Supersymmetric quantum mechanics: superpotential, supercharges, and the Witten index

Intuition Beginner

Picture a quantum system that comes in two copies side by side, labelled "bosonic" and "fermionic." Supersymmetric quantum mechanics is a recipe for building two such copies so that their energy levels match up almost perfectly. Every energy a particle can have in the first copy is also available in the second copy — with one possible exception, the very bottom of the energy ladder. The two copies are stitched together by an operation that turns a bosonic state into a fermionic one and back again.

The single input to the whole construction is one function of position, called the superpotential and written $W(x)$. From its slope you build the potential energy landscape of each copy. The two landscapes are close cousins: they differ only by a term set by the curvature of $W$, and this small difference is exactly what shifts the bottom rung.

The interesting question is whether there is a state sitting at exactly zero energy. If there is, the system is called unbroken; if the would-be zero-energy state fails to be a proper normalisable wavefunction, the supersymmetry is spontaneously broken and the true ground state floats up to positive energy.

A robust integer keeps score. Count the zero-energy states in the bosonic copy, subtract the zero-energy states in the fermionic copy, and you get a whole number called the Witten index. Because every state above zero energy is paired between the two copies, those paired states cancel in the count and never affect the answer. The index is therefore stubborn: you can deform the system, change its parameters, even reshape $W$, and the integer does not budge. It is a counting invariant that survives almost any change you make.

Visual Beginner

Two vertical energy ladders drawn side by side: the left ladder is the bosonic copy with Hamiltonian $H_{+}$, the right is the fermionic copy with $H_{-}$. Above zero energy each rung on the left lines up with a rung on the right, joined by a horizontal arrow labelled with the operation that maps one copy to the other. At the very bottom, the left ladder has an extra rung resting exactly on the zero-energy line while the right ladder's lowest rung sits above it — the unmatched ground state that the Witten index counts.

The picture captures the whole story: two energy ladders that agree everywhere except possibly at the bottom rung, and an integer that records how many bottom rungs go unmatched.

Worked example Beginner

Take the simplest superpotential, the straight line through the oscillator, and read off both partner potentials.

Step 1. Choose $W(x)=\frac{1}{2}\omega x^2$, the superpotential whose slope is $W^{\prime }(x)=\omega x$ and whose curvature is $W^{\prime \prime }(x)=\omega$. This is the choice that reproduces the quantum harmonic oscillator of 12.04.02.

Step 2. The two partner potentials are $V_{+}(x)=\frac{1}{2}(W^{\prime }{)}^2+\frac{1}{2}{W}^{\prime \prime }$ and $V_{-}(x)=\frac{1}{2}(W^{\prime }{)}^2-\frac{1}{2}{W}^{\prime \prime }$. Substituting gives $V_{+}(x)=\frac{1}{2}{\omega }^2{x}^2+\frac{1}{2}\omega$ and $V_{-}(x)=\frac{1}{2}{\omega }^2{x}^2-\frac{1}{2}\omega$.

Step 3. Both are oscillator potentials of the same shape; they differ only by the constant shift $\omega$. The bosonic copy is lifted by $\frac{1}{2}\omega$, the fermionic copy is lowered by $\frac{1}{2}\omega$. The bottom of the fermionic copy therefore sits at zero energy.

Step 4. The zero-energy state lives in whichever copy admits a normalisable wavefunction $e^{-W(x)}=e^{-\omega x^2/2}$. This Gaussian decays at large distance, so it is a genuine bound state. The system is unbroken, with exactly one zero-energy ground state.

Step 5. Count for the index. The bosonic copy has one zero-energy state and the fermionic copy has none, so the Witten index is $1-0=1$. Every higher level is shared between the two copies, so those levels never enter the count.

What this tells us: a single quadratic superpotential reproduces the oscillator, splits it into two shifted copies, and leaves one unmatched state at the bottom. The integer that counts unmatched states comes out to one, and no small change to $\omega$ or to the shape of $W$ near infinity will alter that.

Check your understanding Beginner

Formal definition Intermediate+

Fix the one-dimensional configuration space $\mathbb{R}$ with coordinate $x$ and momentum operator $p=-id/dx$ (units $\hslash =1$, mass $=1$). Adjoin a single fermionic mode: operators $\psi ,{\psi }^{†}$ obeying the anticommutation relations $$ {\psi, \psi^\dagger} = 1, \qquad \psi^2 = (\psi^\dagger)^2 = 0. $$ The fermionic Fock space is two-dimensional, spanned by an empty state $|0⟩$ with $\psi |0⟩=0$ and a filled state ${\psi }^{†}|0⟩$. The graded Hilbert space of the theory is $$ \mathcal{H} = \mathcal{H}+ \oplus \mathcal{H}-, \qquad \mathcal{H}+ = L^2(\mathbb{R}) \otimes |0\rangle, \quad \mathcal{H}- = L^2(\mathbb{R}) \otimes \psi^\dagger|0\rangle, $$ with the fermion-number operator $F={\psi }^{†}\psi$ acting as $0$ on ${\mathcal{H}}_{+}$ and $1$ on ${\mathcal{H}}_{-}$. The fermion parity is $(-1{)}^F$, equal to $+1$ on ${\mathcal{H}}_{+}$ and $-1$ on ${\mathcal{H}}_{-}$.

