research28 min studyLesson 782 in Physics

Supersymmetric quantum mechanics and the Witten index

Factor partner Hamiltonians, identify normalizable zero modes, and state the spectral and boundary assumptions behind an index argument.

On this page

What you’ll understand

  • Derive paired partner Hamiltonians from one first-order operator.
  • Determine zero modes from normalizability at both infinities.
  • Explain why a zero Witten index is inconclusive and why boundary changes matter.

Supersymmetric quantum mechanics compresses many important ideas into a solvable system: positive energy, paired excited states, unpaired ground states, and sensitivity to behavior at infinity. It is a mathematical model rather than a claim about new elementary particles. We will use a dimensionless coordinate and oscillator energy units, explicitly choosing a one-dimensional supercharge normalization that differs from the four-dimensional algebra. The zero-mode calculation will show why solving a differential equation is only half the work; its solution must also belong to the Hilbert space.

Factorization creates two related Hamiltonians

Let A=(∂x+u(x))/2A=(\partial_x+u(x))/\sqrt2 on square-integrable wavefunctions on the real line, where u is a real smooth function. With a common supersymmetry-preserving domain and vanishing integration-by-parts boundary terms, its adjoint is A†=(−∂x+u)/2A^\dagger=(-\partial_x+u)/\sqrt2. Define Q to map the upper, even-parity sector into the lower, odd-parity sector through A. Set H={Q,Q†}H=\{Q,Q^\dagger\} in this lesson. This normalization has no extra factor of two and fixes the oscillator energy spacings quoted below.

Multiplication gives partner potentials involving u squared and opposite signs of its derivative. Both Hamiltonians are nonnegative because they are of the forms A dagger A and A A dagger. If an upper-sector eigenstate has positive energy E, applying A produces a lower-sector state of the same energy; its norm squared is E times the original norm. Division by square root E normalizes the partner. At zero energy this construction can vanish, allowing an unpaired state rather than forcing equal numbers of zero modes.

Q=(00A0),H={Q,Q†}=(A†A00AA†),HB,F=12[−∂x2+u2∓u′]Q=\begin{pmatrix}0&0\\A&0\end{pmatrix},\qquad H=\{Q,Q^\dagger\}=\begin{pmatrix}A^\dagger A&0\\0&AA^\dagger\end{pmatrix},\qquad H_{B,F}=\frac12[-\partial_x^2+u^2\mp u']
B and F name the even and odd graded sectors. They need not represent actual bosonic or fermionic particles in space. The dimensionless units and common operator domain are part of this particular quantum-mechanical model.↔ Scroll wide equations horizontally.

A formal zero mode may fail at infinity

An even zero mode obeys AψB=0A\psi_B=0, giving a wavefunction proportional to the exponential of minus the integral of u. An odd zero mode obeys A dagger psi F equals zero and has the opposite exponential. Normalization requires the integral of the squared magnitude over the entire real line to converge. A solution decaying at one end but growing at the other is not a ground state. Boundary conditions on a finite interval would also have to preserve both the adjoint relation and the supersymmetry action.

For u equal to x, the even solution is Gaussian and normalizable, while the odd solution grows. The even Hamiltonian has energies zero, one, two, and so on, and the odd Hamiltonian starts at one. Each positive energy is paired and the ground state is unpaired. Reversing the sign of u interchanges which parity contains the Gaussian. These examples show that the sign of an index is a grading convention and zero-mode orientation, rather than a negative number of physical states.

ψB∝e−∫xu(s) ds,ψF∝e+∫xu(s) ds,I(β)=Tr⁡[(−1)Fe−βH]=nB(0)−nF(0)\psi_B\propto e^{-\int^x u(s)\,ds},\qquad \psi_F\propto e^{+\int^x u(s)\,ds},\qquad I(\beta)=\operatorname{Tr}[(-1)^F e^{-\beta H}]=n_B^{(0)}-n_F^{(0)}
The trace identity assumes paired positive spectrum and a well-defined regulated trace with no unaccounted continuum or boundary contribution. Beta is inverse energy in the selected units; n-zero counts normalizable zero-energy states.↔ Scroll wide equations horizontally.

