bartorch.optim.iterators.PRIDUIteration

bartorch.optim.iterators.PRIDUIteration#

class bartorch.optim.iterators.PRIDUIteration(terms, image_shape, primal=None, **kwargs)#

BART’s primal-dual iteration, one step of it.

italgos.c’s chambolle_pock, which pics --pridu runs. The data term is carried as its own dual variable rather than differentiated: A^H u is updated through the resolvent

A^H u <- (sigma A^H A x_avg + A^H u - sigma A^H y) / (1 + sigma)

and each regularization term gets a dual of its own, updated through its proximal operator’s conjugate. The primal step is a descent on the duals followed by prox2, and x_avg extrapolates.

pics takes sigma = sqrt(step) * ratio and tau = sqrt(step) / ratio with theta = 1, and hogwild there is a decay of 0.95 a step rather than a halving.

The first term stands apart, as it does in iter2_chambolle_pock: a term whose transform is the identity becomes the primal prox2 and the rest become duals. Without such a term prox2 is the identity, which is what prox_zero_create is.

Notes

This is the one iteration here whose answer depends on how BART was compiled. vecops.c has a single kernel behind axpy, xpay and axpbz, dst[i] = a1 * src1[i] + a2 * src2[i], and clang folds the first product into the add where the hardware has a fused multiply-add – arm64 does, the x86-64 baseline does not. A fused multiply-add does not round the product, and torch cannot fuse across two kernels.

Every other iteration escapes it because its updates are axpy, whose a1 is one: folding an exact product in changes nothing. The data term’s resolvent here is an xpay and an axpbz with two real coefficients, so on a platform that folds them this iteration is within a few times 1e-7 of the library rather than the same bits.

__init__(terms, image_shape, primal=None, **kwargs)#

Initialize internal Module state, shared by both nn.Module and ScriptModule.

Methods

__init__(terms, image_shape[, primal])

Initialize internal Module state, shared by both nn.Module and ScriptModule.

add_module(name, module)

Add a child module to the current module.

apply(fn)

Apply fn recursively to every submodule (as returned by .children()) as well as self.

bfloat16()

Casts all floating point parameters and buffers to bfloat16 datatype.

buffers([recurse])

Return an iterator over module buffers.

children()

Return an iterator over immediate children modules.

compile(*args, **kwargs)

Compile this Module's forward using torch.compile().

cpu()

Move all model parameters and buffers to the CPU.

cuda([device])

Move all model parameters and buffers to the GPU.

double()

Casts all floating point parameters and buffers to double datatype.

eval()

Set the module in evaluation mode.

extra_repr()

Return the extra representation of the module.

float()

Casts all floating point parameters and buffers to float datatype.

forward(X, cur_data_fidelity, cur_prior, ...)

General form of a single iteration of splitting algorithms for minimizing \(F = f + \lambda \regname\), alternating between a step on \(f\) and a step on \(\regname\).

get_buffer(target)

Return the buffer given by target if it exists, otherwise throw an error.

get_extra_state()

Return any extra state to include in the module's state_dict.

get_parameter(target)

Return the parameter given by target if it exists, otherwise throw an error.

get_submodule(target)

Return the submodule given by target if it exists, otherwise throw an error.

half()

Casts all floating point parameters and buffers to half datatype.

ipu([device])

Move all model parameters and buffers to the IPU.

load_state_dict(state_dict[, strict, assign])

Copy parameters and buffers from state_dict into this module and its descendants.

modules([remove_duplicate])

Return an iterator over all modules in the network.

mtia([device])

Move all model parameters and buffers to the MTIA.

named_buffers([prefix, recurse, ...])

Return an iterator over module buffers, yielding both the name of the buffer as well as the buffer itself.

named_children()

Return an iterator over immediate children modules, yielding both the name of the module as well as the module itself.

named_modules([memo, prefix, remove_duplicate])

Return an iterator over all modules in the network, yielding both the name of the module as well as the module itself.

named_parameters([prefix, recurse, ...])

Return an iterator over module parameters, yielding both the name of the parameter as well as the parameter itself.

parameters([recurse])

Return an iterator over module parameters.

register_backward_hook(hook)

Register a backward hook on the module.

register_buffer(name, tensor[, persistent])

Add a buffer to the module.

register_forward_hook(hook, *[, prepend, ...])

Register a forward hook on the module.

register_forward_pre_hook(hook, *[, ...])

Register a forward pre-hook on the module.

register_full_backward_hook(hook[, prepend])

Register a backward hook on the module.

register_full_backward_pre_hook(hook[, prepend])

Register a backward pre-hook on the module.

register_load_state_dict_post_hook(hook)

Register a post-hook to be run after module's load_state_dict() is called.

register_load_state_dict_pre_hook(hook)

Register a pre-hook to be run before module's load_state_dict() is called.

register_module(name, module)

Alias for add_module().

register_parameter(name, param)

Add a parameter to the module.

register_state_dict_post_hook(hook)

Register a post-hook for the state_dict() method.

register_state_dict_pre_hook(hook)

Register a pre-hook for the state_dict() method.

relaxation_step(u, v, beta)

Performs a relaxation step of the form \(\beta u + (1-\beta) v\).

requires_grad_([requires_grad])

Change if autograd should record operations on parameters in this module.

set_extra_state(state)

Set extra state contained in the loaded state_dict.

set_submodule(target, module[, strict])

Set the submodule given by target if it exists, otherwise throw an error.

share_memory()

See torch.Tensor.share_memory_().

state_dict(*args[, destination, prefix, ...])

Return a dictionary containing references to the whole state of the module.

to(*args, **kwargs)

Move and/or cast the parameters and buffers.

to_empty(*, device[, recurse])

Move the parameters and buffers to the specified device without copying storage.

train([mode])

Set the module in training mode.

type(dst_type)

Casts all parameters and buffers to dst_type.

xpu([device])

Move all model parameters and buffers to the XPU.

zero_grad([set_to_none])

Reset gradients of all model parameters.

Attributes

T_destination

call_super_init

dump_patches

training