DATA FOR THE FIGURES OF
  Title:   Discrete nonlinear-spectrum editing with a Fourier-domain transfer function
  Authors: Minji Shi, Mingming Tan, Auro M. Perego

This archive contains the numerical values plotted in Figs. 2-8 of the
article, one CSV file per panel. Fig. 1 is a schematic and contains no data.

FILE FORMAT
  UTF-8, comma-separated, one header row, no index column.
  Numbers are written with 10 significant digits.
  An empty field means that no point is plotted at that abscissa
  (Fig. 7: offsets at which the number of eigenvalues of the output
  differs from that of the target).

UNITS AND NORMALISATION
  Unless stated otherwise, all quantities are dimensionless, in the
  normalised units of the nonlinear Schroedinger equation used in the article:
  time t in units of T0, ordinary frequency omega/(2*pi) in units of 1/T0,
  field q in units of sqrt(P), with P = |beta2|/(gamma*T0^2).
  log10_* columns contain base-10 logarithms of the named quantity.
  Columns ending in _over_GHz contain log10 of a physical frequency divided
  by 1 GHz, evaluated for T0 = 3 ps.

MAIN PARAMETERS
  T0 = 3 ps; beta2 = -21.7 ps^2/km; gamma = 1.3 /W/km
  EDFA field gain g = 5; minimum SLM field transmission T_min = 0.02;
  dark-frequency threshold epsilon = 1e-4
  Simulation grid: 4096 points, t in [-30, 30]
  Fig. 6: K = 100 noise realisations per noise level

ORIGIN OF THE DATA
  All data are obtained numerically. The eigenvalues of the input waveforms
  (Figs. 2d-5d, "input") are the prescribed values used to synthesise the
  waveforms; all other eigenvalues are computed by forward NFT.

==========================================================================
FILE                     FIGURE / PANEL     COLUMNS
==========================================================================
Fig2a.csv  .. Fig5a.csv  Figs. 2-5 (a)      Ordinary Fourier spectral intensities
    omega_over_2pi              normalised ordinary frequency, omega/(2*pi) [1/T0]
    spectral_intensity_in       spectral intensity of the input
    spectral_intensity_tar      spectral intensity of the target
    spectral_intensity_out      spectral intensity of the output
  The data cover a wider frequency range than shown in the figures.

Fig2b.csv  .. Fig5b.csv  Figs. 2-5 (b)      Effective transfer function
    omega_over_2pi              normalised ordinary frequency [1/T0]
    H_squared_dB                |H(omega)|^2 in dB (left axis)
    arg_H_rad                   arg H(omega) in rad (right axis)
  Between the frequency points of the design, |H| and the phase are
  interpolated for display.

Fig2c.csv  .. Fig5c.csv  Figs. 2-5 (c)      Temporal intensities
    t                           normalised time [T0]
    intensity_in                |q_in(t)|^2
    intensity_tar               |q_tar(t)|^2
    intensity_out               |q_out(t)|^2
  The data cover a wider time range than shown in the figures.

Fig2d_input.csv .. Fig5d_input.csv    Figs. 2-5 (d)   Eigenvalues of q_in (circles)
Fig2d_output.csv .. Fig5d_output.csv  Figs. 2-5 (d)   Eigenvalues of q_out (crosses)
    Re_lambda, Im_lambda        real and imaginary parts of the eigenvalues

Fig6a.csv  (injection, eigenvalues)         Additive input noise
Fig6b.csv  (injection, norming constants)
Fig6c.csv  (removal, eigenvalues)
Fig6d.csv  (removal, norming constants)
    EsN0_dB                     input energy to noise spectral density ratio [dB]
    log10_R_lambda_* / log10_R_b_*
                                log10 of the relative RMSE of the eigenvalues /
                                norming constants, for the transfer-function
                                method and the coupling method

Fig7a.csv  (injection, eigenvalues)         Carrier-frequency offset
Fig7b.csv  (removal, eigenvalues)
Fig7c.csv  (injection, norming constants)
Fig7d.csv  (removal, norming constants)
    log10_abs_delta_omega       log10 |Delta omega| (normalised angular frequency)
    log10_abs_delta_f_over_GHz  log10(|Delta f|/GHz) for T0 = 3 ps (top axis)
    log10_E_lambda_* / log10_E_b_*
                                log10 of the relative eigenvalue / norming-constant
                                error (larger of the values for +|Delta omega| and
                                -|Delta omega|), transfer-function and coupling methods

Fig8a.csv  (injection)                      Finite spectral resolution
Fig8b.csv  (removal)
    log10_delta                 log10 of the normalised resolution delta [1/T0]
    log10_delta_f_over_GHz      log10(delta_f/GHz) for T0 = 3 ps (top axis)
    log10_E_lambda, log10_E_b   log10 of the relative eigenvalue / norming-constant error
Fig8a_ideal_mask.csv, Fig8b_ideal_mask.csv  Horizontal reference lines (red)
    log10_E_lambda_ideal_mask, log10_E_b_ideal_mask
                                errors for the ideal mask (delta = 0)
