Table 1. ^a. Cardiac myosin flux through in vivo active cycle in 3 phases.

phase	f1	f2	f3	f4	f5	f6	f7	f8	f9	f10
Unloaded	39±7	0	5±8	56±9	43±11	13±9	13±9	0	0	18±12
Auxotonic	10±5	2±1	31±23	59±23	37±21	29±11	21±9	3±4	13±11	52±22
Near-isometric	2±1	1±1	80±5	18±5	12±3	15±4	6±2	42±5	52±4	86±3
notes	input & output	≤f1	input	input	output f4-f7	≤f5	5→3 f4-f5	≤f10	f2+f6+f8	output f3+f7

^a Flux quantities, f_i, for the 4 step-size network defined in the Fig 8 model. Several flux values relate to known step-frequencies using Eq 5 and where f₁ = x₈, f₄ = x₅, f₉ = x₀, and f₁₀ = x₃. Other fluxes are surmised by using constraints. Flux conservation equality constraints include (total input) f₁+f₃+f₄ = f₁+f₅+f₁₀ (total output), f₄ = f₅+f₇ (5 nm step input detaches or continues to 3 nm step), f₃+f₇ = f₁₀ (3 nm step input sums with 5/3 nm step conversion then detaches with a 3 nm step), and f₂+f₆ −f₇+f₈ = f₉ (total 0 length steps). The problem is under determined by equality constraints hence it is solved in two steps: first using equality constraints eliminating 4 parameters, second using 3 equality constraints relating f₂, f₃, f₅, f₆, f₇, and f₈ and inequality constraints for these variables. Inequality constraints are f₂ ≤ x₈, f₂ ≤ x₀, f₃ ≥ x₃-x₅, f₃ ≤ x₃, f₅ ≤ x_5, f₆ ≤ x₅, f₇ ≤ f_6, f₇ ≤ x_5, f₈ ≤ x₃, f₈ ≤ x₀, and all unknowns ≥ 0. The latter equality and inequality constraints are sufficiently restrictive to identify convergent solutions for the fluxes for each phase using constrained linear programing in Mathematica. We estimate standard deviations for fluxes within all constraints by generating random variates using normal distributions for {ω₀, ω₃, ω₅, ω₈} in Table 2, computing x₀, x₃, x₅, and x₈ using Eq 5, then solving for the unknown flux values. Flux errors are standard deviation for (n) replicates. Total input (or equivalently total output) is re-normalized to 100% post hoc facilitating comparison between phases.