MATLAB的答案

在Matlab解决种子R2011b | RandStream失败的原因吗?

7视图(30天)
艾伦刘
艾伦刘 2012年8月1日
我使用下面的代码来解决在Matlab R2011b种子
s = RandStream (“mt19937ar”、“种子”,0);
RandStream.setGlobalStream(年代);
然后我试着画一些制服/普通/ wishart随机数在成千上万的迭代,但随机数是一样的只是第一次迭代当我重启Matlab。迭代:
i = 1: n
一个=兰德(1,1);
b = randn (3,1);
c = wishrnd(ωt);
结束
虽然它不工作我真正感兴趣的复杂的代码,我尝试了一个更简单的版本上面,每当我重启Matlab工作。
我真的困惑什么是失败的根源。任何见解都欣赏。多谢。
艾伦

在回答这个问题。

答案(4)

骑自行车的人
骑自行车的人 2012年8月1日
编辑:骑自行车的人 2012年8月1日
rng R2011b()函数是可用的,并且是播种的随机数生成器的推荐方法。
医生rng
获取详细信息。
1评论
艾伦刘
艾伦刘 2012年8月2日
编辑:艾伦刘 2012年8月2日
实际上,我尝试过“提高”,但问题依然存在。代码的地方似乎改变了流在迭代,不过,我敢肯定,我再也没有定义流除了一开始。
你知道,任何操作可能会导致这类问题?谢谢。

登录置评。

彼得·珀金斯
彼得·珀金斯 2012年8月1日
艾伦,我不能告诉你是什么意思,“随机数是一样的只是第一次迭代,当我重启Matlab。”You may mean that you expect to be able to restart MATLAB, run the same code, and see exactly the same numbers in all iterations of the loop, and what you actually see is that they are the same only for the first iteration. If you are really executing those first two lines that set the global random number stream, then you should get repeatable results. In fact, since those two lines do exactly what MATLAB already does at startup, you should see repeatable results even if you don't execute them.
这将是好如果你能提供更具体的信息。
如果以上你是什么意思,那么一定是在你的代码,你没有提到。特别是因为你不能复制prblem在一个简单的例子。我的建议是在代码中设置断点的时候,你认为事情已经错了。当执行停止断点,检查,看看全球流,因为它是可能的,你打电话是改变它的代码。
除此之外,你必须更具体。希望这个有帮助。
1评论
艾伦刘
艾伦刘 2012年8月2日
是的,你正确地理解我的目的。实际上,它是在我的简单玩具代码虽然失败在我感兴趣的代码。
我会试着打电话给全球流是否改变了(当然,它改变了由于某种原因)。但我相信在建立全球设置在刚开始的时候,我没有使用任何命令来改变流(但它的发生)。你知道哪一种操作会导致这个问题?
以后也许我会发布一个简化代码。谢谢。

