DesignProxy.computeDesign() does not return a Matrix?

[mbaudin47]

What happened?

The matrix returned by DesignProxy.computeDesign() fails to be used as an input argument to QRMethod.

The script (see below for the full script):

designMatrix = design_proxy.computeDesign(functionIndices)
leastSquaresMethod = ot.QRMethod(designMatrix)  # Fails!

produces:

TypeError: Wrong number or type of arguments for overloaded function 'new_QRMethod'.

The workaround is to convert the matrix into a Matrix:

leastSquaresMethod = ot.QRMethod(ot.Matrix(designMatrix))  # Workaround

How to reproduce the issue?

# %%
import openturns as ot
from openturns.usecases import ishigami_function
import numpy as np

# %%
print(ot.__version__)

# %%
im = ishigami_function.IshigamiModel()
sampleSize = 20
inputSample = im.distributionX.getSample(sampleSize)
multivariateBasis = ot.OrthogonalProductPolynomialFactory([im.X1, im.X2, im.X3])
selectionAlgorithm = ot.PenalizedLeastSquaresAlgorithmFactory()  # Not sparse
projectionStrategy = ot.LeastSquaresStrategy(selectionAlgorithm)
basisSize = 10  # Caution: set basis size <= sample size
adaptiveStrategy = ot.FixedStrategy(multivariateBasis, basisSize)
functionIndices = ot.Indices(basisSize)
functionIndices.fill()
functionCollection = [multivariateBasis.build(i) for i in functionIndices]
transformation = ot.DistributionTransformation(
    im.distributionX, multivariateBasis.getMeasure()
)
z_train = transformation(inputSample)
basis = ot.Basis(functionCollection)
design_proxy = ot.DesignProxy(z_train, basis)
designMatrix = design_proxy.computeDesign(functionIndices)

# %%
leastSquaresMethod = ot.QRMethod(designMatrix)  # Fails!

# %%
leastSquaresMethod = ot.QRMethod(ot.Matrix(designMatrix))  # Workaround

Version

1.21

Operating System

Linux

Installation media

conda

Additional Context

No response

[sofianehaddad]

with OT 1.22 I can not reproduce