topic Re: Problem with a macro in SAS Programming

Problem with a macro

Num19 — Wed, 10 Jan 2018 21:06:43 GMT

Hi, I made a program which put the x and y coordonates when a condition is verified but I have an error generated in SAS,

here some informations :

%macro mandelbrot(x_min,x_max,y_min,y_max,pas,seuil,n_max);
%let pas_coef=&pas.*100;
%let x_min_coef=(&x_min.*100);
%let x_max_coef=(&x_max.*100);
%let y_min_coef=(&y_min.*100);
%let y_max_coef=(&y_max.*100);
%let u=0;
%let v=0;
         %do i=&x_min_coef. %to &x_max_coef. by &pas_coef.;
             %do j=&y_min_coef. %to &y_max_coef. by &pas_coef.;
                  %let n=0;
                  %let u=0;
                  %let v=0;
                  %let &norm=0;
                  %let x_r=%sysevalf(&i./100);
                  %let y_r=%sysevalf(&j./100);
                  %do %while(&norm.<&seuil. and &n.<&nmax.);
                       %let n=%eval(1+&n.);
                       %let u=%sysevalf((&u.)**2-(&v.)**2+&x_r.);
                       %let v=%sysevalf(2*(&u.)*(&v.)+&y_r.);
                       %let &norm=%module_2(&u.,&v.);
				  %end;
                  
       	 %if (&norm.<&seuil.) %then %do;
			proc sql;
			insert into MANDELBROT values (&x_r.,&y_r.);
			quit;
	    %end;
	   %end;
%end;
%mend mandelbrot;
%mandelbrot(-2,0.5,-1.1,1.1,0.02,2,30);

The error :

ERROR: A character operand was found in the %EVAL function or %IF condition where a numeric operand is required. The condition
was: &x_max_coef. by &pas_coef.
ERROR: The %TO value of the %DO I loop is invalid.
ERROR: The macro MANDELBROT will stop executing.

Re: Problem with a macro

novinosrin — Wed, 10 Jan 2018 21:11:50 GMT

should be %by instead of by coz you are using macro language needing macro syntax for the macro processor to recognize

 %do i=&x_min_coef. %to &x_max_coef. %by &pas_coef.;
             %do j=&y_min_coef. %to &y_max_coef. %by &pas_coef.;

Re: Problem with a macro

Reeza — Wed, 10 Jan 2018 21:11:51 GMT

A %DO loop likely wants a %BY rather than BY.

@Num19 wrote:

Hi, I made a program which put the x and y coordonates when a condition is verified but I have an error generated in SAS,

here some informations :

%macro mandelbrot(x_min,x_max,y_min,y_max,pas,seuil,n_max);
%let pas_coef=&pas.*100;
%let x_min_coef=(&x_min.*100);
%let x_max_coef=(&x_max.*100);
%let y_min_coef=(&y_min.*100);
%let y_max_coef=(&y_max.*100);
%let u=0;
%let v=0;
         %do i=&x_min_coef. %to &x_max_coef. by &pas_coef.;
             %do j=&y_min_coef. %to &y_max_coef. by &pas_coef.;
                  %let n=0;
                  %let u=0;
                  %let v=0;
                  %let &norm=0;
                  %let x_r=%sysevalf(&i./100);
                  %let y_r=%sysevalf(&j./100);
                  %do %while(&norm.<&seuil. and &n.<&nmax.);
                       %let n=%eval(1+&n.);
                       %let u=%sysevalf((&u.)**2-(&v.)**2+&x_r.);
                       %let v=%sysevalf(2*(&u.)*(&v.)+&y_r.);
                       %let &norm=%module_2(&u.,&v.);
				  %end;
                  
       	 %if (&norm.<&seuil.) %then %do;
			proc sql;
			insert into MANDELBROT values (&x_r.,&y_r.);
			quit;
	    %end;
	   %end;
%end;
%mend mandelbrot;
%mandelbrot(-2,0.5,-1.1,1.1,0.02,2,30);

The error :

Re: Problem with a macro

Num19 — Wed, 10 Jan 2018 21:14:07 GMT

Oh, sorry for that mistake.

Now I have :

ERROR: A character operand was found in the %EVAL function or %IF condition where a numeric operand is required. The condition
was: &x_max_coef.
ERROR: The %TO value of the %DO I loop is invalid.
ERROR: A character operand was found in the %EVAL function or %IF condition where a numeric operand is required. The condition
was: &pas_coef.
ERROR: The %BY value of the %DO I loop is invalid.
ERROR: The macro MANDELBROT will stop executing.

Re: Problem with a macro

novinosrin — Wed, 10 Jan 2018 21:17:23 GMT

sorry my bad that i didn't notice at the top, you need %eval in your %let assignments

or to be safe %sysevalf

%let pas_coef=&pas.*100;
%let x_min_coef=(&x_min.*100);
%let x_max_coef=(&x_max.*100);
%let y_min_coef=(&y_min.*100);
%let y_max_coef=(&y_max.*100);

Re: Problem with a macro

Num19 — Wed, 10 Jan 2018 21:19:31 GMT

Error on error I have this one now :

ERROR: A character operand was found in the %EVAL function or %IF condition where a numeric operand is required. The condition
was: 0.02*100
ERROR: The macro MANDELBROT will stop executing.
😞