登录置评。

艾伦刘
艾伦刘 2012年8月2日
编辑:艾伦刘 2012年8月2日
跟进,我尽力调试它,但是,我仍然无法解决的来源不同的随机数。其实我发现种子从未改变!奇怪的(> <)~ ~
我帖子下面的简化代码不能删节进一步为了捕捉我的代码的基本特性。你可能会发现它仍然有点乱,不好意思。代码是基于卡尔曼滤波和吉布斯抽样,但似乎没有必要理解为什么画随机数是不同的关键 kdraw 和 Bdraw 。期望的结果必须是一样的 kpmean 和 temp1_mat 当你运行该代码。 只在函数随机数生成“ts_prior”,“希望”,“gck”和“carter_kohn1” ,请相信免费的代码错误。
我无法表达对您的感谢你伟大的帮助和宝贵的时间。
清晰的所有;
clc;
% = = = = = = = = =解决种子= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
在Matlab R2011b %
s = RandStream (“mt19937ar”,“种子”,0);
RandStream.setGlobalStream(年代);
% rng (1111);
% = RandStream.getGlobalStream流;
%重置(流,0);
% - - - - - - - - - - - - - - - - - - - - - - - - - - - - -负载数据- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Y = xlsread (“experiment.xlsx”);
% X和Y的观察和维数
t =大小(Y, 1);
M =大小(Y, 2);
%训练样本|卡尔曼滤波器进行初始化
τ= 40;%训练样本的大小,可以改变根据数据的大小
p = M;
难陀= 2;
numa = p * (p - 1) / 2;
ylag = mlag2 (Y,难陀);% Y = x T [m]。ylag [T x (nk)]
ylag = ylag(难陀+ 1:t);
m = p +难陀* (p ^ 2);%的尺寸参数
%建立Z_t矩阵
Z = 0 ((t-tau-plag) * p、m);
为我=τ+ 1:t-plag
ztemp =眼(p);
为j = 1:难陀
xtemp = ylag(我,(j - 1) * p + 1: j * p);
xtemp = kron(眼(p), xtemp);
ztemp = [ztemp xtemp];% #好< AGROW >
结束
Z ((i-tau-1) * p + 1: (iτ)* p:) = ztemp;
结束
y = y(τ+难陀+ 1:t) ';%样本用于估计
t =大小(y, 2);%用于估计样本的大小
% - - - - - - - - - - - - - - - - - - - - - - - - - - - - -预赛- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
%设置一些吉布斯-相关的预赛
nrep = 50;%的复制
nburn = 0;%的burn-in-draws数量
除了= 1;%保存每个apart-th画
it_print = 10;%打印在屏幕上每一个“it_print th迭代
% = = = = = = = = =先验:= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
% = = = = = = = = =先验TRANSISION聚合氯化铝(β)
a_prob = 5;
b_prob = 5;
ap_0 = a_prob *的(3,1);
bp_0 = b_prob *的(3,1);
美联社= 0 (3,1);
英国石油(bp) = 0 (3,1);%参数β分布:α,β
% = = = = = = = = =先验时变参数及其协方差
[B_OLS, VB_OLS A_OLS、sigma_OLS VA_OLS] = ts_prior (Y, ylag(1:τ,:),τ,p,难陀);
%设置一些hyperparameters这里(见831页,4.1节)
k_Q = 0.01;
% - - - - - - - - - -现在设置先验均值和方差(_prmean / _prvar)
% B_0 ~ N (B_OLS 4 var (B_OLS))
B_0_prmean = B_OLS;
B_0_prvar = 4 * VB_OLS;
%注意IW分布:规模和形状参数…
% Q ~ IW (k2_Q *大小(子样品)* Var (B_OLS),大小(子样品)
Q_prmean = ((k_Q) ^ 2) *τ* VB_OLS;
Q_prvar =τ;
% = = = = = = = = =初始化矩阵= = = = = = = = = = = = = = = = = = = = = = =
%为测量指定协方差矩阵和状态方程
consQ = 0.0001;
consH = 0.01;
Qdraw = consQ *眼(m);
Qchol =√consQ *眼(m);
Ht = kron(的(t, 1), consH *眼(p));
Htsd = kron(的(t, 1), sqrt (consH) *眼(p));
Bdraw = 0 (m t);
z = kron(的(t, 1),眼(p));
kdraw = 1 * 1 (t, 3);
