LIM Component-Separation 3D Dataset (CII + CO)

This compressed NumPy archive contains the full set of simulated line-intensity-mapping cubes for run 1 of the dataset. It includes, for each sky tile: the [CII] intensity cube, generated using the de Looze [CII] prescription; the CO foreground cube, containing the sum of all modeled CO rotational transitions included in the simulation; the corresponding instrumental noise cube. The cubes were generated from the SIDES–UCHUU pipeline and are stored in units of: MJy/sr\mathrm{MJy/sr}MJy/sr All cubes have been cropped to a common shape: (181, 723, 719) corresponding to: 181 frequency channels 723 pixels along the first spatial axis 719 pixels along the second spatial axis The frequency axis is: 125, 126, ..., 305 GHz with a spacing of 1 GHz. The archive contains data for 13 × 9 = 117 tiles in the run. The file stores: Metadata freq_ghz — frequency grid in GHz tile1_values — first tile index values tile2_values — second tile index values run_id — simulation run identifier units — data units, "MJy/sr" cube_shape — common cube shape, (181, 723, 719) Cube arrays For each tile (tile1, tile2), the archive contains: cii_<tile1>_<tile2> — [CII] cube co_<tile1>_<tile2> — CO cube noise_<tile1>_<tile2> — instrumental noise cube For example, tile (0, 0) is stored as: cii_0_0co_0_0noise_0_0 The corresponding mixed observed cube can be reconstructed as: total=[CII]+CO+noise How to load the file import numpy as nppath = "1_all_tiles_CO_CII_noise_cubes.npz"data = np.load(path) To inspect all stored keys: print(data.files) To read the metadata: freqs = data["freq_ghz"]tile1_values = data["tile1_values"]tile2_values = data["tile2_values"]run_id = data["run_id"]units = data["units"]cube_shape = data["cube_shape"]print(freqs)print(run_id)print(units)print(cube_shape) To load the components of one specific tile, for example tile (0, 0): cii = data["cii_0_0"]co = data["co_0_0"]noise = data["noise_0_0"] Each array has shape: print(cii.shape)# (181, 723, 719) To construct the corresponding mixed cube: total = cii + co + noise Memory note The .npz archive is large. Loading the archive with: data = np.load(path) does not immediately load every cube into memory, but each individual array is decompressed when accessed. It is therefore recommended to load and process one tile at a time rather than reading all cubes simultaneously.