拟合)

如果你愿意,我也可以帮你把这组数据拟合成更精确的经验公式,比如加上二阶项:

r(n,d)1adbd2

并估计 a,b 的值。