Quantum Simulation in Chemistry and Materials

Introduction and Context

Chemistry and materials science are fundamentally concerned with the behaviour of electrons and atomic nuclei. Chemical bonds, molecular reactions, catalytic activity, conductivity, magnetism, corrosion and battery performance all depend on interactions that are quantum mechanical in origin. Classical computational chemistry has achieved remarkable success, but its cost can increase rapidly when many interacting electrons must be represented accurately.

Quantum simulation uses controllable quantum systems to represent and solve models of other quantum systems. In quantum computing, the electronic structure of a molecule or material is mapped onto qubits. A quantum processor then estimates properties such as ground-state energy, excitation energy or reaction energetics, while a classical computer manages data preparation, parameter optimization and interpretation.

The objective is not to replace all classical chemistry. Rather, quantum computing is most promising for problems in which electron correlation, large configuration spaces or difficult reaction mechanisms limit conventional approximations. At diploma level, it is essential to understand both the opportunity and the engineering constraints: a chemically meaningful answer requires accurate problem formulation, suitable basis sets, efficient encoding, reliable measurements and careful validation.

Learning Objectives

By the end of this chapter, a learner should be able to:

1. Explain how molecular and materials problems are represented as quantum Hamiltonians.
2. Distinguish the roles of Hartree–Fock, density functional theory, coupled-cluster methods and quantum algorithms.
3. Describe the workflow of variational quantum eigensolvers and related simulation methods.
4. Apply energy differences to reaction pathways, catalysis, batteries and pharmaceutical screening.
5. Evaluate data-preparation, measurement, noise, scaling and hardware constraints.
6. Interpret quantum-computing results using chemical accuracy, uncertainty and classical validation.

1. The Electronic-Structure Problem

In the Born–Oppenheimer approximation, nuclei are treated as fixed while the electronic problem is solved. This approximation separates the rapid motion of electrons from the much slower nuclear motion. For a specified nuclear geometry, the electronic Hamiltonian may be written conceptually as

Here, Ĥ_e is the electronic Hamiltonian in atomic units, i and j index electrons, A indexes nuclei, ∇_i² represents the kinetic-energy operator of electron i, Z_A is the charge of nucleus A, r_iA is the electron–nucleus distance, r_ij is the electron–electron distance, and E_NN is the classical nuclear–nuclear repulsion energy. Atomic units set ℏ = m_e = e = 4πε₀ = 1.

The time-independent Schrödinger equation is

where |Ψ⟩ is an electronic wavefunction and E is an allowed energy. The ground-state energy is the smallest eigenvalue. Exact solution becomes difficult because the wavefunction depends on the coordinates of all electrons simultaneously. If a one-electron basis has M spin-orbitals and there are N electrons, the number of possible occupation configurations is combinatorial; for a fixed number of electrons it is proportional to C(M,N) = M!/[N!(M−N)!]. This growth motivates quantum representations.

From orbitals to qubits

A common second-quantized Hamiltonian is

The indices label spin-orbitals; a†_p and a_p are fermionic creation and annihilation operators; h_pq are one-electron integrals; and h_pqrs are two-electron integrals. The factor 1/2 prevents double counting under the chosen integral convention. Integrals are determined by the basis functions and nuclear geometry.

Fermionic operators must obey anticommutation relations. Qubit operators, in contrast, act on tensor products of two-level systems. Encodings such as Jordan–Wigner and Bravyi–Kitaev transform fermionic operators into sums of Pauli operators:

where c_k are real coefficients and P_k are tensor products of I, X, Y and Z. For example, X ⊗ Z ⊗ I is a Pauli string acting on three qubits. The number of qubits is usually related to the number of spin-orbitals, although symmetry reduction and active-space selection can reduce it.

2. Classical Baselines and Quantum Advantage

Quantum methods should be compared with appropriate classical baselines rather than with an idealized notion of “classical chemistry.” Important methods include:

A quantum computer is not automatically superior. A quantum calculation is valuable only when its total error and resource cost are competitive with a validated classical method. Hybrid workflows are therefore central: classical chemistry software prepares integrals and initial states, the quantum processor evaluates expectation values, and classical optimization updates parameters.

