Add cp, cf, x, y arrays in the available functions for optimization

[lucaerosBGD]

Description of feature

Add the following slice data in the available function for optimization :
# "cp_visc_upper": viscous CP on the airfoil's upper surface
# "cp_invisc_upper": inviscid CP on the airfoil's upper surface
# "x_upper": x coordinates of the upper surface CP data
# "y_upper": y coordinates of the upper surface CP data
# "cp_visc_lower": viscous CP on the airfoil's lower surface
# "cp_invisc_lower": inviscid CP on the airfoil's lower surface
# "x_lower": x coordinates of the lower surface CP data
# "y_lower": y coordinates of the lower surface CP data
# "cf_upper": skin friction coefficient on the upper surface
# "x_cf_upper": x coordinates of upper surface skin friction coefficient
# "y_cf_upper": y coordinates of upper surface skin friction coefficient
# "cf_lower": skin friction coefficient on the lower surface
# "x_cf_lower": x coordinates of lower surface skin friction coefficient
# "y_cf_lower": y coordinates of lower surface skin friction coefficient
Thus we could optimize and constraint parameters available in this list (max pressure gradients for example)

Potential solution

[A-CGray]

Hey @lucaerosBGD , this shouldn't be too hard to add, you just need to return the values from the real/complex slice data dictionaries in the same way we already do from the real/complex funcs dictionaries,

Here's where we do that in the function evaluation:

CMPLXFOIL/cmplxfoil/CMPLXFOIL.py

Lines 606 to 614 in c1701d9

	for f in evalFuncs:
	# If it's not in the list of available functions
	if f not in self.functionList:
	# Either throw an error (if requested) or skip it
	if not ignoreMissing:
	raise ValueError(f'Supplied function "{f}" is not in the available functions {self.functionList}.')
	else:
	returnFuncs[aeroProblem.name + "_" + f] = self.funcs[aeroProblem.name][f]
	funcs.update(returnFuncs)

And here's where we do that in the derivative calculation:

CMPLXFOIL/cmplxfoil/CMPLXFOIL.py

Lines 763 to 771 in c1701d9

	self.__call__(self.curAP, useComplex=mode == "CS", deriv=True)
	# Compute the Jacobian vector products
	jacVecProd = {}
	for f in self.functionList:
	if mode == "FD":
	jacVecProd[f] = (self.funcs[self.curAP.name][f] - orig_funcs[f]) / h
	else:
	jacVecProd[f] = np.imag(self.funcsComplex[self.curAP.name][f]) / np.imag(h)

[lucaeros]

Hey @A-CGray, @eytanadler , I am trying to implement that feature. This is being done in my fork under cost_func branch (https://github.com/lucaeros/CMPLXFOIL/tree/cost_func)
I added data from sliceData dictionnary to funcs.
I changed evalFunctionsSens to enable jacobian matrix (because some outputs are vector, not scalar)
I am trying to run single_point.py with "cp_visc" in evalFuncs to see this is working. CMPLXFOIL derivative are working. However, I got issue from Multipoint package :
File "/home/lucaeros/packages/multipoint/multipoint/multiPointSparse.py", line 744, in sens gcon[oKey][dvSet] += np.dot(deriv, np.atleast_2d(funcSens[iKey][dvSet][i, :])) ValueError: non-broadcastable output operand with shape (1,1) doesn't match the broadcast shape (1,322)

Is it possible to have vector in evalFuncs or only scalars ?