Nonlinear Operation Codes

Nonlinear expressions are built as a tree of operations, each with an associated operation code (opcode). Each opcode has both a name and an integer value. Our programming language APIs provide predefined constants for all of these, which we recommend you use. These constants are accessed by prepending a fixed prefix (which varies slightly by language) to the opcode name. To access the SQRT operation code (to define the square root function) from our various APIs, for example, you would use the following:

Language	Operation Code
C	GRB_OPCODE_SQRT
C++	GRB_OPCODE_SQRT
Java	GRB.OPCODE_SQRT
.NET	GRB.OPCODE_SQRT
Python [1]	GRB.OPCODE_SQRT

[1]

Note that in Python, we recommend you use arithmetic operators and our nonlinear function helpers to build nonlinear constraints (see NLExpr for details). Using these operation codes directly is considered advanced usage.

The following table lists the available operation codes:

Opcode name	Value	Arity	Data
CONSTANT	0	n/a	value of constant
VARIABLE	1	n/a	index of variable (or variable object)
PLUS	2	n-ary	\(-1.0\)
MINUS	3	binary	\(-1.0\)
MULTIPLY	4	n-ary	\(-1.0\)
DIVIDE	5	binary	\(-1.0\)
UMINUS	6	unary	\(-1.0\)
SQUARE	7	unary	\(-1.0\)
SQRT	8	unary	\(-1.0\)
SIN	9	unary	\(-1.0\)
COS	10	unary	\(-1.0\)
TAN	11	unary	\(-1.0\)
POW	12	binary	\(-1.0\)
EXP	13	unary	\(-1.0\)
LOG	14	unary	\(-1.0\)
LOG2	15	unary	\(-1.0\)
LOG10	16	unary	\(-1.0\)
LOGISTIC	17	unary	\(-1.0\)

These operation codes will also appear in MPS and LP files (using the names from the table above).

See our section Operation Codes for a detailed description of each operation code.