input_output.py

from pathlib import Path
from typing import Tuple

import numpy as np
from scipy import sparse


def read_input(dataset: str) -> Tuple[np.ndarray, np.ndarray, np.ndarray, np.ndarray, np.ndarray, np.ndarray]:
    def load(filename: str) -> np.ndarray:
        return np.loadtxt(Path('input') / dataset / filename, delimiter=',')

    l = load('rays.dat').T
    Sll = np.array([[
        [line[0], line[3], line[5]],
        [line[3], line[1], line[4]],
        [line[5], line[4], line[2]],
    ] for line in load('covariances.dat')])
    ijt = load('linkage.dat').astype(int) - 1  # NOTE: from base-1 to base-0
    Xa = load('points.dat').T
    Ma = load('motions.dat').reshape(-1, 4, 4)
    P = load('projections.dat').reshape(-1, 3, 4)

    return l, Sll, ijt, Xa, Ma, P


def write_output(dataset: str,
                 ld: np.ndarray,
                 Xd: np.ndarray,
                 Md: np.ndarray,
                 Sdd: sparse.spmatrix,
                 vr: np.ndarray,
                 w: np.ndarray) -> None:
    def save(filename: str, X, fmt: str = '%.6f') -> None:
        np.savetxt(Path('results') / dataset / filename, X, delimiter=',', fmt=fmt)

    save('rays.dat', ld.T)
    save('points.dat', Xd.T)
    save('motions.dat', Md.reshape(-1, 4))
    save('motioncovariance.dat', Sdd.toarray(), '%.12f')
    save('corrections.dat', vr.T, '%.12f')
    save('weights.dat', w)