The strong force does the opposite of gravity or electromagnetism: it gets weaker at short distance / high energy, and gets stronger without bound as quarks are pulled apart. This is asymptotic freedom (Gross, Politzer, Wilczek — Nobel 2004), computed here from the real QCD one-loop beta function:
β0 = 11 − (2/3) n_f
α_s(Q) = 4π / [ β0 · ln(Q² / Λ²) ] (Λ_QCD ≈ 0.21 GeV)
As the probe energy Q rises, ln(Q²/Λ²) grows, so α_s(Q) falls — quarks inside a fast-moving proton barely feel each other, exactly as seen in deep-inelastic electron-proton scattering at SLAC and HERA. As Q drops toward Λ_QCD the formula diverges (a "Landau pole") — perturbation theory breaks down and the true non-perturbative physics is confinement: quarks are bound by a flux tube whose potential energy grows linearly with separation,
V(r) = − (4/3) α_s ħc / r + σ·r (σ ≈ 0.9 GeV/fm, string tension)
The resolved distance shown is r = ħc/Q with ħc = 0.1973 GeV·fm — the de Broglie wavelength of the virtual photon or gluon doing the probing. High Q ⇒ short r ⇒ you are "looking" deep inside the proton, where quarks act almost like free particles.
- Q slider — probe energy in GeV, log-spaced from 0.35 to 500 GeV.
- nf slider — how many quark flavors are light enough to contribute virtual loops at this scale; changing it visibly steepens or flattens the running of α_s.
- Auto-sweep — animates Q from low to high so you can watch the flux tube relax from a tight, turbulent, rigid bundle (confined, large α_s) into a loose cloud of nearly independent quarks (asymptotically free, small α_s).
- Drag the main view — the three quarks live at fixed 3D coordinates (a stylized uud baryon); dragging rotates that 3D arrangement and a manual perspective projection redraws it on the 2D canvas every frame, exactly like orbiting a 3D camera.
The three colored disks are a stylized proton (uud) projected from 3D onto the 2D canvas; disk depth-shading, tube brightness/thickness and gluon-spark turbulence all scale directly with the computed α_s(Q) — this is an illustrative rendering of the mechanism, not a lattice-QCD wavefunction.
Note: the one-loop formula above is evaluated exactly as published (verified numerically against the known α_s(M_Z)≈0.118 benchmark, giving ≈0.135 with this illustrative Λ_QCD — the small offset is expected at one-loop order with a simplified Λ, not a bug).