3. Variational Quantum Eigensolver

The variational quantum eigensolver (VQE) is a hybrid method designed for estimating ground-state energies on near-term quantum hardware. A parameterized circuit prepares

where |ψ₀⟩ is an initial state, U(θ) is a quantum circuit and θ is a vector of adjustable parameters. The energy objective is

The variational principle states that

for every normalized trial state, where E₀ is the exact ground-state energy of the represented Hamiltonian. Equality is reached only if the trial state has the ground-state component needed to achieve the minimum. This inequality assumes a Hermitian Hamiltonian and a normalized state; it does not guarantee that an optimizer finds the global minimum.

Each Pauli expectation value is estimated from repeated measurements. If a Pauli observable has outcomes ±1 and N independent shots, the sample mean ⟨P_k⟩ has standard error approximately

This expression assumes independent measurements and an ideal binary measurement. More generally, uncertainty also includes state-preparation, readout and calibration errors. Since many Pauli terms may occur in the Hamiltonian, measurement grouping and adaptive shot allocation are important.

A typical VQE cycle is:

1. Select a molecule, geometry, basis set and active space.
2. Compute one- and two-electron integrals classically.
3. Map the fermionic Hamiltonian to qubit operators.
4. Prepare a physically motivated initial state, often a Hartree–Fock state.
5. Measure Pauli expectations and calculate energy.
6. Use a classical optimizer to update circuit parameters.
7. Repeat until the energy and parameters meet stopping criteria.
8. Validate against classical calculations, symmetry checks and uncertainty estimates.

Ansatz design

An ansatz is the family of states accessible to the circuit. Hardware-efficient ansätze use native gates and shallow layers, but can suffer from poor chemical interpretability or barren plateaus, in which gradients become very small. Chemistry-inspired ansätze, such as unitary coupled-cluster-style circuits, encode excitation structure but may require deeper circuits.

The ansatz must balance expressibility and trainability. An ansatz that is too restricted produces bias because the ground state is inaccessible. One that is excessively deep increases gate errors, optimization difficulty and circuit execution time.

4. Other Quantum-Simulation Approaches

Quantum phase estimation

Quantum phase estimation (QPE) can estimate eigenvalues with high precision when a sufficiently accurate eigenstate is available. It uses controlled time evolution or controlled unitary operations and requires fault-tolerant quantum hardware for demanding chemistry problems. QPE is theoretically powerful but generally beyond the capabilities of noisy near-term devices.

Quantum imaginary-time and real-time simulation

Imaginary-time methods suppress higher-energy components and can prepare ground states under suitable conditions. Real-time simulation applies unitary evolution to study dynamics, spectra and reaction processes. These methods are relevant to photochemistry and excited-state phenomena, but state preparation, long coherent circuits and measurement remain challenging.

Quantum machine learning and generative methods

Quantum machine-learning models may assist property prediction, feature reduction or sampling. However, the input-data bottleneck is important: encoding classical molecular data into quantum states can require substantial circuit depth. A quantum model is not useful merely because it contains qubits; it must improve accuracy, cost, interpretability or scientific insight over a classical model.

5. Geometry, Potential-Energy Surfaces and Reaction Pathways

A molecule’s energy depends on nuclear coordinates. The potential-energy surface (PES) is the function

for a nonlinear molecule with N nuclei after removing three translational and three rotational degrees of freedom. For a linear molecule, the number is 3N−5. These counts assume an isolated molecule and exclude special constraints or external fields.

A reaction pathway is a sequence of geometries connecting reactants and products. A transition state is a first-order saddle point on the PES: it has one imaginary vibrational frequency in a harmonic analysis and corresponds to a maximum along the reaction coordinate but minima in perpendicular directions.

The reaction energy is

and the activation energy from a reactant minimum is

These are electronic energy differences at a specified level of theory. Experimental thermochemistry may require zero-point energy, thermal, entropic and solvation corrections. At temperature T, a free-energy difference is commonly represented as ΔG = ΔH − TΔS, where ΔH is enthalpy change and ΔS is entropy change. Energies must be compared using the same geometry conventions, basis set, charge, spin state and environmental model.

Quantum simulation can evaluate energies at multiple geometries, while a classical optimizer searches the PES. In practice, a quantum processor may be used only for the difficult electronic component, with geometry optimization and pathway algorithms remaining classical.

