sporco_cuda.cbpdn module¶

cbpdn(D, s, lmbda, opt, dev=0)¶

A GPU-accelerated version of sporco.admm.cbpdn.ConvBPDN.

Parameters:

Darray_like(float32, ndim=3): Dictionary array (three dimensional)
sarray_like(float32, ndim=2): Signal array (two dimensional)
lmbdafloat32: Regularisation parameter
optdict or sporco.admm.cbpdn.ConvBPDN.Options object: Algorithm options (see Algorithm Options)
devint: Device number of GPU device to use

Returns:

Xndarray: Coefficient map array (sparse representation)

cbpdnmsk(D, s, w, lmbda, opt, dev=0)¶

A GPU-accelerated version of sporco.admm.cbpdn.AddMaskSim used together with sporco.admm.cbpdn.ConvBPDN, providing a spatial mask in the data fidelity term of the functional minimized by this class. Since the spatial mask is implemented via the Additive Mask Simulation (AMS) method (see doi:10.1109/ICIP.2016.7532675 and the documentation for sporco.admm.cbpdn.AddMaskSim), the entries must be in \(\{0,1\}\). Note that this GPU version differs from the Python code in its handling of the L1Weight option: this version automatically adjusts this array to account for the AMS impulse filter that is inserted into the dictionary, while the Python version requires this to be handled by the calling function. In addition, this version prepends the AMS impulse filter at the start of the dictionary, while the Python version appends it at the end.

Parameters:

Darray_like(float32, ndim=3): Dictionary array (three dimensional)
sarray_like(float32, ndim=2): Signal array (two dimensional)
warray_like: Mask array (two dimensional)
lmbdafloat32: Regularisation parameter
optdict or sporco.admm.cbpdn.ConvBPDN.Options object: Algorithm options (see Algorithm Options)
devint: Device number of GPU device to use

Returns:

Xndarray: Coefficient map array (sparse representation)

cbpdngrd(D, s, lmbda, mu, opt, dev=0)¶

A GPU-accelerated version of sporco.admm.cbpdn.ConvBPDNGradReg.

Parameters:

Darray_like(float32, ndim=3): Dictionary array (three dimensional)
sarray_like(float32, ndim=2): Signal array (two dimensional)
lmbdafloat32: Regularisation parameter (l1)
mufloat: Regularisation parameter (gradient)
optdict or sporco.admm.cbpdn.ConvBPDNGradReg.Options object: Algorithm options (see Algorithm Options)
devint: Device number of GPU device to use

Returns:

Xndarray: Coefficient map array (sparse representation)

cbpdngrdmsk(D, s, w, lmbda, mu, opt, dev=0)¶

A GPU-accelerated version of of sporco.admm.cbpdn.AddMaskSim used together with sporco.admm.cbpdn.ConvBPDNGradReg, providing a spatial mask in the data fidelity term of the functional minimized by this class. Since the spatial mask is implemented via the Additive Mask Simulation (AMS) method (see doi:10.1109/ICIP.2016.7532675 and the documentation for sporco.admm.cbpdn.AddMaskSim), the entries must be in \(\{0,1\}\). Note that this GPU version differs from the Python code in its handling of the L1Weight and GradWeight options: this version automatically adjusts these arrays to account for the AMS impulse filter that is inserted into the dictionary, while the Python version requires this to be handled by the calling function. In addition, this version prepends the AMS impulse filter at the start of the dictionary, while the Python version appends it at the end.

Parameters:

Darray_like(float32, ndim=3): Dictionary array (three dimensional)
sarray_like(float32, ndim=2): Signal array (two dimensional)
warray_like: Mask array (two dimensional)
lmbdafloat32: Regularisation parameter (l1)
mufloat: Regularisation parameter (gradient)
optdict or sporco.admm.cbpdn.ConvBPDNGradReg.Options object: Algorithm options (see Algorithm Options)
devint: Device number of GPU device to use

Returns:

Xndarray: Coefficient map array (sparse representation)

Algorithm Options¶

The algorithm options parameter may either be an appropriate sporco options object (sporco.admm.cbpdn.ConvBPDN.Options or sporco.admm.cbpdn.ConvBPDNGradReg.Options), or a dict with the following entries:

Verbose : Flag determining whether iteration status is displayed.

MaxMainIter : Maximum main iterations.

AbsStopTol : Absolute convergence tolerance (see the docs for sporco.admm.admm.ADMM.Options).

RelStopTol : Relative convergence tolerance (see the docs for sporco.admm.admm.ADMM.Options).

RelaxParam : Relaxation parameter (see the docs for sporco.admm.admm.ADMM.Options). Relaxation is disabled by setting this value to 1.0.

rho : ADMM penalty parameter \(\rho\).

AutoRho : Options for adaptive rho strategy. The value of this dict key should itself be a dict with the following entries (see the docs for sporco.admm.admm.ADMM.Options for more detail):

Enabled : Flag determining whether adaptive penalty parameter strategy is enabled.

Period : Iteration period on which rho is updated. If set to 1, the rho update test is applied at every iteration.

Scaling : Multiplier applied to rho when updated.

RsdlRatio : Primal/dual residual ratio in rho update test.

RsdlTarget : Residual ratio targeted by auto rho update policy.

AutoScaling : Flag determining whether RhoScaling value is adaptively determined. If enabled, Scaling specifies a maximum a llowed multiplier instead of a fixed multiplier.

StdResiduals : Flag determining whether standard residual definitions are used instead of normalised residuals.

AuxVarObj : Flag indicating whether the objective function should be evaluated using variable X (False) or Y (True) as its argument. Setting this flag to True often gives a better estimate of the objective function, but at additional computational cost.

HighMemSolve : Flag indicating whether to use a slightly faster algorithm at the expense of higher memory usage.

NonNegCoef : Flag indicating whether to force solution to be non-negative.

NoBndryCross : Flag indicating whether all solution coefficients corresponding to filters crossing the image boundary should be forced to zero.

L1Weight : An array of weights for the \(\ell_1\) norm (see the docs for sporco.admm.cbpdn.GenericConvBPDN.Options for more detail).

GradWeight : An array of weights \(w_m\) for the term penalising the gradient of the coefficient maps (see the docs for sporco.admm.cbpdn.ConvBPDNGradReg.Options for more detail). NB: This option is only relevant to cbpdngrd() and cbpdngrdmsk().

Entries in the sporco options objects that are not listed above are silently ignored.