Use Ipopt

Choose Ipopt for continuous, nonlinear prescriptive problems. It is a gradient-based local solver, so it can converge to a locally optimal solution rather than a global optimum.

Capabilities

• Supports continuous nonlinear programs (NLP), including continuous quadratic programs (QP) and quadratically constrained programs (QCP) as special cases.

Limitations

• Requires a single, scalar objective.
• Does not support discrete decision variables.
• Does not support indicator constraints like implies().
• Does not support SOS constraints such as special_ordered_set_type_1() and special_ordered_set_type_2().

Example

Pass "ipopt" to Problem.solve() to use the Ipopt backend:

from relationalai.semantics import Float, Model
from relationalai.semantics.reasoners.prescriptive import Problem

m = Model("NonlinearExample")

# ... model definition ...

p = Problem(m, Float)

# ... problem definition ...

p.solve("ipopt")

Related resources:

• Solve a decision problem
• Browse prescriptive reasoning templates

Note that:

• It is common for Ipopt to terminate with LOCALLY_SOLVED. Treat that as “converged to a local minimum” rather than “certified global optimum.”

Version and licensing

Ipopt is bundled and does not require a license key or additional Snowflake setup.

Detail	Info
Version	3.14.4
License	Eclipse Public License (free and open source)

Resources

• Ipopt documentation

Where to go next

Solve a decision problemLearn the prescriptive workflow for declaring decision variables, adding requirements and an objective, solving, and reading results.

Browse prescriptive reasoning templatesExplore prescriptive templates you can adapt to your own problems.