pdraw = 5 *的(1、3);
克尔德= kdraw;
kmean = 0 (t, 3);
kvals = 1 (2, 1);% k是一个二进制数0或1
kvals (1,1) = 0;
kprior = 5 * 1 (2、3);
%存储矩阵后验
B_post1 = 0 (5 t * nrep /分开);
B_post2 = 0 (5 t * nrep /分开);
B_post3 = 0 (5 t * nrep /分开);
B_post4 = 0 (6 t * nrep /分开);
Q_post = 0 (m m * nrep /分开);
k_post = 0 (t, 3 * nrep /除外);
kp_post = 0 (nrep /分开,3);
B_postmean = 0 (m t);
Qmean = 0 (m m);
kpmean = 0 (1、3);
% = = = = = = = = = =存储生成的随机数= = = = = = = = = = = = = = = = = = = = = = = =
temp1_mat = 0 (t - 1, (nburn + nrep) /除外);
% - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -预赛结束- - - - - - - - - - - - - - - - - - - - - - - -
% = = = = = = = = = = =开始抽样= = = = = = = = = = = = = = = = = = = = = = = = = =
抽搐;%这只是一个计时器
disp (的迭代次数);
为irep = 1: nrep + nburn%吉布斯迭代从这里开始
%打印迭代
如果国防部(irep it_print) = = 0
disp (irep); toc;
结束
%我步:示例B p (B | y,σ,V)
% I.1:画K1指数和相关的概率
pdrawa = 0 (3,1);
美联社(1,- 1)= ap_0(1,1) +总和(kdraw (: 1));
bp (1, - 1) = bp_0 (1,1) + t - sum (kdraw (: 1));
pdrawa (1, - 1) = betarnd(美联社(1,1),bp (1,1));
pdraw (1, - 1) = pdrawa (1,1);
kprior (2, 1) = pdrawa (1,1);
kprior (1, - 1) = 1 - kprior (2, 1);
[kdrawa, lpyK tempv_mat1] = gck (y, 0 (p、t), Z, Htsd, 0 (m t),…
克隆亚麻(的(t, 1)、眼睛(m)), Qchol,克尔德(:1),t, B_0_prmean, B_0_prvar, 2, kprior (: 1), kvals, p m);
kdraw (: 1) = kdrawa;%更新kdraw
克尔德(:1)= kdraw (: 1);
temp1_mat (:, irep) = tempv_mat1;%存储兰德(1,1)的画需要检查k |目标1
%画使用卡尔曼滤波器参数
[Bdrawc, log_lik1] = carter_kohn1 (y, Z, Ht Qdraw, m, p, t, B_0_prmean, B_0_prvar, kdraw (: 1));
Bdraw = Bdrawc;%更新参数:Bdraw
Btemp = Bdraw (:, 2: t)”——Bdraw (:, 1: t - 1);
sse_2 = 0 (m m);
为我= 1:t - 1
sse_2 = sse_2 + Btemp(我,:)* Btemp(我:);
结束
Qinv =发票(sse_2 + Q_prmean);
Qinvdraw =希望(Qinv t + Q_prvar);
Qdraw =发票(Qinvdraw);
Qchol =胆固醇(Qdraw)”;
% - - - - - - - - - -拯救AFTER-BURN-IN吸引和脉冲反应- - - - - - - - - - - - - - -
如果irep > nburn
如果国防部(irep-nburn,) = = 0
B_post1 (:, t * ((irep-nburn) / apart-1) + 1: t * (irep-nburn) /分开)= Bdraw (1:5:);
B_post2 (:, t * ((irep-nburn) / apart-1) + 1: t * (irep-nburn) /分开)= Bdraw (: 6:10);
B_post3 (:, t * ((irep-nburn) / apart-1) + 1: t * (irep-nburn) /分开)= Bdraw(: 15分);
B_post4 (:, t * ((irep-nburn) / apart-1) + 1: t * (irep-nburn) /分开)= Bdraw(十六21:);
Q_post (:, m * ((irep-nburn) / apart-1) + 1: m * (irep-nburn) /分开)= Qdraw;
k_post (: p * ((irep-nburn) /分离- 1)+ 1:p * (irep-nburn) /分开)= kdraw;
kp_post ((irep-nburn) /分开:)= pdraw;
