Self-Assessment Quiz: Quantum Simulation

Twenty questions on Hamiltonian simulation, Trotterization, encodings, and why simulation is the most defensible quantum application. Aim for 16+.


Question 1

Feynman's original motivation for quantum computers was:

A) Breaking cryptography B) Simulating quantum systems, which costs classical resources exponential in particle count C) Faster arithmetic D) Random number generation

Question 2

Simulating $n$ interacting spin-½ particles classically requires storing:

A) $n$ numbers B) $n^2$ C) $2^n$ complex amplitudes D) $n!$

Question 3

Quantum simulation is considered the most defensible application because:

A) It is easiest to implement B) The classical baseline is genuinely exponential, with no structure to exploit C) It needs fewest qubits D) It requires no error correction

Question 4

Time evolution under a Hamiltonian $H$ is:

A) $e^{-iHt}$ B) $Ht$ C) $H^t$ D) $e^{Ht}$

Question 5

Trotterization approximates $e^{-i(A+B)t}$ by:

A) $e^{-iAt}e^{-iBt}$ repeated in small steps B) $e^{-iABt}$ C) $A + B$ D) $e^{-iAt} + e^{-iBt}$

Question 6

The first-order Trotter error per step scales as:

A) $O(t)$ B) $O(t^2[A,B])$ C) $O(1)$ D) Zero

Question 7

Trotter error vanishes when:

A) $A$ and $B$ commute B) $t$ is large C) Never D) The system is classical

Question 8

Second-order (symmetric) Trotter formulas improve error to:

A) $O(t^2)$ B) $O(t^3)$ per step C) $O(t)$ D) No improvement

Question 9

The Jordan–Wigner transformation maps:

A) Fermionic operators to qubit operators B) Qubits to bits C) Bosons to fermions D) Hamiltonians to circuits

Question 10

A drawback of Jordan–Wigner is:

A) It is not exact B) It introduces long strings of $Z$ operators, raising gate cost C) It requires more qubits than orbitals D) It only works for 1D systems

Question 11

The Bravyi–Kitaev encoding improves on Jordan–Wigner by:

A) Reducing operator weight to $O(\log n)$ B) Using fewer qubits C) Eliminating the need for simulation D) Being classical

Question 12

Qubitization and LCU methods offer:

A) Better asymptotic scaling than Trotterization for many Hamiltonians B) Fewer qubits always C) Classical speedup D) Exact simulation with one gate

Question 13

A quantum simulation's output is typically:

A) The full wavefunction B) Expectation values of observables, or an energy via QPE C) All amplitudes D) A classical trajectory

Question 14

Analog quantum simulation differs from digital in that it:

A) Engineers a physical system to mimic the target Hamiltonian directly, without gates B) Uses analog electronics C) Is always less accurate D) Requires error correction

Question 15

The most cited near-term target for quantum simulation is:

A) Weather forecasting B) Molecular electronic structure and correlated materials C) Financial modelling D) Video rendering

Question 16

True or false: Quantum simulation gives exponential speedup for all physical systems.

Question 17

True or false: Classical methods like DMRG and quantum Monte Carlo are already exact for many systems.

Question 18

True or false: A quantum simulator gives you the full wavefunction to inspect.

Question 19

Short answer. Explain why simulation avoids the data-loading bottleneck that defeats many other proposed quantum speedups.

Question 20

Short answer. Why is the fermionic sign problem the specific reason quantum Monte Carlo fails where quantum simulation might succeed?


Answer Key

Q Ans Note
1 B The 1981 lecture; the founding argument for the field.
2 C The exponential wall.
3 B Unlike optimization or machine learning, there is no clever classical algorithm exploiting structure — the difficulty is intrinsic.
4 A The Schrödinger evolution operator.
5 A Split into small steps, alternating the pieces.
6 B Error is governed by the commutator; non-commuting terms are the source.
7 A Commuting terms factor exactly.
8 B Symmetric (Strang) splitting cancels the leading error term.
9 A Maps creation/annihilation operators to Pauli strings, handling fermionic antisymmetry.
10 B The $Z$-strings can be $O(n)$ long, making each term expensive.
11 A Logarithmic operator weight, at the cost of a less intuitive mapping.
12 A Qubitization achieves optimal scaling in several parameters and is standard in modern resource estimates.
13 B You measure observables; you never extract the state.
14 A Cold atoms in optical lattices are the canonical example — no gates, no error correction, but limited programmability.
15 B Chemistry and materials, where correlated electrons defeat classical methods.
16 False Many systems are efficiently simulable classically — weakly correlated, low-entanglement, sign-problem-free. The advantage is specific to strongly correlated, high-entanglement, sign-problem-afflicted systems.
17 True For 1D and weakly entangled systems DMRG is essentially exact, and sign-problem-free QMC handles many others. Quantum advantage lives in the complement of those classes, which must be argued rather than assumed.
18 False You get measurement statistics like any other quantum computation. Extracting all $2^n$ amplitudes needs exponentially many measurements.
19 The input to a simulation is a Hamiltonian — a compact description with $O(n^4)$ terms for chemistry — not an exponentially large data set. The state is generated by evolving under that Hamiltonian, so no $2^n$-amplitude loading is required. Compare quantum machine learning on classical data, where the exponential input must somehow be loaded and the loading dominates. Simulation is native: quantum input, quantum processing, classical output of modest size.
20 Quantum Monte Carlo samples configurations weighted by their contribution. For fermionic systems, antisymmetry makes those weights alternate in sign, so contributions cancel and the statistical variance grows exponentially with system size and inverse temperature — the sign problem. It is not an artifact of any particular algorithm; it is NP-hard in general. A quantum computer represents the antisymmetric state directly in its amplitudes, with cancellation handled by interference rather than by sampling, so the sign problem simply does not arise. This is the most concrete statement of where quantum simulation's advantage comes from.