vegan (version 2.7-6)

optspace: optspace: algorithm for matrix reconstruction from a partially revealed set

Description

This function was adapted from the original source code in the Roptspace R package (version 0.2.3; MIT License) by Raghunandan H. Keshavan, Andrea Montanari, Sewoong Oh (2010). See ROptSpace::OptSpace for more information. Let's assume an ideal matrix \(M\) with \((m\times n)\) entries with rank \(r\) and we are given a partially observed matrix \(M\_E\) which contains many missing entries. Matrix reconstruction - or completion - is the task of filling in such entries. optspace is an efficient algorithm that reconstructs \(M\) from \(|E|=O(rn)\) observed elements with relative root mean square error (RMSE) $$RMSE \le C(\alpha)\sqrt{nr/|E|}$$.

Usage

optspace(x, ropt = 3, niter = 5, tol = 1e-5, verbose = FALSE)

Value

Returns a named list containing:

X

an \((n \times r)\) matrix as left singular vectors.

S

an \((r \times r)\) matrix as singular values.

Y

an \((m \times r)\) matrix as right singular vectors.

dist

a vector containing reconstruction errors at each successive iteration.

M

an \((n \times m)\) imputed matrix, with columns and rows centered to zero.

Details

This implementation removes the trimming step of the original Roptspace::OptSpace code in order to leave feature filtering to the user. Some of the defaults have been adjusted to better reflect ecological data. The implementation has been adjusted for ecological applications as in Martino et al. (2019). The imputed matrix (M) in the optspace output includes matrix reconstruction (XSY'), with subsequent centering for the columns and rows.

References

Keshavan, R. H., Montanari, A., Oh, S. (2010). Matrix Completion From a Few Entries. IEEE Transactions on Information Theory 56:2980--2998. tools:::Rd_expr_doi("10.1109/TIT.2010.2046205").

Martino, C., Morton, J.T., Marotz, C.A., Thompson, L.R., Tripathi, A., Knight, R. & Zengler, K. (2019) A novel sparse compositional technique reveals microbial perturbations. mSystems 4, 1. tools:::Rd_expr_doi("10.1128/msystems.00016-19").

Examples

data(varespec)
# rclr transformation with no matrix completion for the 0/NA entries
x <- decostand(varespec, method = "rclr", impute = FALSE)
# Add matrix completion
xc <- optspace(x, ropt = 3, niter = 5, tol = 1e-5, verbose = FALSE)$M