Math Needed for Computational QM

This crosswalk is for readers preparing for computational quantum mechanics: Hamiltonian matrix construction, grids, basis truncations, diagonalization, time evolution, wave-packet propagation, scattering numerics, many-body exact diagonalization, open-system solvers, quantum-circuit simulation, benchmark problems, and reproducible notebooks.

The planned Computational QM volume will own reusable algorithms, notebooks, testing policies, and software-facing guidance. Computational Quantum Matter owns material-facing workflow and evidence routing. This page gathers the mathematical tools that determine whether a calculation is meaningful: numerical linear algebra, discretization, ODE and PDE solvers, FFTs, Monte Carlo, stability, convergence, and error estimates.

Minimum Tools

Start with states and operators as arrays, basis choice, inner products, norms, Hermitian matrices, tensor products, matrix diagonalization, sparse matrices, sparse eigensolvers, discretization, finite differences, spectral methods, numerical quadrature, FFTs, matrix exponentials, ODE solvers, PDE solvers, Monte Carlo, floating-point arithmetic, conditioning and stability, error estimates, convergence tests, and benchmark problems.

The basic computational move is to replace an abstract quantum problem by a controlled finite representation:

$$Hc=Ec, \qquad r=Hc-Ec, \qquad \lVert r\rVert \ll \lVert H\rVert\,\lVert c\rVert.$$

The residual $r$ is one of the simplest honesty checks: an eigenvalue printed by software is not yet a validated quantum result.

For dynamics, the corresponding move is to approximate the propagator or its action:

$$\psi(t+\Delta t) \approx U_{\Delta t}\psi(t), \qquad U_{\Delta t} \approx \exp\left(-\frac{iH\Delta t}{\hbar}\right).$$

Different algorithms preserve different structures: norm, phase accuracy, energy conservation, positivity, trace preservation, locality, or symplectic structure. A reliable calculation states which structure matters and how it was checked.

Recommended Tools by Topic

Computational topic	Mathematical tools
States and operators as arrays	finite-dimensional Hilbert spaces, bases and coordinates, matrices as linear maps, inner products
Basis choice and truncation	orthonormal bases, change of basis, direct sums, error estimates
Dense eigenproblems	eigenvalues and eigenvectors, Hermitian operators, matrix diagonalization, residual checks
Large sparse Hamiltonians	sparse matrices, sparse eigensolvers, conditioning and stability, convergence tests
Grid wavefunctions	discretization, finite difference methods, boundary conditions, numerical quadrature
Spectral and pseudospectral methods	spectral methods, Fourier series, Fourier transform, fast Fourier transform
Time evolution	matrix exponentials, matrix exponentials numerically, time-stepping methods, ODE solvers
Schrödinger PDEs	partial differential equations, PDE solvers, finite difference methods, spectral methods
Wave packets and FFT workflows	wave packets, momentum representation, FFT, convolution
Variational numerics	variational principle, Rayleigh–Ritz method, trial wavefunctions, basis-set convergence
Scattering numerics	boundary conditions, ODE solvers, Phase-Shift Extraction, benchmark problems
Many-body exact diagonalization	Math Needed for Many-Body QM, tensor products, occupation-number basis, sparse eigensolvers
Open-system simulation	Math Needed for Open Systems, density operators, matrix exponentials numerically, Monte Carlo basics
Quantum-information simulation	Math Needed for Quantum Information, tensor products, Pauli matrices, Bloch sphere geometry
Statistical estimates	Monte Carlo basics, expectation values, variance and covariance, error estimates
Validation and reproducibility	floating-point arithmetic, conditioning and stability, convergence tests, benchmark problems

Suggested Reading Order

For representation and algebra, read Dirac Notation as Linear Algebra, Bases and Coordinates, Matrices as Linear Maps, Inner Products, Norms and Metrics, Hermitian Operators, Unitary Operators, Tensor Products, and Direct Sums.

For numerical linear algebra, read Floating-Point Arithmetic, Conditioning and Stability, Matrix Diagonalization, Sparse Matrices, Sparse Eigensolvers, Matrix Functions and Exponentials, and Matrix Exponentials Numerically.

For grid, basis, and continuum calculations, read Discretization, Boundary Conditions, Finite Difference Methods, Spectral Methods, Numerical Quadrature, ODE Solvers, PDE Solvers, Fast Fourier Transform, and Fourier Transform Conventions.

For uncertainty and reproducibility, read Error Estimates, Convergence Tests, Benchmark Problems, Monte Carlo Basics, Expectation Values, and Variance and Covariance.

For a Monte Carlo estimate, the statistical object is usually a sample mean such as

$$\bar f_N = \frac{1}{N}\sum_{j=1}^N f(X_j), \qquad \operatorname{SE}(\bar f_N) \approx \sqrt{\frac{\operatorname{Var}(f)}{N_{\mathrm{eff}}}}.$$

The effective sample size $N_{\mathrm{eff}}$ may be much smaller than $N$ when samples are correlated. A computational result should not hide that distinction.