Let $W:\mathbb{R}\to \mathbb{R}$ be a smooth superpotential. The supercharge and its adjoint are $$ Q = \psi^\dagger,(p - i,W'(x)), \qquad Q^\dagger = \psi,(p + i,W'(x)), $$ where $W^{\prime }(x)=dW/dx$. They are nilpotent, $Q^2=(Q^{†}{)}^2=0$, because $({\psi }^{†}{)}^2={\psi }^2=0$. The Hamiltonian is their anticommutator, $$ H = {Q, Q^\dagger} = QQ^\dagger + Q^\dagger Q. $$ Working out the anticommutator and using $[p,W^{\prime }]=-i{W}^{\prime \prime }$ gives the explicit form $$ H = \tfrac{1}{2}\big(p^2 + W'(x)^2\big) - \tfrac{1}{2},W''(x),[\psi^\dagger, \psi] = H_+ \oplus H_-, $$ where the two partner Hamiltonians are the one-dimensional Schrödinger operators $$ H_\pm = \tfrac{1}{2}p^2 + V_\pm(x), \qquad V_\pm(x) = \tfrac{1}{2},W'(x)^2 \pm \tfrac{1}{2},W''(x). $$ Here $V_{\pm }$ are the partner potentials: $H_{+}$ acts on the bosonic sector ${\mathcal{H}}_{+}$ and $H_{-}$ on the fermionic sector ${\mathcal{H}}_{-}$. The factor $\frac{1}{2}(W^{\prime }{)}^2$ is common; the curvature term $\pm \frac{1}{2}{W}^{\prime \prime }$ splits them.

The structure is the $\mathcal{N}=2$ supersymmetry algebra: $Q^2=(Q^{†}{)}^2=0$, $\{Q,Q^{†}\}=H$, and $[H,Q]=[H,Q^{†}]=0$, the last because $H$ is built from $Q$ and $Q^{†}$. Since $H=\{Q,Q^{†}\}$ is a sum of $A^{†}A$-type terms, it is a positive operator: $⟨\chi |H|\chi ⟩=\Vert Q\chi {\Vert }^2+\Vert Q^{†}\chi {\Vert }^2\ge 0$ for every $\chi \in \mathcal{H}$.

A state $|\Omega ⟩$ has zero energy iff $H|\Omega ⟩=0$, which by positivity is equivalent to $Q|\Omega ⟩=Q^{†}|\Omega ⟩=0$. Such a state is supersymmetric: it is annihilated by both supercharges. Supersymmetry is unbroken when a normalisable zero-energy state exists, and spontaneously broken when none does, in which case the ground-state energy is strictly positive and serves as the order parameter for breaking.

The Witten index is the graded trace $$ \Delta = \mathrm{tr},(-1)^F = n_B^{E=0} - n_F^{E=0}, $$ the number of bosonic zero-energy states minus the number of fermionic zero-energy states. The regularised form $\Delta =\mathrm{tr}(-1{)}^F{e}^{-\beta H}$ is independent of $\beta >0$ because every nonzero-energy state is paired between the sectors (shown in the next section).

Counterexamples to common slips

• The sign in front of $W^{\prime \prime }$ tracks the sector, not a free choice. $H_{+}$ carries $+\frac{1}{2}{W}^{\prime \prime }$ and acts on the bosonic sector ${\mathcal{H}}_{+}$; flipping the convention silently swaps which copy hosts the zero-energy state and flips the sign of the index.
• Normalisability is decided at both ends of the line. The zero-energy candidate is ${\psi }_0\propto e^{-{\int }^x{W}^{\prime }}=e^{-W(x)}$ in the bosonic sector and $e^{+W(x)}$ in the fermionic sector. At most one of $e^{\pm W}$ decays at both $\pm \infty$; if $W(\pm \infty )$ have the same sign, neither does and supersymmetry is broken.
• A nonzero index forbids breaking, but a zero index does not force it. $\Delta =0$ is consistent with either an unbroken vacuum (equal bosonic and fermionic zero-energy counts) or a broken one (no zero-energy states at all); the index alone cannot distinguish these.