6. Catalysis and Reaction Mechanisms

Catalysts alter reaction rates by providing a pathway with a lower activation barrier; they do not change the overall thermodynamic equilibrium between reactants and products. Quantum simulation can support catalyst discovery by estimating:

• adsorption energies of reactants and intermediates;
• bond-breaking and bond-forming energies;
• oxidation and spin-state energetics;
• transition-state barriers;
• selectivity between competing pathways; and
• effects of ligand, surface, solvent and electric-field changes.

For heterogeneous catalysis, a finite cluster or periodic model may represent an active surface site. The model must be large enough to capture relevant locality and charge effects, but small enough for computation. For transition-metal catalysts, near-degenerate orbitals and multiple spin states create strong-correlation problems that can challenge standard DFT and small quantum devices.

A practical workflow is to screen many candidates with inexpensive classical methods, select a smaller set of difficult or high-value cases, and apply quantum simulation to active spaces. Results should be benchmarked against larger classical calculations wherever possible.

7. Batteries and Energy Materials

Battery performance depends on electronic structure, ionic transport, phase stability, interfaces and degradation. Quantum simulation may contribute to:

• voltage estimation from formation energies;
• redox energetics;
• lithium, sodium or proton insertion;
• defect and vacancy formation;
• electrode–electrolyte reactions;
• solid-electrolyte stability; and
• magnetic and electronic properties of electrode materials.

A simplified average cell voltage can be related to a Gibbs free-energy change by

where V is voltage in volts, ΔG is the reaction Gibbs-energy change in joules per mole, n is the number of electrons transferred per formula unit, and F is the Faraday constant, approximately 96485 C mol⁻¹. The relation assumes a reversible cell and consistent molar stoichiometry. At the electronic-energy level, replacing ΔG with ΔE neglects thermal, entropic and solvation effects and therefore gives only an approximation.

For example, if a reversible reaction has ΔG = −193000 J mol⁻¹ and transfers n = 2 electrons, then

This calculation is chemically meaningful only if the reaction equation and electron count are correctly balanced.

8. Pharmaceuticals and Molecular Design

Drug discovery involves molecular recognition, conformational flexibility, solvation, protonation and sometimes covalent reaction mechanisms. Quantum simulation may help estimate:

• relative energies of molecular conformers;
• binding-relevant electronic properties;
• proton-transfer barriers;
• reaction mechanisms in enzymatic catalysis;
• properties of fragments and active pharmaceutical ingredients; and
• descriptors used in hybrid quantum–classical machine learning.

A major challenge is that biologically relevant binding free energies include solvent, entropy and protein dynamics. A quantum calculation of an isolated ligand’s ground-state energy is not equivalent to a binding-affinity prediction. Proper workflows combine molecular mechanics, molecular dynamics, continuum or explicit solvent models and quantum calculations on selected active regions.

Data preparation is especially important. Molecular structures must be standardized for atom ordering, charge, spin multiplicity, tautomeric state, stereochemistry and protonation. Incorrect chemical identity can produce a numerically precise but scientifically meaningless result.

9. Materials Discovery

Materials problems may involve periodic boundary conditions, band structures, defects, surfaces, phonons and strongly correlated electrons. Molecular qubit mappings designed for finite molecules do not transfer automatically to infinite crystals. Periodic materials may require k-point sampling, embedding, fragment methods or specialized quantum algorithms.

Useful target properties include band gaps, magnetic exchange, dielectric response, adsorption, defect levels and phase stability. A material’s performance is multi-objective: low energy alone may conflict with toxicity, scarcity, manufacturability, stability or environmental impact.

A discovery pipeline should therefore include:

1. Define the target property and acceptable uncertainty.
2. Assemble verified structures and metadata.
3. Use classical screening to remove implausible candidates.
4. Select active spaces and quantum-relevant subproblems.
5. Estimate properties with uncertainty and compare baselines.
6. Test promising materials experimentally or with higher-fidelity simulation.
7. Record provenance so the result can be reproduced.

10. Data Preparation and Hardware Constraints

Quantum chemistry begins with classical data. Typical inputs include molecular geometry, atomic numbers, basis functions, electron count, charge, spin multiplicity and environmental assumptions. Errors in any of these can change the Hamiltonian.

Active-space selection

An active space retains orbitals expected to contribute most strongly to the property of interest. Core orbitals may be frozen, while orbitals far from the frontier may be omitted. Active-space reduction lowers qubit and measurement requirements but introduces model error. The selected space should be justified using orbital energies, occupations, symmetry, chemical intuition and convergence tests.

