{smcl}
{bf:Mata course:  Translating from FORTRAN, example 2}
{hline}

{title:The problem}

{p 8 12 2}
1.
Calculate spearman rank correlation 

{p 8 12 2}
2.  
Calculate significance using Statalib routine 89


{title:Step 1:  Obtain FORTRAN code}

{p 8 12 2}
1.
I visited Statlib site, found routine, and copied it
to {bf:{view prho.f}}.


{title:Step 2:  Translate to Mata}

{p 8 12 2}
1.  Copy {bf:prho.f} to {bf:prho.do}

{p 8 12 2}
2.  Edit to create Mata program -- see {bf:{view prho1.do}}.

{p 12 16 2}
a.  Included {cmd:mata set matastrict on}


{title:Step 3:  Run do-file (compile), see what happens}


	{txt}
	{com}. do prho1
	{txt}
	{com}. clear
	{txt}
	{com}. 
	. mata:
	{txt}{hline 29} mata (type {cmd:end} to exit) {hline}
	{com}: 
	: mata set matastrict on

	: 
	: 
	: real scalar prho(real scalar n, real scalar is, real scalar ifault)
	> {c (-}
	> {it:(output omitted)}
	> {c )-}
	{err}variable js undeclared
	variable nfac undeclared
	variable i undeclared
	variable ifr undeclared
	variable m undeclared
	variable ise undeclared
	variable n1 undeclared
	variable mt undeclared
	variable nn undeclared
	(2 lines skipped)
	{txt}{hline}
	{search r(3000):r(3000);}

	end of do-file
	{search r(3000):r(3000);}{txt}

{p 4 4 2}
Comments:

{p 8 12 2}
1.  I checked that variables were not typographical errors.

{p 8 12 2}
2.  I declared them:

		{cmd:real scalar js, nfac, i, ifr, m, ise, n1, mt, nn}

{p 8 12 2}
3.  I compiled again.


	{com}. do prho2
	{txt}
	{com}. clear
	{txt}
	{com}. 
	. mata:
	{txt}{hline 29} mata (type {cmd:end} to exit) {hline}
	{com}: 
	: mata set matastrict on

	: 
	: 
	: real scalar prho(real scalar n, real scalar is, real scalar ifault)
	> {c -(}
	> {it:(output omitted)}
	> {c )-}
	{txt}note: variable z unused

	{com}: 
	: end
	{txt}{hline}

{p 8 12 2}
3.  I checked that variable {cmd:z} really was not used, then 
    removed its declaration, and compiled again.

	{com}. do prho3
	{txt}
	{com}. clear
	{txt}
	{com}. 
	. mata:
	{txt}{hline 29} mata (type {cmd:end} to exit) {hline}
	{com}: 
	: mata set matastrict on

	: 
	: 
	: real scalar prho(real scalar n, real scalar is, real scalar ifault)
	> {c -(}
	> {it:(output omitted)}
	> {c )-}

	{com}: 
	: end
	{txt}{hline}

{p 4 4 2}
Notice, there are no messages.


{title:Step 4:  Make prho() easier to test}

{p 8 12 2}
1.
Rename {cmd:rpho()} and sl89_prho()}

{p 8 12 2}
2.
Add new routine {cmd:scalar_prho()}

	    {cmd}real scalar scalar_prho(real scalar n, real scalar r)
	    {
        	    real scalar     ifault
        	    real scalar     res

        	    res = sl89_prho(n, (n^3-n)*(1-r)/6, ifault)
        	    if (ifault) res = .
        	    return(res)
	    }{txt}

{p 8 12 2}
3.  Result in {bf:{view prho4.do}}


{title:Step 5:  Tested result interactively}

{p 8 12 2}
1.  But just a little.


{title:Step 6.  Make an ado-file}

{p 8 12 2}
1.  See {bf:{view prho.ado}}

{p 8 12 2}
2.  Try it with 

		. {cmd:clear}
		. {cmd:set seed 29393}
		. {cmd:set obs 100}
		. {cmd:gen u1 = uniform()}
		. {cmd:gen u2 = uniform()}
		. {cmd:prho u1 u2}

{p 12 12 2}
Result:

		{com}. prho u1 u2

		{txt} Number of obs = {res}    100
		{txt}Spearman's rho = {res}     -0.0345

		{txt}Test of Ho: u1 and u2 are independent
		    Prob > |t| = {res}      0.7335
		{txt}      New test = {res}      0.7330{txt}


{hline}
{bf:{view talk.smcl:Top}}