Key theorem with proof Intermediate+

Theorem (partner isospectrality and the SUSY pairing; Witten 1981 ^{[Witten-1981]}, Cooper-Khare-Sukhatme 1995 ^{[Cooper-Khare-Sukhatme]}). Let $H_{\pm }=\frac{1}{2}{p}^2+V_{\pm }$ with $V_{\pm }=\frac{1}{2}(W^{\prime }{)}^2\pm \frac{1}{2}{W}^{\prime \prime }$ be the partner Hamiltonians of a superpotential $W$, and set $A=p-i{W}^{\prime }(x)$, so that $H_{+}=\frac{1}{2}{A}^{†}A$ and $H_{-}=\frac{1}{2}A{A}^{†}$. Then $H_{+}$ and $H_{-}$ have identical spectra on the strictly positive part of the line: for every eigenvalue $E>0$ with $H_{+}{\psi }_{+}=E{\psi }_{+}$, the state ${\psi }_{-}=A{\psi }_{+}$ satisfies $H_{-}{\psi }_{-}=E{\psi }_{-}$, and the map ${\psi }_{+}\mapsto A{\psi }_{+}$ is a bijection between the $E$-eigenspaces of $H_{+}$ and $H_{-}$. The only possible spectral asymmetry is at $E=0$.

Proof. Write $A=p-i{W}^{\prime }$ and compute $A^{†}A$ and $A{A}^{†}$. With $p=-id/dx$ one has $A^{†}=p+i{W}^{\prime }$, and $$ A^\dagger A = (p + iW')(p - iW') = p^2 + (W')^2 + i[W', p] = p^2 + (W')^2 - W'', $$ using $[W^{\prime },p]=i{W}^{\prime \prime }$. Likewise $$ A A^\dagger = (p - iW')(p + iW') = p^2 + (W')^2 - i[W', p] = p^2 + (W')^2 + W''. $$ Halving gives $H_{+}=\frac{1}{2}{A}^{†}A$ with potential $\frac{1}{2}(W^{\prime }{)}^2+\frac{1}{2}{W}^{\prime \prime }=V_{+}$ and $H_{-}=\frac{1}{2}A{A}^{†}$ with potential $\frac{1}{2}(W^{\prime }{)}^2-\frac{1}{2}{W}^{\prime \prime }=V_{-}$. (The role of $\pm$ is fixed by the convention that $H_{+}$ acts on the empty fermionic state.)

Now suppose $H_{+}{\psi }_{+}=E{\psi }_{+}$, that is $\frac{1}{2}{A}^{†}A{\psi }_{+}=E{\psi }_{+}$, with $E>0$ and ${\psi }_{+}$ normalisable. Apply $A$ on the left: $$ H_- (A\psi_+) = \tfrac{1}{2}A A^\dagger A \psi_+ = A\big(\tfrac{1}{2}A^\dagger A\big)\psi_+ = A (E\psi_+) = E,(A\psi_+). $$ So $A{\psi }_{+}$ is an eigenstate of $H_{-}$ with the same eigenvalue $E$. It is nonzero: if $A{\psi }_{+}=0$ then $H_{+}{\psi }_{+}=\frac{1}{2}{A}^{†}A{\psi }_{+}=0$, contradicting $E>0$. Its norm is $\Vert A{\psi }_{+}{\Vert }^2=⟨{\psi }_{+}|A^{†}A|{\psi }_{+}⟩=2E\Vert {\psi }_{+}{\Vert }^2<\infty$, so $A{\psi }_{+}$ is normalisable. The same argument with $A^{†}$ maps $H_{-}$-eigenstates back to $H_{+}$-eigenstates, and $A^{†}A{\psi }_{+}=2E{\psi }_{+}$ shows the two maps are mutually inverse up to the constant $2E$. The eigenspaces at energy $E>0$ are therefore in bijection, and the only place the spectra can differ is the kernel at $E=0$. $\square$

Bridge. This pairing is the foundational reason the Witten index is well defined and builds toward its deformation invariance: every positive-energy state lives in a boson-fermion doublet, so it contributes $(+1)+(-1)=0$ to $\mathrm{tr}(-1{)}^F$, and only the unpaired $E=0$ states survive in the count. This is exactly the cancellation that makes $\mathrm{tr}(-1{)}^F{e}^{-\beta H}$ independent of $\beta$, a structure that appears again in the heat-kernel computation of the index below and that generalises far beyond quantum mechanics. The same intertwining operator $A$ is dual to the boundary operator of Hodge theory: putting these together, the factorisation $H_{\pm }=\frac{1}{2}{A}^{†}A,\frac{1}{2}A{A}^{†}$ is the operator-theoretic shadow of the geometric statement that the deformed exterior derivative and its adjoint share their nonzero spectrum, the central insight that the companion Morse-theory reading 03.15.12 turns into the Morse inequalities. The bridge is the recognition that the drift $-W^{\prime }$ inside $A$ is the very drift of a Langevin process, so the same factorisation appears again in the Fokker-Planck operator of 08.10.02 — the link that carries SUSY-QM into stochastic quantisation and forward to the Nicolai map of 08.10.12.