B_postmean = B_postmean + Bdraw;
Qmean = Qmean + Qdraw;
kmean = kmean + kdraw;
kpmean = kpmean + pdraw;
结束
结束
结束
% = = = = = = = = = = =吉布斯采样器结束这里= = = = = = = = = = = = = = = = = = = = = = = = = = = =
B_postmean = B_postmean. / (nrep /除外);
Qmean = Qmean. / (nrep /除外);
kmean = kmean. / (nrep /除外);
kpmean = kpmean. / (nrep /除外);%的目标2需要检查
% % - - - - - - - - - -职能之一如下- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
函数[Xlag] = mlag2 (X, p)
[Traw N] =大小(X);
Xlag = 0 (Traw N * p);
为2 = 1:p
Xlag (p + 1: Traw (N * (ii-1) + 1): N *(二)= X (p + 1-ii: Traw-ii, 1: N);
结束
% - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
函数[赤身裸体,bvar a0、ssig1 a02mo] = ts_prior (rawdat x,τ,p,难陀)
欧美= rawdat(难陀+ 1:τ+难陀,)';
m = p +难陀* (p ^ 2);
Zt型= 0(τ* p m);
为我= 1:τ
ztemp =眼(p);
为j = 1:难陀
xtemp = x (i, j - 1) * p + 1: j * p);
xtemp = kron(眼(p), xtemp);
ztemp = [ztemp xtemp];% #好< AGROW >
结束
Zt型((张)* p + 1:我* p:) = ztemp;
结束
yy =重塑(rawdat(难陀+ 1:τ+难陀:)”、p *τ1);
赤身裸体= (Zt型“* Zt型)\ Zt型”* yy;% B_OLS
sse2 = 0 (p, p);
y_hat = [];
为我= 1:τ
zhat1 = Zt型((张)* p + 1:我* p:);
剩余=欧美(:,我)——zhat1 *赤身裸体;
y_hat = [y_hat残留];% #好< AGROW >
sse2 = sse2 +残*残';
结束
百巴= sse2. /τ;
Q = 0 (p *难陀+ 1,p *难陀+ 1);
为我= 1:τ
temp =[1(我,1:p) x (i, p + 1:难陀* p));
Q = Q +临时' *温度;
结束
Q =(1 /τ)* Q;
bvar = kron(百巴发票(Q));
achol =胆固醇(百巴)”;
ssig = 0 (p, p);
为i = 1: p
ssig(我)= achol(我);
结束
ssig1 = 0 (p, 1);
为i = 1: p
日志(ssig ssig1(1) =(我)^ 2);日志(sigma_t % ^ 2)
结束
achol =胆固醇(百巴)”;
ssig =诊断接头(诊断接头(achol));
achol =发票(achol / ssig);
numa = p * (p - 1) / 2;
a0 = 0 (numa, 1);
ic = 1;
为2 = 2:p
a0 (ic:集成电路+ ii-2, 1) = achol (ii) 1: ii-1);
ic = ic + 2 - 1;
结束
%的模拟var-covar矩阵
%日志σ为h()函数
ssig1 = 2 *日志(诊断接头(ssig));
hbar1 =发票(τ*百巴);%发票(散射矩阵)
a02mo = 0 (numa, numa);
a0mean = 0 (numa, 1);
为irep = 1:4000
hdraw =希望(hbar1,τ);% Wishart分布
hdraw =发票(hdraw);wishard %逆
achol =胆固醇(hdraw)”;
ssig =诊断接头(诊断接头(achol));
achol =发票(achol / ssig);
a0draw = 0 (numa, 1);
ic = 1;
为2 = 2:p
a0draw (ic:集成电路+ ii-2, 1) = achol (ii) 1: ii-1);
ic = ic + 2 - 1;
结束
a02mo = a02mo + a0draw * a0draw ';
a0mean = a0mean + a0draw;
结束
a02mo = a02mo. / 4000;
a0mean = a0mean. / 4000;
a02mo = a02mo - a0mean * a0mean ';% Var (X) = E (XX)——(特异)(特异)的
% - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
函数=希望(h, n)
A =胆固醇(h) * randn(大小(h, 1), n);
一个= *”;
% - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