Re: Problem with a macro

Reeza — Wed, 10 Jan 2018 21:20:16 GMT

1. Replace BY with %BY

2. Set parameters using %SYSEVALF()

3. Fix Norm value declaration (not &norm, remove &).

4. NMAX is undeclared/not calculated???

%macro mandelbrot(x_min,x_max,y_min,y_max,pas,seuil,n_max);
%let pas_coef=%sysevalf(&pas.*100);
%let x_min_coef=%sysevalf(&x_min.*100);
%let x_max_coef=%sysevalf(&x_max.*100);
%let y_min_coef=%sysevalf(&y_min.*100);
%let y_max_coef=%sysevalf(&y_max.*100);
%let u=0;
%let v=0;
         %do i=&x_min_coef. %to &x_max_coef. %by &pas_coef.;
             %do j=&y_min_coef. %to &y_max_coef. %by &pas_coef.;
                  %let n=0;
                  %let u=0;
                  %let v=0;
                  %let norm=0;
                  %let x_r=%sysevalf(&i./100);
                  %let y_r=%sysevalf(&j./100);
                  %do %while(&norm.<&seuil. and &n.<&nmax.);
                       %let n=%eval(1+&n.);
                       %let u=%sysevalf((&u.)**2-(&v.)**2+&x_r.);
                       %let v=%sysevalf(2*(&u.)*(&v.)+&y_r.);
                       %let &norm=%module_2(&u.,&v.);
				  %end;
                  
       	 %if (&norm.<&seuil.) %then %do;
			proc sql;
			insert into MANDELBROT values (&x_r.,&y_r.);
			quit;
	    %end;
	   %end;
%end;
%mend mandelbrot;
%mandelbrot(-2,0.5,-1.1,1.1,0.02,2,30);

A little bit closer....it seems like this would be easier/faster in a data step or PROC IML.

It says NMAX isn't declared, which is true.

Re: Problem with a macro

novinosrin — Wed, 10 Jan 2018 21:20:21 GMT

Mate!, use %sysevalf instead of %eval

Re: Problem with a macro

Num19 — Wed, 10 Jan 2018 21:23:26 GMT

I think that's so close, I have this now :

ERROR: Expecting a variable name after %LET.
ERROR: Symbolic variable name 0 must begin with a letter or underscore.
ERROR: The macro MANDELBROT will stop executing.

Re: Problem with a macro

novinosrin — Wed, 10 Jan 2018 21:24:41 GMT

problem is this

  %let norm=0; causing vairable name issue hahahahha

Re: Problem with a macro

Reeza — Wed, 10 Jan 2018 21:25:57 GMT

And n_max in the macro list versus nmax used in the code.....

Re: Problem with a macro

novinosrin — Wed, 10 Jan 2018 21:26:14 GMT

that is causing error in macro var creation here

    %let &norm=%module_2(&u.,&v.);

all same names begin with a letter or underscore

Re: Problem with a macro

Num19 — Wed, 10 Jan 2018 21:36:39 GMT

No error displayed but I have nothing displayed by the macro

Re: Problem with a macro

ballardw — Wed, 10 Jan 2018 21:44:13 GMT

@novinosrin wrote:

sorry my bad that i didn't notice at the top, you need %eval in your %let assignments

or to be safe %sysevalf

%let pas_coef=&pas.*100;
%let x_min_coef=(&x_min.*100);
%let x_max_coef=(&x_max.*100);
%let y_min_coef=(&y_min.*100);
%let y_max_coef=(&y_max.*100);

AND

%eval only does integers.

See

2740  %let junk= %eval(0.5 *100);
ERROR: A character operand was found in the %EVAL function or %IF condition where a numeric
       operand is required. The condition was: 0.5 *100

Yes you would need:

%let junk= %sysevalf(0.5 *100);

HOWEVER since you are using those values in %do loops ALL the values must be integers on the %do statements. The macro language doesn't work with %do x= .5 %to 10 %by .5;

Of course the question is why are you doing data processing in the macro language to begin with???

Especially to do an INSERT into a set using Proc SQL? I don't know what your %module_2 macro does but I'm willing to be this can all be turned into a data step relatively easily.

Re: Problem with a macro

Reeza — Wed, 10 Jan 2018 21:44:39 GMT

The macro doesn't display anything, it inserts value into your table. Check your table.

Re: Problem with a macro

Num19 — Wed, 10 Jan 2018 21:51:26 GMT

Sorry I meant when I check my table I have a lot of -2, 0 -1, 1 but I don't have decimal numbers.

Re: Problem with a macro

ballardw — Wed, 10 Jan 2018 22:06:22 GMT

@Num19 wrote:

Sorry I meant when I check my table I have a lot of -2, 0 -1, 1 but I don't have decimal numbers.

Since you only insert a couple of variables into your table you won't have much info for diagnostics.

Likely to generate a lot of junk but try running

Options mprint symbolgen;

before the macro. I could show where your values aren't being kept or such.

Also what does the %module_2 macro actually look like? That might be the cause since it is setting the value of one of your variables. If it sets a large value for norm then you may not have many times when norm< seuil at the test for insertion.

Re: Problem with a macro

WarrenKuhfeld — Wed, 10 Jan 2018 22:12:12 GMT

I completely agree with @ballardw . You are making this an order of magnitude harder than it needs to be by using the macro language instead of a DATA step.

Re: Problem with a macro

Num19 — Wed, 10 Jan 2018 22:20:08 GMT

Here is module's code :

%macro module_2(x_1,y_2);
%global norme_2;
%let xcarre=(&x_1**2);
%let ycarre=(&y_2**2);
%let normecarre=%sysevalf(&xcarre+&ycarre);
%let norme_2=%sysfunc(sqrt(&normecarre));
%mend module_2;

%module_2(2,3);

Re: Problem with a macro

novinosrin — Wed, 10 Jan 2018 22:40:28 GMT

missing %sysevalf here mate!

%let xcarre=(&x_1**2);
%let ycarre=(&y_2**2);