Exercises Intermediate+

Advanced results Master

Theorem (the Witten index is a deformation invariant; Witten 1982 ^{[Witten-1982-constraints]}). Let $H(\lambda )$ be a smooth one-parameter family of SUSY-QM Hamiltonians with supercharge $Q(\lambda )$, $\{Q(\lambda ),Q(\lambda {)}^{†}\}=H(\lambda )$, such that the spectrum is discrete near $E=0$ and the essential spectrum stays bounded away from $0$ along the family. Then the Witten index $\Delta =\mathrm{tr}(-1{)}^F$ is independent of $\lambda$. Equivalently, the regularised trace $\Delta (\beta )=\mathrm{tr}(-1{)}^F{e}^{-\beta H}$ is independent of both $\beta >0$ and $\lambda$.

The argument has two halves. First, $\beta$-independence: differentiate $\Delta (\beta )=\mathrm{tr}(-1{)}^F{e}^{-\beta H}$ in $\beta$ to get $-\mathrm{tr}(-1{)}^{F}H{e}^{-\beta H}$. On every nonzero-energy doublet $\{|b⟩,|f⟩=E^{-1/2}{Q}^{†}|b⟩\}$ the operator $(-1{)}^F$ takes opposite signs while $H{e}^{-\beta H}$ takes equal values $E{e}^{-\beta E}$, so the doublet contributes $E{e}^{-\beta E}(+1-1)=0$. Zero-energy states are killed by $H$. Hence $d\Delta /d\beta =0$. Second, $\lambda$-independence: as $\lambda$ varies, energy levels move continuously, but a state can leave or enter $E=0$ only in a boson-fermion pair (because nonzero levels are paired by $Q^{†}$), so $n_B^{E=0}-n_F^{E=0}$ cannot jump. The index is therefore a robust integer, computable in any convenient limit of the family.

Theorem (the topological index formula; Witten 1981 ^{[Witten-1981]}). For a superpotential $W$ with $W^{\prime }(x)\ne 0$ outside a compact set and $|W^{\prime }(x)|\to \infty$ as $|x|\to \infty$, the Witten index equals $$ \Delta = \tfrac{1}{2}\big(\mathrm{sgn},W'(+\infty) - \mathrm{sgn},W'(-\infty)\big) = -\tfrac{1}{2}\big(\mathrm{sgn},W(+\infty)^{-} - \cdots \big), $$ so $\Delta =\pm 1$ when $W^{\prime }$ has opposite signs at the two ends (a $W$ growing to $+\infty$ one way and $-\infty$ the other) and $\Delta =0$ when $W^{\prime }$ has the same sign at both ends (an even-degree $W$). The index depends only on the asymptotic behaviour of the superpotential, not on its detailed shape.

The proof reduces to the deformation invariance: scale $W\to sW$ and send $s\to \infty$, localising the ground-state wavefunction at the zeros of $W^{\prime }$, then count the contribution of each zero by the sign of $W^{\prime \prime }$ there. The net count telescopes to the boundary signs of $W^{\prime }$. This is the quantum-mechanical avatar of the Poincaré-Hopf theorem: $\Delta$ is the signed count of critical points of $W$, which is the Euler characteristic of the relevant configuration in the Morse-theory reading of 03.15.12.

Theorem (factorisation and the SUSY-QM hierarchy; Cooper-Khare-Sukhatme 1995 ^{[Cooper-Khare-Sukhatme]}). If $H_{+}=\frac{1}{2}{A}^{†}A$ has discrete spectrum $0=E_0<E_1<E_2<\cdots$ with normalisable ground state, the partner $H_{-}=\frac{1}{2}A{A}^{†}$ has spectrum $E_1<E_2<\cdots$ — the same as $H_{+}$ with the ground state deleted. Iterating the factorisation produces a hierarchy of Hamiltonians $H^{(1)}=H_{+},H^{(2)},H^{(3)},\dots$, each the partner of the last with its ground state removed, so the entire spectrum of $H_{+}$ is generated algebraically when the potentials are shape invariant, $V_{-}(x;a_0)=V_{+}(x;a_1)+R(a_0)$ for parameters $a_1=f(a_0)$.