Noise and error sources

Current processors experience gate errors, decoherence, crosstalk, leakage and readout errors. Chemical energies are often differences between large quantities, so small systematic errors can obscure the desired signal. Error mitigation methods include measurement-error calibration, zero-noise extrapolation, symmetry verification and probabilistic error cancellation. These methods can reduce bias but generally increase sampling cost and do not replace fault-tolerant error correction.

Resource measures

Important resources include:

• number of logical and physical qubits;
• circuit depth and two-qubit-gate count;
• number of Hamiltonian terms;
• measurement shots;
• classical optimization iterations;
• wall-clock execution time; and
• uncertainty in the final observable.

A small qubit count is not sufficient evidence of feasibility if the circuit is too deep or requires millions of measurements. Conversely, a larger fault-tolerant algorithm may eventually be preferable if it provides predictable accuracy.

Chemical accuracy

The phrase “chemical accuracy” is often used informally for an energy error near 1 kcal mol⁻¹, approximately 0.043 eV per molecule. It is not a universal guarantee: the acceptable tolerance depends on the reaction, property and comparison. For a reaction energy, errors in several component energies can accumulate or cancel. Reports should state absolute errors, relative errors, statistical uncertainty and systematic uncertainty separately.

11. Worked Example: Measuring a Two-Term Hamiltonian

Consider the illustrative Hamiltonian

The coefficients are in an arbitrary consistent energy unit. Suppose a prepared state produces measured expectation values ⟨Z₀⟩ = −0.8, ⟨Z₁⟩ = 0.5 and ⟨Z₀Z₁⟩ = −0.4. Then

The result is −0.25 energy units. This is an expectation value for the prepared state, not necessarily the ground-state energy. If the state is normalized and the Hamiltonian is Hermitian, it cannot lie below the exact ground-state eigenvalue.

If the three expectation values are estimated independently, uncertainty propagation for a linear combination can be approximated by

where σ₀, σ₁ and σ₀₁ are standard errors. Independence is an assumption; correlations require covariance terms.

12. Illustrative Case Study: Quantum-Assisted Catalyst Screening

Background

This fictional case concerns a research team investigating catalysts for electrochemical carbon-dioxide reduction. The team wants to compare two transition-metal complexes whose activity may depend on competing spin states and a proton-coupled electron-transfer step.

Problem

DFT calculations disagree on the relative energies of two intermediates. The predicted activation barriers differ by approximately 0.08 eV, which changes the ranking of the catalysts. The complexes contain many orbitals, but only a metal d-shell, ligand π orbitals and the transferring proton are expected to dominate the disputed step.

Analysis

The team first standardizes geometries, charge and spin multiplicities. It performs classical calculations with several functionals and identifies an active space containing 12 spin-orbitals. A fermionic Hamiltonian is generated and reduced using particle-number and spin symmetries. The resulting quantum model uses fewer qubits than the unreduced representation.

A Hartree–Fock state initializes a chemistry-inspired VQE ansatz. Measurements are grouped by compatible Pauli operators. Readout calibration and symmetry checks are applied. The team runs several optimizers from different initial parameters to identify failures caused by local minima. For each intermediate, it reports the mean energy, measurement uncertainty and variation across independent runs.

Approach and solution

The quantum calculation is not used to model the entire catalytic cycle. It is used as a targeted benchmark for the disputed active-space energy difference. Geometry optimization, solvent corrections and most screening remain classical. The calculated barrier is combined with a separately estimated solvation correction, with the combination clearly labelled as approximate.

Outcome

The quantum-assisted estimate supports the ordering predicted by two of the classical methods and indicates that the alternative spin state is important. The result does not establish experimental catalytic performance, because surface effects, solvent dynamics and mass transport are not represented. It does, however, identify which catalyst deserves more detailed classical and experimental study.

Learning points

1. Quantum simulation can be most useful as a focused correction or benchmark rather than a complete replacement for a workflow.
2. Active-space choices are scientific assumptions, not merely software settings.
3. Energy differences require consistent treatment of every intermediate.
4. Uncertainty and validation are part of the result.
5. A computational ranking is not the same as demonstrated laboratory performance.

13. Professional, Ethical and Governance Considerations

Quantum chemistry tools may influence pharmaceutical, energy and environmental decisions. Researchers should document data provenance, software versions, Hamiltonian construction, basis sets, circuit parameters, random seeds where relevant and hardware calibration conditions. Claims should distinguish simulated evidence from experimental evidence.

