The following plots are in www.stat.ucla.edu/~frederic/sums . I'll refer
to plots like p3.ps and p3.pdf simply as p3.
1) sum1.
ratio (est/sim) versus n, for n = 1 to 100,
alpha=.66, q=.02, nsim=10K.
Estimates are made using p1=sqrt(q).
Estimates and simulated quantiles
are reported in out1.txt.
2) sum2.
ratio (est/sim) versus n, for n = 1 to 100,
alpha=.66, q=.02, nsim=100K.
Estimates are made using p1=sqrt(q). Estimates and simulated quantiles
are reported in out2.txt.
3) sum3.
ratio (est/sim) versus n, for n = 1 to 100,
alpha=.66, q=.02, nsim=1M.
Estimates are made using p1=sqrt(q).
Estimates and simulated quantiles
are reported in out3.txt.
4) p1.
Optimal value of the parameter p_1 versus n,
for n = 1 to 100, alpha=.66, q=.02, nsim=10K.
The line Y=sqrt(q) is overlaid.
5) p2.
Optimal p_1 versus n, for n = 1 to 100,
alpha=.66, q=.02, nsim=100K. The line Y=sqrt(q) is overlaid.
6) p3.
Optimal p_1 versus n, for n = 1 to 100,
alpha=.66, q=.02, nsim=1M. The line Y=sqrt(q) is overlaid.
7) p1_v_q_1.
Optimal p_1 versus q, where q is the fraction
corresponding to the quantile we're after (i.e. if we want the 0.05
quantile, then q = 0.05). alpha=.66, n=20, nsim=1M.
The line Y = X is overlaid.
8) ratio_v_quant_1.
ratio (est/sim) versus q.
Values plotted are q = 0.0001, 0.0010, 0.005,
0.01, 0.02, ..., 0.09, .1,.2,...,.9, for alpha=.66, n=20, nsim=1M.
Estimates are made using p1=sqrt(q). Estimates and simulated quantiles are
reported in out4.txt.
9) sum4.
ratio (est/sim) versus n, for n =
1,2,...,100,1000,2000.
alpha=.66, q=.02, nsim=1M.
Estimates are made using p1=sqrt(q).
Estimates and simulated quantiles
were added to the end of out3.txt.
10) p4.
Optimal p_1 versus n, for n =
1,2,...,100,1000,2000.
alpha=.66, q=.02, nsim=1M. The line Y=sqrt(q) is overlaid.
11) sum5.
ratio (est/sim) versus n, for n =
1 to 100,
alpha=.80, q=.02, nsim=1M.
Estimates are made using p1=sqrt(q).
Estimates and simulated quantiles
are reported in out5.txt.
12) p5.
Optimal p_1 versus n, for n =
1 to 100,
alpha=.80, q=.02, nsim=1M. The line Y=sqrt(q) is overlaid.
13) p1_v_q_2.
Optimal p_1 versus q,
for alpha=.80, n=20, nsim=1M.
The line Y = X is overlaid.
14) ratio_v_quant_2.
ratio (est/sim) versus q,
for alpha=.80, n=20, nsim=1M.
Estimates are made using p1=sqrt(q). Estimates and simulated quantiles are
reported in out6.txt.
15) ratio_v_a_1.
ratio (est/sim) versus alpha,
for n=20, nsim=1M.
Estimates are made using p1=sqrt(q). Estimates and simulated quantiles are
reported in out7.txt.
16) p1_v_a_1.
Optimal p_1 versus alpha,
for n=20, nsim=1M. The line Y = sqrt(q) is overlaid.
17) p1_v_q_3.
Optimal p_1 versus q,
for alpha=.66, n=20, nsim=1M, for q between 0.01 and 0.35.
The line Y = X is overlaid.
17) p1_v_q_4.
Optimal p_1 versus q,
for alpha=.66, n=20, nsim=1M, for q between 0.01 and 0.35.
A quadratic curve is overlaid.