China DOS Union

I don't know if this post is appropriate to be posted here. The post I made can find the trace of the command line..

This tutorial is very exciting to me because I've been looking for it. Let me tell you, this is not cmd, but more fun than cmd----MATLAB

If you like mathematics, this software is a must.. Of course, this tutorial is about MATLAB, very suitable for getting started..

The early versions of MATLAB were developed for DOS. Using this software is like doing math problems in a notebook. Everything can be command line. Many operation commands are very similar to DOS commands..

I just found a command "!" today. Using this "!" can realize the complete "portability" of batch commands on MATLAB. You can treat it as a special CMD. Just add a "!" in front when calling external commands.
If you are interested, install MATLAB and play around,

Tutorial download:
http://upload.cn-dos.net/img/376.rar

[ Last edited by plp626 on 2008-5-14 at 09:36 PM ]

In the computer science major, the course "Image Processing" also uses MATLAB

Well, it can also be used to write "pornographic movies",

Thanks for sharing the good stuff!

%Print T-distribution table... 19:32 2008-5-13
function main()
%Define main function
!set/p=%time%<nul>time
v=[0.25 0.10 0.05 0.025 0.01 0.005];%Set quantile vector
fprintf('n\\α')
for i=1:1:6
fprintf('%10.4f',v(i))
end

fprintf('\n')
for j=1:1:1-15 %Set degree of freedom range 1-15
fprintf('%5.1d ',j)
for i=1:1:6
fprintf('%10.4f',dt(v(i),j))
if i==6 fprintf('\n')
end
end
end
!echo\ %time%>>time
!for /f "tokens=*" %a in (time)do timediff %a 0

%For degree of freedom n, find the lower limit of integration when the integral value is v (quantile).
function r=dt(v,n) %%//////dt(v,n)
h=1; %Initial step size
a=0.5; %Initial value a, then iterate

%Print T Distribution Table {19:32 2008-5-13|plp626}

function m() %Define main function

!set/p=%time%<nul>time

v=;%Set quantile vector

fprintf('n╲α')

for i=1:1:6

fprintf('%10.4f',v(i))

end



fprintf('\n')

for j=1:1:45                       %Set degree of freedom range 1-45

fprintf('%5.1d  ',j)

   for i=1:1:6

      fprintf('%10.4f',dt(v(i),j))

      if i==6 fprintf('\n')

      end

   end

end

!echo\ %time%>>time

!for /f "tokens=*" %a in (time)do timediff %a 0



%For degree of freedom n, find the lower limit of integration when the integral value is v (quantile).

function r=dt(v,n)   %%//////dt(v,n)

h=1;                 %Initial step size

a=0.5;               %Initial value a, then iterate to the value we need.

ta=t(a,n);           %Initial integral value: 0.5 is the lower limit of integration.

%%%%Here use variable step size 3-4-5 approximation method (original)

while abs(ta-v)>0.000001 %Set precision

        while ta>v

		a=a+h;ta=t(a,n);

        end

	h=0.3*h;a=a-h;

	while ta<v

		a=a-h;ta=t(a,n);

        end

	h=0.4*h;a=a+0.5*h;

end

r=a;



%Calculate the t integral from  when degree of freedom is n. For specified n, t(a,n) is a decreasing function.

function r=t(a0,n)                %%/////t(a0,n)

syms x

f=gamma((n+1)/2)/(sqrt(n*pi)*gamma(n/2))*(1+x*x/n)^(-(n+1)/2);%T distribution density function



f=simplify(f);

g=int(f,x,a0,inf);

r=numeric(g);

Attach another printed result:


[ Last edited by plp626 on 2008-5-15 at 07:55 AM ]

Good stuff... Scilab is really a pain to use.. This one is a bit simpler.