Index protection requires control of the spectrum

A nonzero well-defined index proves at least one supersymmetric zero-energy state exists. A zero index does not prove supersymmetry breaking because equal numbers of even and odd zero modes can cancel. To conclude breaking one must show the actual absence of normalizable zero-energy states and the existence of a suitable ground state. An infimum reached only at infinity is not a normalizable vacuum. These logical distinctions reappear in field theories with flat directions, runaway potentials, and continuous spectra.

The usual deformation-invariance argument cancels positive-energy pairs while varying parameters smoothly. It assumes states cannot escape to infinity, the relevant domains remain compatible, and the trace or spectral regulator behaves properly. Noncompact directions or a continuum reaching zero can violate those assumptions. For example, changing the asymptotic growth of u can send a normalizable Gaussian into a nonnormalizable limit. Declaring the index topological without examining this boundary behavior would hide the very mechanism by which its apparent value can change.

A doubled oscillator slope

Take u(x)=2xu(x)=2x on the full real line with the stated factorization and normalizable oscillator domain. Find both zero-mode candidates, the Witten index, and the first paired positive energy.

  1. The integral of u is x squared. Thus the even candidate is e−x2e^{-x^2} and the odd candidate is e+x2e^{+x^2}. Only the first has finite squared integral at both ends.

  2. The even partner is HB=[−∂x2+4x2−2]/2H_B=[-\partial_x^2+4x^2-2]/2. The oscillator of frequency two has energies 2n plus one before the subtraction, leaving energies 2n.

  3. The odd spectrum is 2n plus two. Therefore the index is +1 and the first positive pair has energy two, from even level n = 1 and odd level n = 0.

The result

There is one even Gaussian zero mode, no odd zero mode, index +1, and a first paired energy of two in the chosen units.

Explore the model

Interactive lab

Interactive lab · oscillator

Frequency scales a classical period

Reset and set BOTH Damping ratio ζ = 0 and Natural frequency ω₀ = 1; predict period 2π. For a fresh comparison, Reset and set ζ = 0 and ω₀ = 2; predict period π. Play or Step each reset trajectory from x = 1 and velocity zero.

Frequency scales a classical periodUnit mass, x(0) = 1 and v(0) = 0. The analytic solution includes underdamped, critically damped and overdamped cases; time is measured in seconds. Current readings: Displacement x(t): 1; Velocity v(t): 0; Regime: underdamped. All parameters are adjustable with the labeled controls below.02.557.510−1.15−0.5700.571.15Time · sDisplacementunderdamped
Displacement x(t)Current time
Displacement x(t)
1
Velocity v(t)
0
Regime
underdamped
t = 0 s

Unit mass, x(0) = 1 and v(0) = 0. The analytic solution includes underdamped, critically damped and overdamped cases; time is measured in seconds. Parameter changes pause playback and recompute the model at the current elapsed time. Reset restores both parameters and returns time to zero. Playback starts paused. Adjust either parameter, then choose Play or Step.

Look forThis analytic classical oscillator illustrates a frequency scale. It does not calculate quantum zero modes, partner spectra, fermion grading, or a Witten index, and supplies no SUSY solver.

Check your understanding

What follows from a well-defined Witten index equal to zero?

Try it yourself

Replace the worked example by u(x)=−3xu(x)=-3x on the same real-line Hilbert space. Determine the normalizable zero-mode parity and index. Explain why the growing candidate cannot be retained as a second ground state.

Take a moment to make a prediction or work it out on paper. Open a hint when you need one.

Give me a hint

Integrate u first, then inspect each exponential at both positive and negative infinity.

Compare with the solution

The integral is −3x2/2-3x^2/2. The even candidate grows as e+3x2/2e^{+3x^2/2}, while the odd candidate decays as e−3x2/2e^{-3x^2/2}. Only the odd one is square integrable, giving zero even zero modes, one odd zero mode, and index minus one. The growing function solves the formal first-order equation but lies outside the Hilbert space and the normalizable oscillator domain, so it contributes neither a physical state nor an index term.

Take it with you

  • Factorized partner Hamiltonians pair positive-energy states.
  • Zero-mode equations require a separate normalizability and domain check.
  • A protected index needs controlled boundaries and spectrum; zero alone does not imply breaking.

Read further

Original sources and deeper explanations to follow at your own pace.

Optional progress, saved in this browser when available. Revisit any lesson at any time.

Lesson not marked complete.