pypty.iterative

run

Launch an iterative ptychographic reconstruction.

PARAMETER	DESCRIPTION
pypty_params	Reconstruction parameters dictionary or path to a saved parameters file. Must include experiment setup and algorithm settings (see custom_presets.md and reference/initialize.md). TYPE: dict or str

Notes

• The function mutates global state, including obj, probe, and history_bfgs.
• Parameters can be dynamically varied via lambda functions in pypty_params.
• Checkpoints and logs are saved to output_folder as specified in the parameters.

bfgs_update

x Internal function of PyPTy. Perform a single L-BFGS update step for all active reconstruction parameters.

PARAMETER	DESCRIPTION
Way

RETURNS	DESCRIPTION
total_loss	Total objective loss for this epoch. TYPE: float
this_sse	Sum of squared errors for this epoch. TYPE: float
constraint_contributions	Regularization term contributions. TYPE: list of float
actual_step	Final step length chosen by line search. TYPE: float
count	Number of line search iterations performed. TYPE: int
d_value	Directional derivative before update. TYPE: float
new_d_value	Directional derivative after update. TYPE: float
warnings	Warnings generated during update. TYPE: str

Notes

• This function updates global variables: obj, probe, positions, etc.
• Handles both first-time and history-based L-BFGS updates.

reset_bfgs_history

Reset the global BFGS history and free GPU memory.

Notes

• Clears all stored vectors and scalars in history_bfgs.
• Attempts to free memory pools for CuPy and pinned memory.