%Print normal distribution table  

%$Author: ****** Date: 2008-5-23 $

function f()

p=0:9;                                        %The second decimal place

fprintf('   z  ')

for i=1:10

fprintf('%-8.1d',p(i))

end

fprintf('\n')

for z=0:0.1:2.0                               %The first decimal place

   fprintf('%4.1f',z)

   for i=1:1:10

      fprintf('%8.4f',int_f(z+p(i)/100))

      if i==10 fprintf('\n')

      end

   end

end

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

function r=int_f(a)

syms x

r=double(int(1/sqrt(2*pi)*exp(x^2/-2),-inf,a));

[ Last edited by plp626 on 2010-10-5 at 15:29 ]

The problem has been solved.

My graduation thesis title: Parameter Estimation of the Coefficient of Variation (Original)

I studied software M crazily for half a month, but I still don't know how to use the command of @ anonymous function and the setting of global variables.

Now I post the graduation project program for everyone to share. If there are any deficiencies or errors in the program, please point them out. I will be very grateful.

D:\MATLAB65\WORK\H Distribution
├─Upper Quantile Table of H Distribution
│ date.txt
│ finda.m
│ findp.m
│ main.m
│ print_h.m
│ readme.txt
│
└─Related Function Figures
cdf_cont.m
cdf_mesh.m
plot_pdf.m
plot_pdfab.m
readme.txt
---------------------------------------------------------------------------------------------------------
---------- README.TXT
//** findp.m is an independent program, respectively called by finda.m and print_h.m **//
findp.m Find probability given random variable value
Parameter list (coefficient of variation, degrees of freedom, random variable value)

When n is large (greater than 287), gm_n is directly displayed as inf (infinity) by the computer, so 287 is the upper limit of degrees of freedom that the computer can calculate. First, integrate the z int, then integrate the x quadl, and then let (ninf → ∞). Here, ninf is determined according to the values of n and v. Due to the particularity of the function (it has extremely decreased near 3, refer to cdf3.m), the value of ninf can be greatly reduced to reduce the calculation time and ensure the accuracy.

finda.m Find quantile (random variable value) given probability p
Parameter list: (coefficient of variation, degrees of freedom, probability value)
Call the findp function for bisection approximation. Here, first change the step size to make two initial values have opposite signs relative to probability P, and then perform bisection approximation. The efficiency of the program largely depends on the step size growth coefficient K (different coefficients of variation and degrees of freedom are more sensitive to this coefficient) and the efficiency of the findp function.

On average, it takes about 20 approximations to get a 4-digit precision quantile. The reliability of the data depends on the findp function.

print_h.m Print distribution table
Parameter list: (coefficient of variation, degrees of freedom vector, probability vector)
The program depends on the finda.m file (calculate the quantile of the specified probability) and the findp.m file (calculate the probability value of random variable a)
Program flow: Call the finda function to obtain the quantile of the specified probability and then make a table and print it.

When v is large, it is very close to the T distribution table. From the data, H(+∞, n, a) = T(n+1, a)
For example, when v = 1000, the quantile table of the H distribution with degrees of freedom 20 is the quantile table of the T distribution with degrees of freedom 21.

The program uses the inline function. It is known from searching the Internet that this function may report an error when running the m file for the first time due to version issues. It doesn't matter. Run it again after clc, and it won't report an error.