Shape invariance is the algebraic engine behind every exactly-solvable one-dimensional potential. The oscillator, the Coulomb problem, the Pöschl-Teller and Morse potentials, the Rosen-Morse and Eckart potentials all satisfy a shape-invariance relation, and their full spectra follow from the SUSY-QM ladder without solving a differential equation. The harmonic oscillator of 12.04.02 is the simplest instance: $W=\frac{1}{2}\omega x^2$ gives shape-invariant partners differing by the constant $R=\omega$, and the hierarchy reproduces the spectrum $E_n=n\omega$ rung by rung.

Theorem (the index as a regularised supertrace; Cecotti-Girardello 1982 ^{[Cecotti-Girardello]}). The Witten index admits the path-integral representation $$ \Delta = \mathrm{tr}(-1)^F e^{-\beta H} = \int_{\text{periodic}} \mathcal{D}x,\mathcal{D}\psi,\mathcal{D}\psi^\dagger; e^{-S_E[x, \psi, \psi^\dagger]}, $$ where the Euclidean action $S_E$ is the SUSY-QM action with periodic boundary conditions in imaginary time for both bosons and fermions, and the $(-1{)}^F$ insertion is implemented by the periodic (rather than antiperiodic) fermion boundary condition. The result is $\beta$-independent and localises onto the constant (zero-mode) configurations, reducing to the index integral $\frac{1}{2\pi }\int dx{W}^{\prime \prime }{e}^{-W^{\prime 2}/2}(\cdots )$ that evaluates to the boundary-sign formula above.

This representation is the quantum-mechanical seed of the supersymmetric localisation programme: the index is computed exactly by a finite-dimensional integral over zero modes, with all nonzero modes cancelling between bosons and fermions by the pairing theorem. The same localisation, lifted to field theory, gives the Nicolai map of 08.10.12 and the Parisi-Sourlas dimensional reduction of 08.10.13.

Synthesis. Supersymmetric quantum mechanics puts these results together into a single algebraic machine: from one function $W$ it manufactures two partner Hamiltonians whose spectra are forced to agree except at zero energy, and the foundational reason for that agreement is the factorisation $H_{\pm }=\frac{1}{2}{A}^{†}A,\frac{1}{2}A{A}^{†}$ with the supercharge $A$ intertwining the two sectors. This is exactly the structure that makes the Witten index a deformation invariant, and the central insight is that a graded trace cancels every paired state and registers only the unpaired zero-energy kernel — a quantity that generalises from this two-state model to the Euler characteristic of a manifold in 03.15.12 and to anomaly indices in field theory. The bridge runs in three directions at once: putting these together, the factorisation is dual to the Hodge-theoretic splitting of the de Rham complex, the index is dual to the Poincaré-Hopf count of critical points of $W$, and the drift $-W^{\prime }$ inside $A$ is the very drift that appears again in the Fokker-Planck operator of 08.10.02, carrying the whole construction into stochastic quantisation. The central insight that unifies these faces is that boson-fermion cancellation is a topological statement: it survives any deformation that does not drag spectrum across $E=0$, which is the foundational reason the same index computes shape-invariant spectra, counts Morse critical points, and seeds the supersymmetric-localisation route forward to the Nicolai map 08.10.12 and Parisi-Sourlas reduction 08.10.13.

Full proof set Master

Proposition (positivity of the SUSY Hamiltonian and the vacuum criterion). The Hamiltonian $H=\{Q,Q^{†}\}$ is a positive operator, and a normalisable state $|\Omega ⟩$ has $H|\Omega ⟩=0$ if and only if $Q|\Omega ⟩=Q^{†}|\Omega ⟩=0$.

Proof. For any $|\chi ⟩$ in the domain, $$ \langle \chi | H | \chi \rangle = \langle \chi | QQ^\dagger | \chi \rangle + \langle \chi | Q^\dagger Q | \chi \rangle = |Q^\dagger\chi|^2 + |Q\chi|^2 \geq 0, $$ so $H\ge 0$. If $H|\Omega ⟩=0$ then $⟨\Omega |H|\Omega ⟩=\Vert Q^{†}\Omega {\Vert }^2+\Vert Q\Omega {\Vert }^2=0$, which forces both norms to vanish, hence $Q|\Omega ⟩=Q^{†}|\Omega ⟩=0$. Conversely, if both supercharges annihilate $|\Omega ⟩$, then $H|\Omega ⟩=Q{Q}^{†}|\Omega ⟩+Q^{†}Q|\Omega ⟩=0$. So zero-energy states are exactly the supersymmetric (doubly annihilated) states, and the ground-state energy $E_0\ge 0$ vanishes iff such a state exists. $\square$

Proposition (at most one zero-energy state, and its sector). For a superpotential $W$ with $|W^{\prime }|\to \infty$ at infinity, the kernel of $H$ is at most one-dimensional, occupied by $e^{-W}$ in the bosonic sector when $W\to +\infty$ at both ends, by $e^{+W}$ in the fermionic sector when $W\to -\infty$ at both ends, and empty when $W\to +\infty$ at one end and $-\infty$ at the other does not occur — i.e. when the two limits have the same sign the kernel is one-dimensional and when they differ supersymmetry is broken.

