Unable to use allow_free option in Opti::to_function in Python

[S-Dafarra]

I have the following snippet of code

import casadi as cs

opti = cs.Opti()
v = opti.variable(1)
p = opti.parameter(1)
opti.minimize(v)
opti.solver("ipopt")
opti.to_function("test",[v,p], [v,p], {"allow_free": True})

but it fails with the error

Traceback (most recent call last):
  File "C:\Software\mambaforge\envs\hippopt_dev\Lib\site-packages\IPython\core\interactiveshell.py", line 3577, in run_code
    exec(code_obj, self.user_global_ns, self.user_ns)
  File "<ipython-input-19-fbf0b281ebea>", line 1, in <module>
    opti.to_function("test",[v,p], [v,p], {"allow_free": True})
  File "C:\Software\mambaforge\envs\hippopt_dev\Lib\site-packages\casadi\casadi.py", line 48679, in to_function
    return _casadi.Opti_to_function(self, *args)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
RuntimeError: Error in Opti::to_function [OptiNode] at D:\bld\casadi_1709758764158\work\casadi\core\optistack.cpp:331:
Error in Opti::to_function [OptiNode] at D:\bld\casadi_1709758764158\work\casadi\core\optistack.cpp:331:
Error in Function::Function for 'helper' [MXFunction] at D:\bld\casadi_1709758764158\work\casadi\core\function.cpp:280:
D:\bld\casadi_1709758764158\work\casadi\core\function_internal.cpp:147: Error calling MXFunction::init for 'helper':
D:\bld\casadi_1709758764158\work\casadi\core\mx_function.cpp:409: helper::init: Initialization failed since variables [opti0_p_1] are free. These symbols occur in the output expressions but you forgot to declare these as inputs. Set option 'allow_free' to allow free variables.

I am using version 3.6.5.

[ernestds]

You are not using p in the problem so it is sort of not registered in Opti.

Ex:

opti.minimize(v + p)
opti.solver("ipopt")
opti.to_function("test",[v,p], [v,p],)

or

opti.minimize(v)
opti.solver("ipopt")
opti.to_function("test",[v,p], [v],)

both work.

[S-Dafarra]

Thanks @ernestds, but I thought that was what the allow_free option is meant for.

[ernestds]

Tbh I think that isn't possible with Opti to_function, but I'm not sure.

import casadi as cs

opti = cs.Opti()
v = opti.variable(1)
p = opti.parameter(1)
opti.minimize(v**2+p)
opti.solver("ipopt")
f = opti.to_function("test",[v], [v,p], {"allow_free": True,})
# f = cs.Function("test",[v], [v+p,p], {"allow_free": True})
cs.Function('bla',[v,p],[*f(v)])(1,2)

This raises no error but still doesn't work for the optimization. You might want to play around with this or try the nlpsol function instead (I have no idea whether that will be any better).