---------- DATE.TXT
>> main
/** Upper quantile table with coefficient of variation ν = 0.10 **/
─────────────────────────
n\p 0.5000 0.2500 0.0500 0.0050
─────────────────────────
2 6.8012 30.2295 211.3837 2242.6010
3 3.4702 14.9951 35.7568 35.7568
5 2.0402 9.9275 30.8079 76.2102
20 0.7967 6.4060 16.6854 26.4288
50 0.4824 5.7017 14.3829 24.3061
─────────────────────────
/** Upper quantile table with coefficient of variation ν = 1.00 **/
─────────────────────────
n\p 0.5000 0.2500 0.0500 0.0050
─────────────────────────
2 0.5094 3.0323 21.6851 229.9287
3 0.2764 1.7483 6.9846 13.5901
5 0.1671 1.2971 4.0453 9.7534
20 0.0664 0.9746 2.5435 4.4741
50 0.0404 0.9091 2.2965 3.8206
─────────────────────────
/** Upper quantile table with coefficient of variation ν = 10.00 **/
─────────────────────────
n\p 0.5000 0.2500 0.0500 0.0050
─────────────────────────
2 0.0361 1.1194 7.3685 75.4351
3 0.0223 0.8734 3.2109 11.1905
5 0.0142 0.7736 2.2639 4.9947
20 0.0059 0.7010 1.7741 2.9635
50 0.0036 0.6883 1.7046 2.7408
─────────────────────────
/** Upper quantile table with coefficient of variation ν = 1000.00 **/
────────────────────────────────────────
n\p 0.5000 0.2500 0.1000 0.0500 0.0250 0.0100 0.0050
────────────────────────────────────────
2 0.0004 1.0011 3.0820 6.3237 12.7274 31.8756 63.7682
3 0.0002 0.8170 1.8872 2.9227 4.3072 6.9727 9.9370
4 -0.0002 0.7652 1.6388 2.3550 3.1850 4.5448 5.8466
5 0.0001 0.7410 1.5340 2.1331 2.7783 3.7498 4.6078
6 0.0001 0.7269 1.4765 2.0161 2.5720 3.3671 4.0350
7 0.0001 0.7178 1.4403 1.9440 2.4481 3.1445 3.7098
8 0.0001 0.7113 1.4154 1.8953 2.3657 2.9995 3.5015
9 0.0001 0.7066 1.3973 1.8602 2.3070 2.8979 3.3572
10 0.0001 0.7029 1.3834 1.8338 2.2631 2.8227 3.2515
11 0.0001 0.7000 1.3726 1.8131 2.2290 2.7650 3.1708
12 0.0001 0.6976 1.3638 1.7965 2.2018 2.7192 3.1072
13 0.0001 0.6956 1.3566 1.7828 2.1796 2.6820 3.0559
14 0.0001 0.6940 1.3505 1.7714 2.1611 2.6513 3.0135
15 0.0001 0.6926 1.3454 1.7618 2.1455 2.6254 2.9780
16 0.0001 0.6913 1.3409 1.7535 2.1321 2.6034 2.9478
17 0.0001 0.6903 1.3371 1.7464 2.1205 2.5844 2.9218
18 0.0001 0.6903 1.3336 1.7400 2.1104 2.5678 2.8992
19 0.0001 0.6885 1.3306 1.7345 2.1015 2.5532 2.8794
20 0.0001 0.6878 1.3280 1.7295 2.0936 2.5402 2.8619
21 0.0001 0.6871 1.3256 1.7251 2.0865 2.5287 2.8463
22 0.0001 0.6865 1.3234 1.7211 2.0802 2.5184 2.8322
23 0.0001 0.6859 1.3215 1.7175 2.0744 2.5091 2.8196
24 0.0001 0.6854 1.3197 1.7143 2.0692 2.5006 2.8082
25 0.0001 0.6850 1.3181 1.7113 2.0644 2.4928 2.7977
26 0.0001 0.6845 1.3165 1.7085 2.0600 2.4857 2.7882
27 0.0001 0.6842 1.3152 1.7060 2.0560 2.4793 2.7795
28 0.0001 0.6838 1.3139 1.7036 2.0523 2.4733 2.7714
29 0.0000 0.6834 1.3128 1.7014 2.0488 2.4677 2.7640
30 0.0000 0.6831 1.3117 1.6994 2.0457 2.4626 2.7571
31 0.0000 0.6828 1.3106 1.6976 2.0427 2.4578 2.7507
32 0.0000 0.6826 1.3096 1.6958 2.0399 2.4534 2.7448
33 0.0000 0.6823 1.3088 1.6942 2.0374 2.4492 2.7392
34 0.0000 0.6821 1.3079 1.6926 2.0349 2.4453 2.7340
35 0.0000 0.6818 1.3071 1.6912 2.0326 2.4417 2.7291
36 0.0000 0.6817 1.3064 1.6898 2.0305 2.4383 2.7244
37 0.0000 0.6815 1.3057 1.6886 2.0285 2.4350 2.7201
38 0.0000 0.6812 1.3050 1.6874 2.0266 2.4320 2.7160
39 0.0000 0.6811 1.3044 1.6862 2.0247 2.4291 2.7122
40 0.0000 0.6809 1.3038 1.6851 2.0230 2.4263 2.7085
41 0.0000 0.6807 1.3032 1.6841 2.0214 2.4237 2.7050
42 0.0000 0.6806 1.3027 1.6831 2.0199 2.4213 2.7018
43 0.0000 0.6804 1.3022 1.6822 2.0185 2.4190 2.6986
44 0.0000 0.6803 1.3017 1.6814 2.0170 2.4167 2.6957
45 0.0000 0.6802 1.3012 1.6805 2.0157 2.4146 2.6928
────────────────────────────────────────
---------- FINDA.M
% Return the quantile of the specified probability.
% Parameter list: (coefficient of variation, degrees of freedom, probability value)
% Example: finda(100,5,0.025)