Reproducibility requires publishing input structures, charge and spin states, active spaces, measurement budgets and uncertainty estimates. Proprietary molecular or materials data raise confidentiality and intellectual-property issues. In pharmaceuticals, a computational prediction must not be represented as clinical efficacy or safety evidence. In energy technologies, sustainability claims should include mining, manufacturing, toxicity and lifecycle considerations rather than computational performance alone.

AI-assisted quantum workflows introduce additional risks: training-data leakage, hidden dataset bias, non-reproducible preprocessing and overconfident predictions. Human chemical expertise remains necessary for selecting meaningful targets and detecting chemically impossible outputs.

14. Contemporary Developments and Best Practices

Current development focuses on improved error mitigation, symmetry-preserving circuits, adaptive ansätze, low-rank Hamiltonian factorization, qubit tapering, embedding methods and fault-tolerant algorithms. Hardware progress is measured not only by qubit count but also by fidelity, connectivity, coherence, calibration stability and the ability to execute useful circuits.

Best practices for diploma-level projects are:

• begin with a small molecule or reduced material model;
• define the property before selecting the algorithm;
• establish a classical reference calculation;
• verify charge, spin, geometry and units;
• reduce the active space transparently;
• compare several initial states or optimizers;
• report shot counts and uncertainty;
• test noise-free and noisy simulations separately;
• avoid interpreting a variational energy as a complete chemical explanation; and
• preserve all input and output data for reproducibility.

Chapter Summary

Quantum simulation maps molecular or materials Hamiltonians onto qubits and estimates observables using quantum circuits. The electronic-structure problem is expressed through one- and two-electron integrals, fermionic operators and Pauli strings. VQE provides a practical hybrid approach: a parameterized state is measured, its energy is evaluated and a classical optimizer updates the circuit. The variational principle makes the expectation value an upper bound to the ground-state energy under its stated assumptions, but optimization and hardware errors remain important.

Energy differences support the study of reaction pathways, catalysts, batteries, pharmaceuticals and materials. Reaction barriers, voltage estimates and binding-related properties require consistent geometries, charge states, environmental corrections and uncertainty analysis. Active-space selection can reduce resources but introduces approximation. Present hardware is constrained by noise, circuit depth, measurement cost and limited qubit connectivity. Consequently, the strongest near-term strategy is usually a validated hybrid workflow in which quantum computation targets a difficult electronic subproblem and classical methods provide broader modelling, optimization and validation.

Key Terms

Born–Oppenheimer approximation: Separation of electronic and nuclear motion by treating nuclei as fixed during the electronic calculation.

Electronic Hamiltonian: Operator representing electronic kinetic energy, electron–nucleus attraction, electron–electron repulsion and nuclear repulsion.

Fermionic operator: Creation or annihilation operator obeying anticommutation relations for electrons.

Spin-orbital: A spatial orbital combined with a specified spin state.

Pauli string: Tensor product of single-qubit Pauli operators used to represent a qubit Hamiltonian.

Jordan–Wigner encoding: A mapping from fermionic operators to qubit operators using parity strings.

Active space: Selected orbitals retained explicitly in a reduced electronic-structure model.

VQE: Hybrid algorithm that minimizes a measured energy expectation value using a parameterized quantum circuit.

Ansatz: Parameterized family of quantum states prepared by a circuit.

Variational principle: Principle stating that a normalized trial-state energy is not below the exact ground-state energy.

Expectation value: Average result obtained by repeatedly measuring an observable.

Chemical accuracy: Informal target often associated with an energy error near 1 kcal mol⁻¹.

Potential-energy surface: Energy as a function of nuclear coordinates.

Transition state: First-order saddle point associated with a reaction barrier.

Activation energy: Energy difference between a transition state and a reactant reference.

Catalyst: Substance or system that changes reaction kinetics without being consumed overall.

Redox energy: Energy associated with oxidation–reduction and electron transfer.

Error mitigation: Techniques that reduce estimated noise effects without fully correcting the quantum hardware.

Shot: One repeated circuit execution used to estimate a measurement probability or expectation value.

Symmetry verification: Checking measured results against conserved quantities such as particle number or spin symmetry.

