how to use absolute function in proc model to solve nonlinear equations

[tonghui1127]

how to use absolute function in proc model to solve nonlinear equations. 

[Kurt_Bremser]

---

@tonghui1127 wrote:

how to use absolute function in proc model to solve nonlinear equations. 

---

Like you use it in a data step, it is the same function.

[tonghui1127]

Dear KurtBremser 

Thank you for your rely.

This is my SAS code, when i use proc model to solve pie1 , I think there has some values are negative.  So I want to use ABS() function in my formula to make sure there is no negative values. But this is not work. 

proc model data = data2 out = data3;
eq.f = (p*t*pie1+(1-p)*(1-t)*(1-pie1)-x1/&NumTrials)
+(p*(1-t)*pie1+(1-p)*t*(1-pie1)-x2/&NumTrials)
+((1-p)*t*pie1+p*(1-t)*(1-pie1)-x3/&NumTrials)
+((1-p)*(1-t)*pie1+p*t*(1-pie1)-x4/&NumTrials);
solve pie1;

proc print data = data3;
run;

This is the result i got after run code, the pie1 is .0001 ,  i don't know how to fix this problem.

PREDICT	SIMULATE	0	.0001	0.7	0.7	77	115	86	222
PREDICT	SIMULATE	0	.0001	0.7	0.7	58	104	117	221
PREDICT	SIMULATE	0	.0001	0.7	0.7	58	112	111	219
PREDICT	SIMULATE	0	.0001	0.7	0.7	74	110	109	207
PREDICT	SIMULATE	0	.0001	0.7	0.7	68	93	125	214
PREDICT	SIMULATE	0	.0001	0.7	0.7	62	110	95	233
PREDICT	SIMULATE	0	.0001	0.7	0.7	64	104	113	219
PREDICT	SIMULATE	0	.0001	0.7	0.7	57	103	94	246
PREDICT	SIMULATE	0	.0001	0.7	0.7	70	106	114	210

[Kurt_Bremser]

If you want to prevent negative values of pie1 in your solution, either set them to zero by using the max() function, eg

max(pie1,0)

instead of just pie1, or use a where condition on the dataset to discard such observations.