Proof. A zero-energy state satisfies $Q|\Omega ⟩=Q^{†}|\Omega ⟩=0$. In the bosonic sector ${\mathcal{H}}_{+}$, the condition $Q^{†}|\Omega ⟩=0$ reduces to $A{\psi }_{+}=(p-i{W}^{\prime }){\psi }_{+}=0$, solved by ${\psi }_{+}=e^{-W}$ as in Exercise 3. In the fermionic sector ${\mathcal{H}}_{-}$, the condition $Q|\Omega ⟩=0$ reduces to $A^{†}{\psi }_{-}=(p+i{W}^{\prime }){\psi }_{-}=0$, solved by ${\psi }_{-}=e^{+W}$. Normalisability requires $\int e^{-2W}<\infty$ for the bosonic solution (so $W\to +\infty$ both ways) or $\int e^{+2W}<\infty$ for the fermionic one (so $W\to -\infty$ both ways). These conditions are mutually exclusive, so at most one sector hosts a zero-energy state, and the kernel is at most one-dimensional. When $W\to +\infty$ at one end and $-\infty$ at the other, neither integral converges, no normalisable zero-energy state exists, and supersymmetry is spontaneously broken. $\square$

Proposition ($\beta$-independence of the regularised index). The quantity $\Delta (\beta )=\mathrm{tr}(-1{)}^F{e}^{-\beta H}$ is independent of $\beta >0$ and equals the integer $n_B^{E=0}-n_F^{E=0}$.

Proof. Decompose the trace over the eigenbasis of $H$. Zero-energy states contribute $(-1{)}^F\cdot 1$, summing to $n_B^{E=0}-n_F^{E=0}$. For each energy $E>0$, the pairing theorem provides a bijection $|b⟩\leftrightarrow |f⟩=(2E{)}^{-1/2}{Q}^{†}|b⟩$ between bosonic and fermionic eigenstates of the same $E$. Each such pair contributes $e^{-\beta E}(+1)+e^{-\beta E}(-1)=0$ to the trace. Hence all $\beta$-dependent terms cancel and $$ \Delta(\beta) = \sum_{E = 0} (-1)^F = n_B^{E=0} - n_F^{E=0}, $$ independent of $\beta$. The interchange of trace and sum is justified by the trace-class property of $e^{-\beta H}$ for $\beta >0$ when $H$ has compact resolvent (discrete spectrum tending to $+\infty$), which holds under the confining hypothesis $|W^{\prime }|\to \infty$. $\square$

Proposition (oscillator spectrum from the SUSY ladder). For $W=\frac{1}{2}\omega x^2$, the bosonic Hamiltonian $H_{+}=\frac{1}{2}{p}^2+\frac{1}{2}{\omega }^2{x}^2+\frac{1}{2}\omega$ has spectrum $E_n=n\omega$ for $n=0,1,2,\dots$, recovering the quantum harmonic oscillator of 12.04.02 shifted so the ground state sits at zero.

Proof. Here $A=p-i\omega x$ and $A^{†}=p+i\omega x$ satisfy $[A,A^{†}]=[p,i\omega x]-[i\omega x,p]=2\omega$, the oscillator algebra up to scale. So $H_{+}=\frac{1}{2}{A}^{†}A$ has $A^{†}A=A{A}^{†}-2\omega$, giving the ladder structure with spacing $\omega$. The ground state is $A{\psi }_0=0$, i.e. ${\psi }_0=e^{-\omega x^2/2}$, with $H_{+}{\psi }_0=0$. Acting with $A^{†}$ raises the energy by $\omega$ each time, producing $E_n=n\omega$. The partner $H_{-}=\frac{1}{2}A{A}^{†}$ has spectrum $E_n=(n+1)\omega$, the same ladder with the ground state removed, confirming the isospectrality theorem. The conventional oscillator energies $E_n^{\text{osc}}=(n+\frac{1}{2})\omega$ are recovered by undoing the constant shift $+\frac{1}{2}\omega$ built into $V_{+}$. $\square$

Proposition (the index counts critical points of $W$ with sign). For a Morse superpotential $W$ (isolated nondegenerate critical points $x_i$ where $W^{\prime }(x_i)=0$, $W^{\prime \prime }(x_i)\ne 0$) with $W\to +\infty$ at both ends, the Witten index equals the signed count $\Delta ={\sum }_i\frac{1}{2}(1-\mathrm{sgn}{W}^{\prime \prime }(x_i))$ restricted to give the boundary-sign formula, agreeing with the alternating-sum Morse count.