function result=fun(v,n,p)
h=1; % Initial step size
k=1; % Step size growth coefficient k=1, step size is linearly increasing, and when k>1, it is geometrically increasing.
yihao=0; % Sign opposite flag
if v<1 || n<3 k=3; % When v<1 and n<3, the mean value completely deviates from near 0. The appropriate step size growth coefficient will greatly reduce the number of bisection approximations!
end
a=0; % Iteration initial value
tmp=a; % tmp is used to record a
i=0;
%/*** Variable step size iteration, jump out of the loop when signs are opposite ***/
while findp(v,n,a)>p
tmp=a;
a=a+h;h=k*h;
yihao=1;
i=i+1;
end
while findp(v,n,a)<P
if yihao==1 break;
end
tmp=a;a=a-h;
h=k*h;
i=i+1;
end
b=tmp; % The new a value has opposite sign relative to P compared with tmp, pass this tmp to b, and perform bisection approximation below.
% /*** Bisection approximation ***/
while abs(a-b)>0.0001 % 4-digit precision.
c=(a+b)/2;
i=i+1;
if (findp(v,n,c)-p)*(findp(v,n,a)-p)>0
a=c;
else
b=c;
end
end
%fprintf('Number of iterations:%d\n',i)
result=(a+b)/2;
---------- FINDP.M
% Find the probability value of the random variable of the h distribution function
% Parameter list (coefficient of variation, degrees of freedom, random variable value)
% Example:
% findp(100,5,0)
function result=fun(v,n,a)
%/** xinf--Set integration region
xinf=2.5; % For the case where n>4, take 2.5 to ensure accuracy. When too large, quadl integration will fail.
if n<4 xinf=4.5;
end
if n > 280
disp('Excessive degrees of freedom will lead to overflow error')
end
if n>=2 && n<=280
syms x z ainf
gm_n=((n-1)/2)^((n-1)/2)/sqrt(pi/2)/gamma((n-1)/2);
f=x^(n-1)*exp((((n-1)*x^2+((z*x)+sqrt(n)*(x-1)/v)^2)/-2));
g=int(f,z,a,ainf); % First perform symbolic integration on z
h=subs(g,ainf,inf); % Since numerical integration will be used later, here cannot use h=limit(g,ainf,inf);
result=quadl(inline(h*gm_n),0,xinf); % Integration region is x=, z= Note that gm_n must be inside inline(h*gm_n), so that the truncation error can be reduced (otherwise the error at each level is very large!) result=double(result);
end
if n<2
disp('It is meaningless when degrees of freedom are less than 2')
end
% Remarks:
% There is a sudden change from degrees of freedom 22 to 23.
---------- MAIN.M
p=;
n=;
print_h(0.1,n,p);
print_h(1,n,p);
print_h(10,n,p);
p=;
print_h(1000,2:45,p);
print_h(10000,2:45,p);
---------- PRINT_H.M
% Print the upper quantile table of the h distribution
% Parameter list: (coefficient of variation, degrees of freedom vector, probability vector)
% Degrees of freedom cannot be 1
% Example: print_h(100,,)
function r=main(v,n,p)
fprintf('/** Upper quantile table with coefficient of variation ν=%.2f **/\n',v)
xecho(length(p))
fprintf(' n\\p')
for i=1:length(p)
fprintf('%10.4f',p(i))
end
fprintf('\n')
xecho(length(p)) % Call xecho to print the horizontal line
for i=1:length(n)
fprintf(' %2.1d ',n(i))
for j=1:length(p)
fprintf('%10.4f',finda(v,n(i),p(j))) % Call the finda function (return the quantile corresponding to p(j))
if j==length(p) fprintf('\n')
end
end
end
xecho(length(p)) % Call xecho to print the horizontal line
%/*------------xecho--------------
function xecho(length)
for i=1:length*5+5
fprintf('─')
end
fprintf('\n')
%--------------------------------*/
-----------------------------------------------------------------------------
Figure program:
---------- README,TXT
Each file is an independent program.
cdf_cont.m Output contour of 3D image of the integrand of the distribution function
cdf_mesh.m Output mesh of 3D image of the integrand of the distribution function
plot_pdf.m Output image of the probability density function
plot_pdfab.m Output image of the probability function in the specified interval
plot_cdf.m Output image of the distribution function
Help:
help

