Output files¶

In QuanEstimation, the output data will be saved into files during or after the optimization process. There are two categories: the values of the objective function at each step, and the optimized variables from the corresponding scheme. This guide describes all output file types and how to load them.

Objective function values¶

During optimization, the objective function is evaluated at each step and the values are saved sequentially (one value per line) into f.csv.

Python

import numpy as np
f = np.loadtxt("f.csv")

Control optimization¶

The control optimization results are saved into controls.npy. The shape depends on savefile:

savefile=False (default): the final control coefficients are saved, with shape (n_ctrl, n_tseg) — the first dimension is the number of control Hamiltonians, the second dimension is the number of time segments.
savefile=True: all control coefficients after each round of optimization are saved with shape (n_rounds, n_ctrl, n_tseg).

Python

import numpy as np
controls = np.load("controls.npy")

If the optimized controls are exported as CSV (Julia side), use csv2npy_controls to convert:

from quanestimation import csv2npy_controls
csv2npy_controls(controls_csv_data, num=n_tseg)

This reshapes the CSV array and saves as controls.npy.

State optimization¶

The state optimization results are saved into states.npy.

savefile=False (default): the final optimized state vectors are saved.
savefile=True: all state vectors after each round are saved.

Python

import numpy as np
states = np.load("states.npy")

Use csv2npy_states to convert CSV-based state files:

from quanestimation import csv2npy_states
csv2npy_states(states_csv_data, num=1)

Measurement optimization¶

The measurement optimization results are saved into measurements.npy.

savefile=False (default): the final optimized POVMs are saved.
savefile=True: all POVM lists after each round are saved.

Python

import numpy as np
M = np.load("measurements.npy")

If measurements are exported as CSV via writedlm, use csv2npy_measurements to convert:

from quanestimation import csv2npy_measurements
csv2npy_measurements(M_csv_data, num=n_operators)

Comprehensive optimization¶

Comprehensive optimization combines multiple variable types. The output follows the same conventions as above:

Controls are saved in controls.npy
States are saved in states.npy
Measurements are saved in measurements.npy

Each file is present only if that variable type was part of the optimization. For example, SC optimization produces controls.npy and states.npy, but not measurements.npy.

Bayesian estimation¶

The Bayes() estimator produces the following output files:

File	Contents	Dimensions
`pout.npy`	Posterior probability distributions	`(n_iter, n_grid_points)`
`xout.npy`	Estimated parameter values	`(n_iter,)` or `(n_iter, n_params)`
`Lout.npy`	Likelihood functions (when `continue=True`)	`(n_iter, n_grid_points)`

Python

import numpy as np
pout = np.load("pout.npy")
xout = np.load("xout.npy")

Adaptive measurement schemes¶

Adaptive estimation produces additional files via the Adapt() module:

File	Contents
`pout.csv`	Posterior distributions at each adaptive step
`xout.csv`	Estimated values at each adaptive step
`y.csv`	Experimental data at each step
`f.csv`	Objective function across iterations

Python

import numpy as np
pout = np.loadtxt("pout.csv")
xout = np.loadtxt("xout.csv")
y = np.loadtxt("y.csv")