Planned Computational-QM Targets

When the computational volume is added, these planned pages should use this crosswalk as their prerequisite map:

Computational Many-Body QM is the current physics-facing chapter entry. Its Computational Many-Body Overview uses the tools below to choose among finite-vector, stochastic, compressed-state, and dynamical representations without duplicating their numerical derivations.

Planned page	Current prerequisite homes
foundations/states-operators-arrays	bases, matrices as linear maps, inner products, norms
foundations/basis-choice-representation	orthonormal bases, change of basis, direct sums, truncation error
foundations/dense-vs-sparse	matrix diagonalization, sparse matrices, conditioning
foundations/convergence-studies	convergence tests, error estimates, benchmark problems
single-particle/finite-difference-schrodinger-equation	discretization, finite differences, boundary conditions, PDE solvers
single-particle/benchmark-particle-in-box	benchmark problems, finite differences, quadrature, exact solutions
single-particle/benchmark-harmonic-oscillator	harmonic oscillator, spectral methods, basis truncation
time-dynamics/matrix-exponentials	matrix functions, matrix exponentials numerically, unitarity checks
time-dynamics/split-operator-method	FFTs, Fourier conventions, operator splitting, error estimates
spectral-variational/matrix-diagonalization	Hermitian eigenproblems, residuals, degeneracies
spectral-variational/sparse-eigensolvers	sparse matrices, Krylov methods, residual diagnostics
scattering-continuum/phase-shift-extraction	ODE solvers, boundary conditions, scattering conventions
many-body-exact/exact-diagonalization	tensor products, occupation bases, sparse matrices, symmetry sectors
open-systems-control/lindblad-master-equation-solvers	density operators, matrix exponentials, ODE solvers, trace checks
quantum-matter-computation/chern-number-computation	Berry connection, Chern numbers, convergence tests
reproducibility-benchmarks-data/benchmark-problem-design	benchmark problems, error estimates, metadata, validation

Common Mistakes

• Reporting an eigenvalue without a residual, convergence check, or benchmark comparison.
• Normalizing grid wavefunctions without quadrature weights.
• Confusing basis truncation error with solver tolerance.
• Using a time-stepper that is stable but not appropriate for the conserved quantity being studied.
• Treating a finite box as the continuum without checking box-size dependence.
• Forgetting tensor-product ordering or fermionic sign conventions in many-body code.
• Reporting Monte Carlo error bars without checking autocorrelation and systematic error.
• Comparing results that use different units, Fourier conventions, boundary conditions, or basis orderings.
• Mistaking a package default for a physical convention.

Cross-Links

• Benchmark Problems
• Convergence Tests
• Error Estimates
• Sparse Eigensolvers
• Fast Fourier Transform
• Monte Carlo Basics
• Math Needed for Many-Body QM
• Math Needed for Open Systems

References

• L. N. Trefethen and D. Bau III, Numerical Linear Algebra, SIAM, 1997.
• Y. Saad, Numerical Methods for Large Eigenvalue Problems, 2nd ed., SIAM, 2011.
• R. J. LeVeque, Finite Difference Methods for Ordinary and Partial Differential Equations, SIAM, 2007.
• J. M. Thijssen, Computational Physics, 2nd ed., Cambridge University Press, 2007.
• M. E. J. Newman and G. T. Barkema, Monte Carlo Methods in Statistical Physics, Oxford University Press, 1999.
• H. Fehske, R. Schneider, and A. Weisse, eds., Computational Many-Particle Physics, Springer, 2008.

Exercises

1. A sparse eigensolver returns an approximate eigenpair $(E,c)$. What simple residual should you compute first?

Solution

Compute

$$r=Hc-Ec.$$

Then compare $\lVert r\rVert$ to a relevant scale such as $\lVert H\rVert\lVert c\rVert$ or to the spectral gap when resolving nearly degenerate states. A small change in the printed eigenvalue alone is not enough.

2. Why does a grid representation of $\psi(x)$ require quadrature weights in normalization?

Solution

The continuum norm is an integral:

$$\int \lvert\psi(x)\rvert^2\,dx.$$

A grid approximation replaces this by a weighted sum. On a uniform grid the weight is usually $\Delta x$; on nonuniform grids or quadrature rules the weights vary. Normalizing only the array entries can give the wrong continuum norm.

3. Name three independent checks for a time-evolution simulation.

Solution

Possible checks include norm conservation for closed unitary evolution, energy conservation for a time-independent Hamiltonian when the method should preserve it approximately, convergence under time-step refinement, comparison to an exactly solvable benchmark, agreement between two independent methods, and absence of boundary reflections in a finite computational box.