---------- CDF_CONT.M
% Output contour of 3D image of the integrand of the distribution function
% Parameter list: (coefficient of variation vector, degrees of freedom vector)
% Example:
% v=; %/*Set coefficient of variation vector
% n=2:20,50:10:100;
% cdf_cont(v,n)
function main(v,n)
%v=0.026;n=2:20;
disp('Press any key to continue...')

for i=1:length(v)
for j=1:length(n)
dgx(v(i),n(j))
pause % Automatic playback can be used pause(0.4)
end
end

function r=dgx(v,n,a,b)
syms x z
gm_n=((n-1)/2)^((n-1)/2)/sqrt(pi/2)/gamma((n-1)/2);
f=x^(n-1)*exp((((n-1)*x^2+((z*x)+sqrt(n)*(x-1)/v)^2)/-2));
fxz=gm_n*f;
ezcontour(inline(fxz))
title()

---------- CDF_MESH.M
% Output 3D image of different h integrands.
% Parameter list: (coefficient of variation vector, degrees of freedom vector)
% Example:
% v=;
% n=;
% cdf_mesh(v,n)
function main(v,n)
disp('Press any key to continue...')
for i=1:length(v)
for j=1:length(n)
cdf3(v(i),n(j))
pause % Automatic playback can be used pause(0.4)
end
end

% 2D image of the integrand
%(coefficient of variation, degrees of freedom)
function r=cdf3(v,n,a,b)
syms x z
gm_n=((n-1)/2)^((n-1)/2)/sqrt(pi/2)/gamma((n-1)/2);
f=x^(n-1)*exp((((n-1)*x^2+((z*x)+sqrt(n)*(x-1)/v)^2)/-2));
fxz=gm_n*f;
ezmesh(inline(fxz))
%meshc(inline(fxz),50)
title()

