Hello,

I am exploring gams for a modelling task in irrigation systems. I attempted the attached model of least square fit of the form:

y= a*x^2 + b*x +c to a small data set. where i know that a= 1, b= 0 and c=0.

the model runs well and gives expected results when using variables a, b, c as follows:

LOWER LEVEL UPPER MARGINAL

---- VAR a -INF 1.0000 +INF EPS

---- VAR b -INF 5.551115E-17 +INF EPS

---- VAR c -INF -6.24500E-16 +INF EPS

---- VAR sse -INF . +INF .

however, it gives unexpected results when using variables coef(j) defined over sets j /1*3/ as in:

y= coef(j)*x^2 + coef(j)*x + coef(j)

---- VAR coef

LOWER LEVEL UPPER MARGINAL

a -INF 0.7246 +INF EPS

b -INF 0.7246 +INF EPS

c -INF 0.7246 +INF EPS

what am i missing? I appreciate quick help on this, and thanks.

testFit_LInear.log (2.24 KB)

testFit_LInear.gms (599 Bytes)