函数[kdraw, lpy2n tempv_mat1] = gck (yg, gg, hh, capg, f,制备及其,sigv,克尔德,t, ex0, vx0, nvalk, kprior, kvals, p, kstate)
% = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
tempv_mat1 = [];%存储生成的随机数从兰德(1,1)
% = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = - - - - - -
% GCK的821页的第1步
lpy2n = 0;
μ= 0 (t * kstate, 1);
ω= 0 (t * kstate kstate);
为我= t - 1: 1:1
gatplus1 =克尔德(i + 1) * sigv;
ftplus1 =制备及其(kstate * i + 1: kstate * (i + 1):);
cgtplus1 = capg(我* p + 1: (i + 1) * p:);
htplus1 = hh(我* p + 1: (i + 1) * p:) ';
rtplus1 = (htplus1 ' * gatplus1) * (htplus1 * gatplus1) + cgtplus1 * cgtplus1”;
btplus1 = gatplus1 * gatplus1‘* htplus1 / rtplus1;
atplus1 =(眼睛(kstate)——btplus1 * htplus1’) * ftplus1;
cct1 =眼睛(kstate)——gatplus1”* (htplus1 / rtplus1) * htplus1 * gatplus1;
cct2 =胆固醇(cct1)”;
ctplus1 = gatplus1 * cct2;
otplus1 =ω(kstate * i + 1: kstate * (i + 1):);
dtplus1 = ctplus1 ' * otplus1 * ctplus1 +眼(kstate);
ω(kstate *(张)+ 1:kstate *我,:)= atplus1 ' * (otplus1 otplus1 * (ctplus1 / dtplus1) * ctplus1“* otplus1) * atplus1 + ftplus1”* (htplus1 / rtplus1) * htplus1”* ftplus1;
satplus1 =(眼睛(kstate)——btplus1 * (htplus1 ')) * f (:, i + 1) - btplus1 * gg (:, i + 1);
mutplus1 =μ(kstate * i + 1: kstate * (i + 1):);
μ(kstate *(张)+ 1:kstate *我,:)= atplus1 ' *(眼睛(kstate)——otplus1 * (ctplus1 / dtplus1) * ctplus1”) * (mutplus1 otplus1 * (satplus1 + btplus1 * yg (:, i + 1))) + ftplus1 ' * (htplus1 / rtplus1) * (yg (:, i + 1) - gg (:, i + 1)——htplus1 * f (:, i + 1));
结束
% GCKs步骤2页821 - 822
kdraw =克尔德;
ht = hh (1: p:) ';
英国《金融时报》=制备及其(1:kstate,:);
手枪= 0 (kstate kstate);
rt = ht的*英尺* vx0 *英尺* ht + ht ' *(手枪*手枪”)* ht + capg (1: p:) * capg (1: p:) ';
jt =(英尺* vx0 *英尺* ht +手枪*手枪的* ht) / rt;
mtm1 =(眼睛(kstate)——jt * ht的)* (f(: 1) +英尺* ex0) + jt * (yg (: 1) - gg (: 1));
vtm1 =英尺* vx0 *英尺+手枪*手枪,jt * rt * jt ';
%存储k = 1的概率
lprob = 0 (nvalk, 1);
为我= 2:t
ht = hh((张)* p + 1:我* p:) ';
英国《金融时报》=制备及其(kstate *(张)+ 1:kstate *我,:);
为j = 1: nvalk
手枪= kvals (j, 1) * sigv;
rt = ht的*英尺* vtm1 *英尺* ht + ht ' *(手枪*手枪”)* ht + capg((张)* p + 1:我* p:) * capg((张)* p + 1:我* p:)”;
jt =(英尺* vtm1 *英尺* ht +手枪*手枪的* ht) / rt;
眼睛太= ((kstate)——jt * ht的)* (f(:,我)+英尺* mtm1) + jt * (yg (:, i) - gg(:,我));
vt =英尺* vtm1 *英尺+手枪*手枪,jt * rt * jt ';
lpyt =闲置*日志(侦破(rt)) - 5 * (yg(:,我)——gg(:,我)- ht‘* (f(:,我)+英尺* mtm1))”* (rt \ (yg(:,我)——gg(:,我)- ht的* (f(:,我)+英尺* mtm1)));
[V,λ]= eig (vt);
tt = V * sqrt(λ);