---------- PLOT_PDF.M
% Print the image of the probability function of the h distribution.
% Parameter list: (coefficient of variation vector, degrees of freedom vector)
% Example:
% v=;
% n=;
% plot_pdf(v,n)
function main(v,n)
disp('Fitting data, please wait...')
hold on;grid on;
for i=1:length(v)
for j=1:length(n)
cdf2(v(i),n(j))
%pause % Automatic playback can be used pause(0.4)
end
end

function result=cdf2(v,n)
xinf=2.5; % For the case where n>4, take 2.5 to ensure accuracy. When too large, quadl integration will fail.
if n<4 xinf=4.5;
end
if n>=2 && n<=280
syms x z k
gm_n=((n-1)/2)^((n-1)/2)/sqrt(pi/2)/gamma((n-1)/2);
f=x^(n-1)*exp((((n-1)*x^2+((z*x)+sqrt(n)*(x-1)/v)^2)/-2));
h=inline(gm_n*int(f,'x',0,xinf));
ezplot(h)
title()
end

---------- PLOT_PDFAB.M
% Print the image of the probability function of the h distribution in the specified interval.
% Parameter list: (coefficient of variation vector, degrees of freedom vector, )
% Example:
% plot_pdfab(1,5,)
function main(v,n,x)
disp('Fitting data, please wait...')
for i=1:length(v)
hold off
for j=1:length(n)
hold on;grid on;
cdf2(v(i),n(j),x(1),x(2))
clc;disp('Press any key to continue...')
pause % Automatic playback can be used pause(0.4)
end
end
function result=cdf2(v,n,x1,x2)
xinf=2.5; % For the case where n>4, take 2.5 to ensure accuracy. When too large, quadl integration will fail.
if n<4 xinf=4.5;
end
if n>=2 && n<=280
syms x z k
gm_n=((n-1)/2)^((n-1)/2)/sqrt(pi/2)/gamma((n-1)/2);
f=x^(n-1)*exp((((n-1)*x^2+((z*x)+sqrt(n)*(x-1)/v)^2)/-2));
h=inline(gm_n*int(f,'x',0,xinf));
ezplot(h,)
title()
end









[ Last edited by plp626 on 2010-10-5 at 15:30 ]

```
% Print normal distribution table
% $Author: plp626 Date: 2008-5-23 $
function f() % Main function,
p=0:9; % p from 0 to 9 represents the second decimal place of the random variable value
fprintf(' z ')
for i=1:10
fprintf('%-8.1d',p(i))
end
fprintf('\n')
for z=0:0.1:2.0 % First decimal place
fprintf('%4.1f',z)
for i=1:1:10
fprintf('%8.4f',int_f(z+p(i)/100)) % Call int_f to calculate the probability value corresponding to z+p(i)/100.
if i==10 fprintf('\n')
end
end
end
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
function r=int_f(a) % Function int_f calculates the probability (integral from negative infinity to a)
syms x
r=double(int(1/sqrt(2*pi)*exp(x^2/-2),-inf,a)); % int command calculates the integral, double command converts the calculation result to double type value and returns
```

```
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%Probability function of coefficient of variation H {plp626|19:11 2008-5-24}
%Given z, return the probability density.
%r=pdfh(v,n,z)
%v Coefficient of variation
%n Degrees of freedom
%z Random variable
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
function result=pdfh(v,n,z)
if round(n)-abs(n)~=0||n==0
disp('error: Degrees of freedom can only take positive integers!')
end
xinf=2.5; %For cases where n>4, taking 2.5 is sufficient to ensure accuracy. If too large, quadl integration will fail.
if n<4 xinf=4.5;
end
if round(n)-abs(n)==0&&n~=0
if v==0
disp('Current ν=0')
result=gamma(n./2).*(1+(z.^2)./(n-1)).^(n./-2)./gamma((n-1)./2)./sqrt((n-1).*pi);
result=double(result);
else
gm_n=((n-1)./2).^((n-1)./2)./sqrt(pi./2)./gamma((n-1)./2);
syms x
result=gm_n.*int(x.^(n-1).*exp(((n-1).*x.^2+(z.*x+sqrt(n).*(x-1)./v).^2)./-2),'x',0,xinf);
result=double(result);
end
end
```