Density Functional Theory

Imagine you want to predict whether a drug molecule will bind to a protein, or whether a new alloy will be stronger than steel. Both questions ultimately reduce to quantum mechanics: the electrons in every atom obey the Schrödinger equation, and the answer is written in their collective wavefunction.

The catch is brutal. A system of $N$ electrons needs a wavefunction $\Psi(\mathbf{r}_1, \mathbf{r}_2, \ldots, \mathbf{r}_N)$ — a function of $3N$ continuous coordinates. For a modest fifty-electron molecule, storing that function even at coarse resolution is more data than all computers on Earth can hold. The many-body problem is not merely hard; it is exponentially hard.

In 1964, Pierre Hohenberg and Walter Kohn published a theorem that changed everything. The ground-state energy of any quantum system, they proved, is a functional of the electron density $n(\mathbf{r})$ — a function of just three coordinates, regardless of how many electrons the system contains. The entire quantum information is encoded in that single three-dimensional cloud. One year later, Kohn and Lu Jeu Sham showed how to use that theorem in practice through a set of single-particle equations. Walter Kohn received the Nobel Prize in Chemistry in 1998 for this work.

Density Functional Theory (DFT) is the result: a method that converts the impossible many-body Schrödinger equation into a tractable set of independent-electron equations, each living in an effective potential shaped by all the others.

The demo below solves the Kohn-Sham equations in one dimension for a simple external potential. Each step of the self-consistency loop updates the effective potential from the current electron density, recomputes the lowest orbitals, and rebuilds the density — repeating until nothing changes.

<!-- {{c_html_intro}} -->
<div class="controls">
  <label>{{lbl_potential}} <select id="potSel">
    <option value="harmonic">{{opt_harmonic}}</option>
    <option value="double">{{opt_double}}</option>
    <option value="box">{{opt_box}}</option>
  </select></label>
  <label>{{lbl_electrons}} <input id="nElec" type="number" min="1" max="6" value="2" style="width:3rem"></label>
  <button id="runBtn" type="button">{{btn_run}}</button>
  <button id="resetBtn" type="button" class="ghost">{{btn_reset}}</button>
</div>
<canvas id="canvas" width="560" height="280" aria-label="{{canvas_aria}}"></canvas>
<div id="status" class="status">{{status_ready}}</div>
<div class="legend">
  <span class="leg-item"><span class="leg-line pot"></span> {{leg_potential}}</span>
  <span class="leg-item"><span class="leg-line dens"></span> {{leg_density}}</span>
  <span class="leg-item"><span class="leg-line orb"></span> {{leg_orbitals}}</span>
</div>

* { box-sizing: border-box; }
body { font-family: system-ui, sans-serif; color: #222; margin: 0; background: #fff; }
.controls { display: flex; flex-wrap: wrap; gap: .5rem; align-items: center; margin-bottom: .6rem; }
label { font-size: .88rem; display: flex; align-items: center; gap: .3rem; }
select, input[type=number] { font: inherit; border: 1px solid #bbb; border-radius: 6px; padding: .2rem .4rem; }
button { font: 600 14px system-ui, sans-serif; padding: .4rem .85rem; border: 1px solid #1d3557;
         background: #1d3557; color: #fff; border-radius: 8px; cursor: pointer; }
button.ghost { background: #fff; color: #1d3557; }
canvas { display: block; border: 1px solid #dde3ea; border-radius: 10px; max-width: 100%; }
.status { font-size: .92rem; font-weight: 600; min-height: 1.4em; margin: .4rem 0; color: #555; }
.status.ok { color: #0a7d33; }
.status.err { color: #c92f3c; }
.legend { display: flex; gap: 1.2rem; flex-wrap: wrap; font-size: .82rem; margin-top: .25rem; }
.leg-item { display: flex; align-items: center; gap: .35rem; }
.leg-line { display: inline-block; width: 24px; height: 3px; border-radius: 2px; }
.leg-line.pot { background: #e63946; }
.leg-line.dens { background: #1d7fd4; }
.leg-line.orb { background: #2d9c6f; }

// Code not found

Watch how the electron density (solid curve) converges as the loop iterates. The density is simply the sum of the squares of the occupied orbitals: $n(x) = \sum_{i} |\phi_i(x)|^2$. This is the quantity that — according to the Hohenberg-Kohn theorem — contains all the information needed to determine the ground-state energy, without ever touching a many-body wavefunction.

DFT is celebrated for what it avoids, but the story has a sting in its tail.

• Exponential to cubic. The many-body Schrödinger equation scales as $O(e^{N})$ in the worst case. Kohn-Sham DFT scales as $O(N^{3})$ in standard implementations (diagonalizing an $N \times N$ matrix each iteration), making thousands-of-atom simulations routine. Linear-scaling DFT variants exist for large systems.
• The exchange-correlation problem. The Hohenberg-Kohn theorem guarantees the exact functional $E_{xc}[n]$ exists, but no one knows what it is. Every practical DFT calculation uses an approximation: Local Density Approximation (LDA), Generalized Gradient Approximation (GGA), hybrid functionals mixing in exact exchange, and so on. This is the central open problem of the field — sometimes called climbing Jacob's ladder of approximations.
• v-representability. Not every density $n(\mathbf{r})$ can be produced by some external potential $v(\mathbf{r})$ — a subtle mathematical gap between the theorem and its use. Most practical densities are fine, but the gap matters in formal developments.
• Strongly correlated systems. When electron-electron interactions dominate — Mott insulators, high-temperature superconductors, heavy-fermion materials — standard DFT approximations fail badly. Capturing strong correlation is an active research frontier connecting DFT to quantum computing.

The result is a method that is exact in principle and approximate in practice, with accuracy that is often remarkable and occasionally wrong in important ways.

DFT is the most widely used method in computational science today — the workhorse behind:

• Drug and protein design: computing binding energies and reaction barriers for candidate molecules before synthesizing them in the lab. DFT screens millions of compounds in silico, cutting the cost of early-stage drug discovery.
• Materials science: predicting crystal structures, band gaps, and mechanical properties of alloys, ceramics, and two-dimensional materials like graphene and MoS₂.
• Catalysis: understanding how molecules adsorb on metal surfaces and which sites lower activation barriers — essential for designing better industrial catalysts and electrocatalysts for green hydrogen production.
• Battery and solar cell research: calculating ion migration barriers in electrolytes, and the electronic structure of photovoltaic absorbers to optimize efficiency.
• Semiconductor physics: modeling dopant behavior, defect levels, and interface properties in silicon, III-V semiconductors, and emerging two-dimensional materials for next-generation transistors.

The connection to computational complexity is not accidental: DFT is effective precisely because it trades an exponentially hard problem for a polynomial one — at the cost of an approximation whose quality must always be validated.

Density Functional Theory is one of the great intellectual shortcuts of modern science. By swapping an exponentially large wavefunction for a three-dimensional electron density, it made the quantum mechanics of real materials computable — and earned Walter Kohn a Nobel Prize.

The catch is honest: the exact exchange-correlation functional is unknown, and the approximations we use work brilliantly in many regimes and fail quietly in others. The field advances by climbing Jacob's ladder, developing better functionals, and connecting DFT to machine learning and quantum computing for the hardest correlated systems.

Every time a chemist computes a reaction barrier, or an engineer predicts a band gap, they are betting on that foundational theorem: the ground-state energy is a functional of the density, and the density alone is enough.