Closed rileyjmurray closed 1 year ago
Hi pyGSTI developers. This is just a note that the following use of CVXPY can be made more efficient.
https://github.com/pyGSTio/pyGSTi/blob/dd4ad669931c7f75e026456470cf33ac5b682d0d/pygsti/tools/optools.py#L326-L348
Specifically, the constraints sig0 == sig0.T, sig1 == sig1.T can be removed if the lines
sig0 == sig0.T, sig1 == sig1.T
sig0 = _cvxpy.Variable(shape=(smallDim, smallDim)) # rho0.real sig1 = _cvxpy.Variable(shape=(smallDim, smallDim)) # rho1.real
are changed to
sig0 = _cvxpy.Variable(shape=(smallDim, smallDim), symmetric=True) # rho0.real sig1 = _cvxpy.Variable(shape=(smallDim, smallDim), symmetric=True) # rho1.real
Using full matrices for the antisymmetric variables tau0 and tau1 is also inefficient, but that doesn't have a one-line fix.
tau0
tau1
Reopening until the next release.
Closed with release of 0.9.12.
Hi pyGSTI developers. This is just a note that the following use of CVXPY can be made more efficient.
https://github.com/pyGSTio/pyGSTi/blob/dd4ad669931c7f75e026456470cf33ac5b682d0d/pygsti/tools/optools.py#L326-L348
Specifically, the constraints
sig0 == sig0.T, sig1 == sig1.T
can be removed if the linesare changed to
Using full matrices for the antisymmetric variables
tau0
andtau1
is also inefficient, but that doesn't have a one-line fix.