1. Localized orbitals and generalized eigenproblems
Position in the course: Lesson 1 of 10. Complete the preceding derivation and use the explained exercises to check understanding.
1. Purpose and assumptions
Tight binding replaces a continuous orbital problem by a selected set of localized functions. Those functions can resemble atomic orbitals, be empirically defined, or arise from a projection of first-principles states. The basis determines which states the model can describe. A one-orbital model does not secretly contain all the missing atomic orbitals; its parameters summarize their effects only within a intended energy and geometry range.
Expand an electronic state in the chosen functions and project the Hamiltonian. If the basis functions overlap, their metric is S rather than the identity. Physical normalization is c†Sc=1. Varying the Rayleigh quotient with a constraint gives Hc=ESc, a generalized eigenproblem. Solving Hc=Ec while retaining a nonidentity S changes the model. The overlap matrix should be Hermitian positive definite when the basis is linearly independent.
For an orthogonal two-site model with equal onsite energy ε and real hopping t, the symmetric and antisymmetric combinations diagonalize the Hamiltonian. Their energies are ε+t and ε−t. Which combination is lower depends on the sign convention for t and orbital phases. A phase change on one basis orbital flips its hopping sign while leaving the physical eigenvalues unchanged. Do not infer an observable sign from a matrix entry without defining the gauge.
The projected Rayleigh principle gives an upper-bound property for a true fixed Hamiltonian in a finite variational basis. An empirically fitted Hamiltonian is not necessarily that projection, so its eigenvalues do not inherit the same rigorous bound. Tight binding can be both a teaching model and a practical electronic-structure approximation, but those uses have different validation requirements. Before fitting parameters, define the target subspace, electron count, symmetry, boundary conditions, and which observables must be reproduced.
2. Derivation step by step
Read each equality with its assumptions. Atomic units are used for DFT equations unless another unit is stated; TB parameters retain explicit energy and length units. The conjugate transpose is denoted by a dagger, and a prime on a coordinate denotes a separate integration variable.
2.1. Orthogonalization changes the Hamiltonian too
Diagonalize S=Udiag(s)U† and form S^-1/2=Udiag(s^-1/2)U† when all s are positive. Defining d=S^1/2c gives d†d=c†Sc and Horth=S^-1/2HS^-1/2. The generalized problem becomes Horth d=Ed. Keeping H unchanged while replacing S by the identity is not orthogonalization. Small overlap eigenvalues amplify matrix errors through the inverse square root. Removing nearly dependent basis directions is sometimes necessary, but it changes the represented subspace and should be documented. Population analysis in the original basis also needs an explicitly defined metric convention.
3. Worked example
For ε=0 and t=−2 eV, the symmetric combination has energy −2 eV and the antisymmetric combination +2 eV. With overlap s, the energies become (ε+t)/(1+s) and (ε−t)/(1−s). At s=0.2 they are −1.667 and +2.5 eV. Normalizing coefficient vectors in Euclidean norm would misnormalize physical states.
4. Exercises with explained solutions
Exercise. Find the eigenvalues of S=[[1,s],[s,1]] and its positivity condition.
Explained solution. The symmetric and antisymmetric eigenvectors give 1+s and 1−s. Both are positive only when |s|<1. As |s| approaches one, one linear combination becomes nearly redundant and numerical conditioning deteriorates.
Further check. State the units and the allowed regime for every parameter in the worked example. Change one assumption and identify which derivation step must be revisited. A correct explanation names the affected constraint, operator, or boundary condition rather than merely saying that the answer changes.
5. Misconceptions and limitations
A small matrix is not automatically a faithful model. Missing orbitals, nonorthogonality, and environment dependence can all change the observable of interest.
The illustration is an original teaching schematic. It is not output from a numerical materials simulation.
6. Connections and sources
Related: density-functional foundations · Interacting Monte Carlo methods
- Slater–Koster, original orbital-geometry construction (1954)
- Su–Schrieffer–Heeger, solitons in polyacetylene (1979)
- Marzari et al., maximally localized Wannier functions review (2012)
The explanations, algebra, and invented worked examples are original teaching synthesis. The cited papers establish the underlying theories, not the numerical toy values.
7. Related theory and practice
Quantum mechanics · Density functional theory · ABACUS · DFTB+