Function lbcrypto::Cholesky(const Matrix<int32_t>&)

Function Documentation

Matrix<double> lbcrypto::Cholesky(const Matrix<int32_t> &input)

Gives the Choleshky decomposition of the input matrix. The assumption is that covariance matrix does not have large coefficients because it is formed by discrete gaussians e and s; this implies int32_t can be used This algorithm can be further improved - see the Darmstadt paper section 4.4 http://eprint.iacr.org/2013/297.pdf

Parameters

&input – the matrix for which the Cholesky decomposition is to be computed

Returns

the resulting matrix of floating-point numbers