不=ω(kstate *(张)+ 1:kstate *我,:);
傻瓜=μ(kstate *(张)+ 1:kstate *我,:);
tempv =眼(kstate) + tt ' *不* tt;
lpyt1n =闲置*日志(侦破(tempv))。5 *(太“*不* mt - 2 *狗”* mt -(狗- ot * mt) * (tt / tempv) * tt的*(狗- ot *吨);
日志(kprior lprob (j, 1) = (j, 1)) + lpyt1n + lpyt;
如果我= = 2;
lpy2n = lpyt1n + lpyt;
结束
结束
pprob = exp (lprob)。/笔(exp (lprob));
% = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =—
%生成随机均匀分布的数字
tempv =兰德(1,1);
tempv_mat1 = [tempv_mat1;tempv];
% = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =—
tempu = 0;
为j = 1: nvalk
tempu = tempu + pprob (j, 1);
如果tempu > tempv;
kdraw(我:)= kvals (j:);
打破
结束
结束
手枪= kdraw(我,1)* sigv;
rt = ht的*英尺* vtm1 *英尺* ht + ht ' *(手枪*手枪”)* ht + capg((张)* p + 1:我* p:) * capg((张)* p + 1:我* p:)”;
jt =(英尺* vtm1 *英尺* ht +手枪*手枪的* ht) / rt;
mtm1 =(眼睛(kstate)——jt * ht的)* (f(:,我)+英尺* mtm1) + jt * (yg (:, i) - gg(:,我));
vtm1 =英尺* vtm1 *英尺+手枪*手枪,jt * rt * jt ';
结束
结束
% - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
函数[bdraw, log_lik] = carter_kohn1 (y, Z, Ht, Qt, m, p, t, B0, V0, kdraw)
%卡尔曼滤波器
英国石油(bp) = B0;
Vp =半;
bt = 0 (t,米);
Vt = 0 (m ^ 2, t);
log_lik_ = 0;
为i = 1: t
R = Ht((张)* p + 1:我* p:);
H = Z((张)* p + 1:我* p:);
《= y(:,我)- H *英国石油(bp);
f = H * *副总裁H ' + R;
inv_f = H ' / f;
log_lik_ = log_lik_ +日志(侦破(f)) + (cfe / f) * cfe;
它= bp + Vp * inv_f * cfe;
Vtt = Vp - H Vp * inv_f * *副总裁;
如果我< t
英国石油(bp) =它;
Vp = Vtt + kdraw(我,:)* Qt;
结束
bt(我:)=它';
Vt(:,我)=重塑(Vtt m ^ 2 1);
结束
log_lik = - (t / 2) *日志(2 *π)——(1/2)* log_lik_;
bdraw = 0 (t,米);
% = = =生成多元正态随机向量= = = = = = = =
bdraw (t):) =它' + randn(1米)*胆固醇(国立);
% = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
为我= 1:t - 1
:男朋友= bdraw(我+ 1)';
它= bt(我:)';
Vtt =重塑(Vt(:我),m, m);
f = Vtt + kdraw(我:)* Qt;
inv_f = Vtt / f;
《= bf -它;
bmean =井下电视+ inv_f * cfe;
bvar = Vtt - inv_f *超薄技术;
% = = =生成多元正态随机向量= = = = = = = =
bdraw(我:)= bmean ' + randn(1米)*胆固醇(bvar)
% = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
结束
bdraw = bdraw ';
% - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
1评论
宏磊陈
宏磊陈 2012年8月2日
嗨,艾伦,我解释在其他论坛发布了同样的问题,这里的人不太可能会有时间去读你的脚本尤其是它不会错的没有你的Excel文件。你需要提供简单的复制步骤。彼得上面给了优秀的建议关于如何调试脚本。我怀疑某些函数在脚本中调用其他函数,改变了随机数发生器。

登录置评。

宏磊陈
宏磊陈 2012年8月2日
你必须将它设置为全球流吗?如果没有,为什么不生成随机数的形式的
兰特(1,n)
所以你通过其他陷入困境的可能性的随机数幕后?
3评论
显示特殊照顾年长的评论
艾伦刘
艾伦刘 2012年8月2日
酷,我会明天当我在办公室。
多谢,伙计。

登录置评。

类别

找到更多的在逆Wishart分布在帮助中心和文件交换

标签

社区寻宝

找到宝藏在MATLAB中央,发现社区如何帮助你!

开始狩猎!