solprop.backends.mlx_gpu.adams#

The exponential Adams history on the Metal device, for the mlx backend.

MlxAdams is an AdamsEngine: the march (AdamsMarch) keeps the controller – the orders, the slots, rejection, restarts – and this engine keeps the history’s untransported source rows on the device and forms each trial there, as HostHistory forms it on the host. The step’s flow exp(x) and its moments k! phi_{k+1}(x) are evaluated on the host in binary64 and rounded once to binary32, as the single-precision oracle rounds them; the predictor, the corrector and the defects are compiled device passes over the rows, with their Lagrange coefficients as inputs; and a trial reads its norms, the anchor’s magnitude, the trial state’s finiteness and the envelope plasma’s failures in one synchronisation. One source evaluation per trial.

It takes the exponential formulation, with or without a carrier frame. A varying operator (a gradient, a taper) adds the transported compensation of SEP 16 section 4.2: each history sample keeps its state on the device and its rate D(z_j) on the host, the state-independent factor exp((1 - theta_j) x) (D(z_j) - x / h) is formed in binary64 and rounded once, and the products with the states and the integrated Adams weights are device passes. The backend refuses the untransported witness by name.

Classes

MlxAdams(backend, method)

The exponential Adams history and its arithmetic, on the Metal device.

class solprop.backends.mlx_gpu.adams.MlxAdams(backend, method)[source]#

Bases: object

The exponential Adams history and its arithmetic, on the Metal device.

Parameters:

backendsolprop.backends.mlx_gpu.backend.MlxBackend: The backend whose state, sources, controller and synchronisation are used.
methodsolprop.stepper.multistep.Multistep: The method, in the exponential formulation.

Attributes:

denseNone: No dense observation: interpolate_planes is refused on mlx.
compensatedbool: Whether a varying operator’s samples carry the transported compensation.
carriersolprop.backends.exponential.CarrierReference or None: The carrier frame the trials remove (SEP 35), as the host engine’s.

dense: Any = None#

exponential_accept(keep, free, *, interpolate)[source]#

Adopt the held trial; its source takes row free.

Parameters:

keepnumpy.ndarray: The rows kept; nothing moves in this formulation.
freeint: The row the new sample takes.
interpolatebool: Must be false: dense observation is refused on mlx.

exponential_trial(z, dz, nodes, slots, order, orders)[source]#

Form one exponential trial on the device and return its defects’ norms.

HostHistory.exponential_trial’s arithmetic over device rows, read in one synchronisation.

Parameters:

z, dzfloat: The anchor and the step, m.
nodeslist of float: The history’s normalised positions, newest first.
slotsnumpy.ndarray: Their rows, int64.
orderint: The step’s order.
orderslist of int: The orders scored, the step’s own first.

Returns:

numpy.ndarray: One scaled norm per scored order; infinite when the trial is not finite.

fused_accept(keep, free, *, interpolate)[source]#

Refuse the interaction formulation, which this engine does not run.

Parameters:

keepnumpy.ndarray: The rows kept.
freeint: The free row.
interpolatebool: Whether the interpolant is kept.

Raises:

BackendError: Always.

fused_trial(*args, **kwargs)[source]#

Refuse the interaction formulation, which this engine does not run.

Parameters:

*args, **kwargsobject: The march’s trial arguments.

Raises:

BackendError: Always.

hold()[source]#

Return the accepted state, the device array itself.

Returns:

object: The state.

load_rows(rows, shape)[source]#

Install rows a snapshot holds, narrowed to complex64 on the device.

Parameters:

rowscollections.abc.Mapping: The rows, by name.
shapetuple of int: The state’s shape.

Raises:

BackendError: If the rows are not a store of this history’s shape, or hold any other name (a compensated or coupled history’s).

property n_slots: int[source]#

The history’s rows: the method’s order and two more.

Returns:

int: order + 2.

observe(position)[source]#

Refuse dense observation, which mlx does not run.

Parameters:

positionfloat: The plane, m.

Raises:

BackendError: Always.

reset()[source]#

Forget any trial in flight.

restore(z, held)[source]#

Make a held state current again at z.

Parameters:

zfloat: Its position, m.
heldobject: What hold() returned.

snapshot_nbytes()[source]#

Return the bytes snapshot_rows() would copy.

Returns:

int: The rows’ size, complex64.

snapshot_rows()[source]#

Return host copies of the history’s rows, in complex64.

Returns:

dict: "store": one row per slot, zero where the history has none yet.

start(z, *, stored)[source]#

Evaluate the source at the accepted state into row zero.

Parameters:

zfloat: The accepted position, m.
storedbool: Must be true: the general path runs on the host and is refused.

Raises:

BackendError: For an unstored sample, which only the general path asks for.

value(slot)[source]#

Return nothing: the history’s rows stay on the device.

Parameters:

slotint: The row.