Estimate the Memory of a Dense Cost Matrix

What building the matrix costs a process: the 8 bytes of each cell, written in place by the distance code, and little besides. Building the matrix and stopping there has peaked at between 1.0 and 1.1 times the raw cell bytes from 2,000 to 20,000 units on the memory benchmark, so the default multiplier is above what has been measured rather than fitted to it. n/m are coerced to double before multiplying so the estimate itself can't overflow.

Usage

estimate_dense_matrix_mb(n, m, overhead_factor = 2)

Arguments

n, m

Problem dimensions.

overhead_factor

Multiplier on the raw cell bytes.

Value

Numeric scalar, the estimated footprint of the matrix in megabytes.

Details

This is the matrix, not the solve. estimate_dense_solve_mb() is what the memory guard reads; see there for why the two differ by more than a copy.

Examples

estimate_dense_matrix_mb(5000, 5000)