Proof. Scale $W\to sW$ and send $s\to \infty$. The ground-state wavefunction $e^{-sW}$ concentrates near the minima of $W$ (where $W^{\prime \prime }>0$), and the harmonic approximation around each critical point $x_i$ reduces the local problem to a shifted oscillator with frequency $|W^{\prime \prime }(x_i)|$. A minimum ($W^{\prime \prime }>0$) contributes a normalisable bosonic zero mode in the $s\to \infty$ limit; a maximum ($W^{\prime \prime }<0$) contributes a fermionic one. The net graded count is the alternating sum over critical points, which by the fundamental theorem of calculus telescopes to the boundary data $\frac{1}{2}(\mathrm{sgn}{W}^{\prime }(+\infty )-\mathrm{sgn}{W}^{\prime }(-\infty ))$. This is the Poincaré-Hopf / Morse identity in one dimension; the geometric generalisation to a manifold is the content of 03.15.12, where the same alternating sum becomes the Euler characteristic. $\square$

Connections Master

• Quantum harmonic oscillator 12.04.02. The oscillator is the simplest realisation of SUSY-QM: the superpotential $W=\frac{1}{2}\omega x^2$ gives partner potentials $V_{\pm }=\frac{1}{2}{\omega }^2{x}^2\pm \frac{1}{2}\omega$, and the ladder operators $A,A^{†}$ of the factorisation $H_{+}=\frac{1}{2}{A}^{†}A$ are exactly the oscillator creation and annihilation operators. The full oscillator spectrum $E_n=n\omega$ is generated by the SUSY ladder, and shape invariance is the abstract statement of why the oscillator is exactly solvable.

• Fokker-Planck equation and equilibrium distribution 08.10.02. The drift $-W^{\prime }$ inside the supercharge $A=p-i{W}^{\prime }$ is the drift of the overdamped Langevin process $\dot{x}=-W^{\prime }(x)+\eta$. Conjugating $H_{+}$ by $e^{\pm W}$ turns the SUSY-QM Hamiltonian into the Fokker-Planck operator of that process, with the zero-energy ground state $e^{-W}$ mapping to the stationary density. This identification makes the Witten index a stochastic-quantisation observable and is the prerequisite link for the Nicolai-map construction.

• Witten's supersymmetric Morse theory 03.15.12. The Hamiltonian SUSY-QM of this unit is the analytic half of the same algebra whose geometric half is the deformed de Rham complex. There the supercharges become $d_t+d_t^{\ast }$ on a Riemannian manifold, zero-energy states become harmonic forms, and the Witten index becomes the Euler characteristic. The boundary-sign formula for $\Delta$ is the one-dimensional Poincaré-Hopf theorem that, generalised, yields the Morse inequalities.

• The Nicolai map and stochastic quantisation of SUSY theories 08.10.12. SUSY-QM is the hard prerequisite for the Nicolai map: the field-theoretic generalisation in which a nonlinear redefinition with Jacobian equal to the fermion determinant reduces the bosonic action to a Gaussian, identifying the Nicolai variable with the equilibrium variable of a Langevin process whose drift is the superpotential gradient. The factorisation and the Fokker-Planck link established here are exactly what that construction lifts to field theory.

• Parisi-Sourlas dimensional reduction 08.10.13. The supersymmetric structure that organises SUSY-QM reappears in the Parisi-Sourlas reduction of a random-field theory in $d$ dimensions to a supersymmetric theory in $d-2$ dimensions. The noise of the Langevin reading here plays the role of the quenched random source there, and the $\mathrm{OSp}$ superrotation is the field-theoretic descendant of the boson-fermion pairing that controls the Witten index.