Review Questions

5 Short Answer Questions

1. What does the Born–Oppenheimer approximation assume?
2. Why are fermionic operators mapped to qubit operators in quantum simulation?
3. State the variational principle used by VQE.
4. What is an active space?
5. Name two hardware-related sources of error in quantum chemistry calculations.

5 Long Answer Questions

1. Explain the complete workflow for applying VQE to a molecular ground-state problem.
2. Compare Hartree–Fock, DFT, coupled-cluster methods and quantum algorithms in terms of strengths and limitations.
3. Explain how quantum simulation can support catalyst discovery while remaining part of a hybrid workflow.
4. Discuss why battery voltage cannot generally be inferred accurately from electronic energy alone.
5. Explain the effect of ansatz choice, measurement cost and noise on the reliability of a quantum chemistry calculation.

3 Analytical / Application Questions

1. A research team obtains different reaction barriers from two active spaces. Design a validation plan that determines whether the difference is caused by active-space truncation, measurement uncertainty or hardware noise.
2. Propose a quantum-assisted workflow for screening electrode materials, identifying the classical and quantum stages and the data that must be recorded for reproducibility.
3. A molecule has a calculated quantum energy lower than a trusted exact small-system benchmark. Diagnose possible causes and recommend checks before accepting the result.

Multiple Choice Questions with Answers

Question 1

What quantity does the variational quantum eigensolver primarily minimize for a ground-state calculation?

A. The circuit depth only

B. The expectation value of the Hamiltonian

C. The number of nuclei

D. The basis-set cardinality

Correct Answer: B. The expectation value of the Hamiltonian

Question 2

In the electronic Hamiltonian, what does the term involving 1/r_{ij} represent?

A. Electron kinetic energy

B. Nuclear repulsion

C. Electron–electron repulsion

D. Electron–nucleus attraction

Correct Answer: C. Electron–electron repulsion

Question 3

If a reversible battery reaction has ΔG = −193000 J mol⁻¹ and transfers two electrons, what is the approximate voltage?

A. 0.50 V

B. 1.00 V

C. 2.00 V

D. 193000 V

Correct Answer: B. 1.00 V

Question 4

Which statement about an active space is correct?

A. It includes every orbital without approximation

B. It selects orbitals treated explicitly in a reduced model

C. It refers only to the physical qubit layout

D. It eliminates the need for validation

Correct Answer: B. It selects orbitals treated explicitly in a reduced model

Question 5

Why are reaction energies often more useful than isolated absolute energies?

A. They never contain modelling error

B. They compare chemically relevant states, although errors may still accumulate or cancel

C. They do not require a Hamiltonian

D. They are independent of geometry

Correct Answer: B. They compare chemically relevant states, although errors may still accumulate or cancel

Question 6

Which feature characterizes a transition state in a harmonic vibrational analysis?

A. No vibrational modes

B. Exactly one imaginary frequency for a first-order saddle point

C. All frequencies imaginary

D. A zero electronic energy

Correct Answer: B. Exactly one imaginary frequency for a first-order saddle point

Question 7

What is a principal limitation of a hardware-efficient ansatz?

A. It cannot contain adjustable parameters

B. It may be difficult to interpret chemically and may suffer optimization problems

C. It always requires fault-tolerant hardware

D. It cannot prepare entangled states

Correct Answer: B. It may be difficult to interpret chemically and may suffer optimization problems

Question 8

For a Pauli observable with outcomes ±1, increasing the number of measurement shots generally does what?

A. Increases sampling uncertainty

B. Reduces sampling uncertainty approximately with the square root of shot count

C. Changes the molecular Hamiltonian

D. Removes all systematic hardware errors

Correct Answer: B. Reduces sampling uncertainty approximately with the square root of shot count

Question 9

Why is a quantum prediction of an isolated ligand energy not by itself a binding-affinity prediction?

A. Binding affinity involves solvent, entropy, protein dynamics and other effects

B. Ligands have no quantum states

C. Quantum computers cannot represent electrons

D. Binding affinity is determined only by nuclear mass

Correct Answer: A. Binding affinity involves solvent, entropy, protein dynamics and other effects

Question 10

Which practice most directly improves reproducibility of a quantum chemistry study?

A. Reporting only the final energy

B. Hiding the active-space choice

C. Recording geometry, charge, spin, Hamiltonian construction, shots and uncertainty

D. Using a different basis set for every intermediate

Correct Answer: C. Recording geometry, charge, spin, Hamiltonian construction, shots and uncertainty