FDTD Stability Calculator — the Courant (CFL) Bound

The finite-difference time-domain method advances an electromagnetic field one time step at a time, and the Courant–Friedrichs–Lewy condition is the bound that keeps it stable: the time step Δt must not exceed the grid step Δs divided by the propagation speed and the dimension factor. Book Three states the condition in both of its lines, Ch 3 — the 4049 line’s 3D magic step, and the b653 line’s 2D at-limit form with its generalized 3D criterion. Section 1 returns the bound itself at your dimension, grid, and speed states; Section 2 returns the Courant number of the step you state, read against the same bound.

Section 1 — The Bound: the Maximum Stable Time Step

The bound stated before the simulation is designed. On the b653 line’s generalized criterion, Δt ≤ 1/(v·√(1/Δx² + 1/Δy² + 1/Δz²)); on a uniform grid Δs the criterion reduces to Δtmax = Δs/(v·√d) at dimension d. The grid stands uniform, as documented.

mm
m/s
—

Section 2 — The Courant Number at Your Step

S = vΔt/Δs: the number your stated step carries, read against Section 1’s bound at the same states. The step this section opens on is the magic-step value Section 1’s opening states imply.

ps
—

The condition stands in Book Three in two lines, and each form’s line state is part of its identity: Line 4049, Ch 3 states the 3D magic step S = cΔt/Δs = 1/√3; Line b653, Ch 3 states the 2D at-limit form Δt = δ/(c√2) and the generalized 3D criterion derived from the dispersion relation. The Book Three census register (R5) holds the two lines aligned — dimension-indexed states of one condition, not competing values — and Book Three’s deduction (D4) carries the entry’s three measurables: the Courant number, the maximum stable Δt, and the dimension factor as computed check. The books state no numeric instance of the bound beyond the factors themselves: the 1 mm grid this page opens on is a stated opening state, not a figure of the book’s, and no default on this page reproduces a printed figure.

Further standings of the same condition stand recorded beside this entry, in other substrates, and are not this page’s ground. Book Five states the condition in Ch 37 §37.3 for explicit substrate-simulation schemes — with a worked instance, ΔtCFL ≤ Δx/vmax = 5.0 fs at Δx = 0.5 nm and vmax = 10⁷ cm/s, its arithmetic checking — and in its electromagnetic form in Ch 40 §40.2, with vmax = 1/√(εmin·μmin) and the cubic-grid simplification. Book Eleven, Ch 23 states the Courant factor λCourant = cΔt/Δx under the same 3D bound, λ ≤ 1/√3 ≈ 0.57735, in the acoustic substrate; Book Nine, Ch 42 states the criterion in the hydraulic identity, Cr = a·Δt/Δx = 1.0 at its stability state. Those standings stand in neither this entry’s source field nor its recurrence record, which names Book Three (both lines) alone; both records stand, neither subordinated, and the calculator is grounded on Book Three. The form is one identity in each book — a time step bounded by a grid step over a propagation speed, under the dimension factor. The dispersion extension of this page (the Book Three deduction’s D16, numerical dispersion and phase error) is a separate candidate and is not built here. The entry’s relation to communications is a measured correspondence, marked as such — FDTD is the standard simulation method for antennas and RF structures, and Book Three’s placement is its computational-electromagnetics span; within SolveForce’s collective work the page corresponds to the Connectivity pillar (RF design simulation) and serves no finder directly.

For the figures: select a field, then tap the keys — or type as usual.