Number

A gamspy.Number() object is needed mainly for two cases.

The first case is to assign records to a symbol conditionally. For example:

import gamspy as gp

m = gp.Container()
k = gp.Set(m, "k", records=["1964-i","1964-ii","1964-iii","1964-iv"])
ki = gp.Set(m, domain=k, description="initial period")
ki[k] = gp.Number(1).where[gp.Ord(k) == 1]

The code snippet above would assign only 1964-i to ki since only the order of 1964-i is equal to 1.

The second case is the modeling of constant right-hand side or left-hand side. Since Python does not provide magic functions for __req__, __rle__ and __rge__ similar to __radd__, it is impossible for GAMSPy to distinguish, for example, 0 <= expression and expression >= 0. For example:

import gamspy as gp

gp.set_options({"ALLOW_AMBIGUOUS_EQUATIONS": "yes"})

m = gp.Container()
c = gp.Parameter(m, name="c")
x = gp.Variable(m, name="x", type="Negative")
e = gp.Equation(m, name="e")
e[...] = (x - c) >= 0
print(e.latexRepr())

f = gp.Equation(m, name="f")
f[...] = 0 <= (x - c)
print(f.latexRepr())

print(e.latexRepr() == f.latexRepr())

Because of the aformentioned limitation of Python, the equation definition of e and f are the same:

x - c \geq 0
x - c \geq 0
True

Due to this you might get marginals with a wrong sign. This implication is especially important for EMP, MCP and MPEC models since the GAMSPy execution system checks appropriate bounds for equation/variable pairs. The option ALLOW_AMBIGUOUS_EQUATIONS controls how GAMSPy reacts to ambiguous equation definitions (when the gamspy.Model.solve() method is called). The default for this option is AUTO. For EMP, MCP and MPEC models AUTO acts like setting the option to NO, for other model types AUTO behaves like setting this option to YES. Instead of figuring out the best option setting, a much better solution is to use gamspy.Number() to ensure that the order of the components are correct:

import gamspy as gp

m = gp.Container()
c = gp.Parameter(m, name="c")
x = gp.Variable(m, name="x", type="Negative")
e = gp.Equation(m, name="e")
e[...] = (x - c) >= gp.Number(0)
print(e.latexRepr())

f = gp.Equation(m, name="f")
f[...] = gp.Number(0) <= (x - c)
print(f.latexRepr())

print(e.latexRepr() == f.latexRepr())

With the usage of gamspy.Number(), the equation components are in order again:

x - c \geq 0
0 \leq x - c
False