Historical & philosophical context Master

Supersymmetric quantum mechanics was introduced by Edward Witten in his 1981 paper Dynamical breaking of supersymmetry (Nucl. Phys. B 188, 513) ^{[Witten-1981]}, not as a subject in its own right but as a controlled laboratory for a hard field-theory question: can supersymmetry, exact in the classical action, be broken by quantum effects? Witten reduced the question to its simplest informative setting — a single bosonic coordinate and a single fermionic mode — where the $\mathcal{N}=2$ algebra $\{Q,Q^{†}\}=H$, $Q^2=0$ can be analysed completely. The criterion that emerged is sharp and geometric: supersymmetry is unbroken precisely when a normalisable zero-energy ground state exists, and the ground-state energy is the order parameter for breaking. The factorisation method that underlies the partner-potential structure was older, going back to Schrödinger's 1940 operator factorisation of the radial Coulomb problem and to Infeld and Hull's 1951 systematic study, but Witten's contribution was to recognise the supersymmetric algebra organising it and to extract a topological invariant from it.

That invariant — the Witten index $\mathrm{tr}(-1{)}^F$ — was developed in the companion 1982 paper Constraints on supersymmetry breaking (Nucl. Phys. B 202, 253) ^{[Witten-1982-constraints]}. The index is a deformation invariant: because nonzero-energy states pair into boson-fermion doublets, the graded trace counts only zero-energy states, and this count cannot change under continuous deformations of the parameters or the volume. A nonzero index is therefore an obstruction to supersymmetry breaking that holds non-perturbatively, surviving the strong-coupling regimes where perturbation theory fails. Cecotti and Girardello in 1982 ^{[Cecotti-Girardello]} recast the index as a regularised functional-integral supertrace with periodic fermion boundary conditions, exhibiting its localisation onto zero modes and its connection to the stochastic functional measure — the bridge that ties SUSY-QM to the Langevin/Fokker-Planck framework and onward to the Nicolai map.

The geometric reading came in Witten's third 1982 paper, Supersymmetry and Morse theory (J. Diff. Geom. 17, 661) ^{[Witten-Morse]}, which turned the same algebra into a proof of the Morse inequalities and made the index the Euler characteristic, a landmark in the cross-pollination of physics and topology. The canonical pedagogical synthesis is the 1995 Physics Reports review by Cooper, Khare, and Sukhatme ^{[Cooper-Khare-Sukhatme]}, which organises the exactly-solvable potentials under shape invariance and remains the standard reference. Philosophically, the unit is a clean case study in how a structure invented to address one problem — dynamical breaking — turns out to encode a robust integer whose stability across deformations reflects a topological truth, a recurring pattern in which physical questions about vacua become mathematical questions about counting.

Bibliography Master

@article{Witten1981,
  author  = {Witten, Edward},
  title   = {Dynamical breaking of supersymmetry},
  journal = {Nuclear Physics B},
  volume  = {188},
  year    = {1981},
  pages   = {513--554}
}

@article{Witten1982constraints,
  author  = {Witten, Edward},
  title   = {Constraints on supersymmetry breaking},
  journal = {Nuclear Physics B},
  volume  = {202},
  year    = {1982},
  pages   = {253--316}
}

@article{Witten1982Morse,
  author  = {Witten, Edward},
  title   = {Supersymmetry and {M}orse theory},
  journal = {Journal of Differential Geometry},
  volume  = {17},
  year    = {1982},
  pages   = {661--692}
}

@article{CooperKhareSukhatme1995,
  author  = {Cooper, Fred and Khare, Avinash and Sukhatme, Uday},
  title   = {Supersymmetry and quantum mechanics},
  journal = {Physics Reports},
  volume  = {251},
  number  = {5-6},
  year    = {1995},
  pages   = {267--385}
}

@book{CooperKhareSukhatme2001,
  author    = {Cooper, Fred and Khare, Avinash and Sukhatme, Uday},
  title     = {Supersymmetry in Quantum Mechanics},
  publisher = {World Scientific},
  year      = {2001}
}

@article{CecottiGirardello1982,
  author  = {Cecotti, Sergio and Girardello, Luciano},
  title   = {Stochastic and parastochastic aspects of supersymmetric functional measures},
  journal = {Physics Letters B},
  volume  = {110},
  year    = {1982},
  pages   = {39--43}
}

@book{Junker1996,
  author    = {Junker, Georg},
  title     = {Supersymmetric Methods in Quantum and Statistical Physics},
  publisher = {Springer-Verlag},
  series    = {Texts and Monographs in Physics},
  year      = {1996}
}

@article{InfeldHull1951,
  author  = {Infeld, Leopold and Hull, T. E.},
  title   = {The factorization method},
  journal = {Reviews of Modern Physics},
  volume  = {23},
  year    = {1951},
  pages   = {21--68}
}

@article{Nicolai1980,
  author  = {Nicolai, Hermann},
  title   = {Supersymmetry and functional integration measures},
  journal = {Nuclear Physics B},
  volume  = {176},
  year    = {1980},
  pages   = {419--428}
}