Source-linked AI summary
Automated Search for new Quantum Experiments
Mario Krenn, Mehul Malik, Robert Fickler, Radek Lapkiewicz, Anton Zeilinger
TL;DR
The paper examines how to construct high-dimensional multipartite quantum states and transformations experimentally. It presents configurations for such states, including a three-dimensional GHZ state, while accounting for post-selection and higher-order emission terms.
Problem
Constructing high-dimensional multipartite entangled states requires experimental configurations beyond straightforward low-dimensional generalizations.
Method
The approach uses experimental configurations involving SPDC sources, optical elements, triggers, and post-selected fourfold coincidences, with experiments simplified after discovery.
Results
The configurations include a state transformable by local unitaries into a three-dimensional GHZ state and an experimentally performed cyclic OAM rotation.
Takeaways & Limitations
The reported configurations provide experimental implementations for high-dimensional entangled states and cyclic quantum transformations.
Takeaways & Limitations
The analysis neglects double emissions and double photons because fourfold coincidence detection is expected to filter them, and higher-order terms are checked only through DC=25.
Abstract
from arXiv · showhide
Quantum mechanics predicts a number of at first sight counterintuitive phenomena. It is therefore a question whether our intuition is the best way to find new experiments. Here we report the development of the computer algorithm Melvin which is able to find new experimental implementations for the creation and manipulation of complex quantum states. And indeed, the discovered experiments extensively use unfamiliar and asymmetric techniques which are challenging to understand intuitively. The results range from the first implementation of a high-dimensional Greenberger-Horne-Zeilinger (GHZ) state, to a vast variety of experiments for asymmetrically entangled quantum states -- a feature that can only exist when both the number of involved parties and dimensions is larger than 2. Additionally, new types of high-dimensional transformations are found that perform cyclic operations. Melvin autonomously learns from solutions for simpler systems, which significantly speeds up the discovery rate of more complex experiments. The ability to automate the design of a quantum experiment can be applied to many quantum systems and allows the physical realization of quantum states previously thought of only on paper.
Supplementary Informations
The supplementary material describes high-dimensional GHZ entanglement and the double-SPDC input used to generate experimental states.
- A 3-partite GHZ state can be generalized to three dimensions using local transformations that preserve entanglement properties.
- The example begins with a double-emission from SPDC, a photon-pair source, and post-selects fourfold coincidences.The state uses photon paths A, B, C and D and a normalization constant N.
- The SPDC input contains paired photons with opposite OAM values across two source paths.
- The experiments consider SPDC orders DC=1 through DC=3 and verify that higher-order terms up to DC=25 do not modify the post-selected output.
- The toolbox descriptions use paths, OAM, and polarization as symbolic degrees of freedom for experimental transformations.
Polarizing beam splitter
The section presents symbolic optical transformations and a catalog of found three-partite entangled-state implementations, including their triggers and resulting states.
- Polarizing beam splitter: A polarizing beam splitter applies path- and polarization-dependent substitutions, including OAM sign reversal for vertically polarized photons.
- Experimental implementations: The catalog records Schmidt-Rank Vectors, experimental setups, triggers in path A, and resulting states for the generated three-partite states.
- Experimental implementations: The catalog includes examples with Schmidt-Rank Vectors such as (10,5,5), (10,6,6), (10,7,5), and (10,9,2).
S5) Example: 3-dimensional GHZ-state (SRV=(3,3,3))
The paper constructs a 3-dimensional GHZ state by combining SPDC photons with OAM-dependent optical operations, interference control, and triggering.
- The target setup uses an OAM-Parity sorter, a mirror, a +2 hologram, and a beam splitter.Normalization constants are omitted for simplicity.
- A trigger in A initially yields SRV (3,3,2), because photon B is perfectly anticorrelated with the 2-dimensional trigger.
- Mixing the trigger with photon C and shifting photon A by -2 prevents unwanted Hong-Ou-Mandel interference and removes the intended state term.
- The mirror protects the construction from higher-order SPDC terms that would otherwise reduce the state to a 2-dimensional GHZ state.
- The triggered state satisfies the high-dimensional GHZ criterion and is locally transformable to |0, 0, 0⟩+ |1, 1, 1⟩+ |2, 2, 2⟩.
6) Cyclic rotations in a high-dimensional space
The section lists cyclic rotations discovered by the algorithm across combinations of OAM, polarization, and path degrees of freedom.
- The listed experiments use OAM alone, OAM with polarization, or OAM with polarization and path.
4-cyclic OAM rotation
The section presents an experimental configuration for a 4-cyclic OAM rotation and notes that the experiment was performed in the authors’ laboratories.
- The section identifies an experimental configuration for the 4-cyclic OAM rotation.
- The experiment was performed in the authors’ laboratories.
3-cyclic OAM+Polarisation rotation
The section describes a 3-cyclic rotation combining OAM and polarisation through a sequence of optical elements.
- The section introduces an experimental configuration for the 3-cyclic OAM-plus-polarisation rotation.
- The configuration applies reflections, PBS operations, beam splitters, and a DP operation in sequence.
- The ket number denotes OAM, while H and V denote horizontal and vertical polarisation.
6-cyclic OAM+Polarisation rotation
The section presents an experimental configuration for a 6-cyclic rotation involving OAM and polarisation labels.
- The section identifies an experimental configuration for the 6-cyclic OAM-plus-polarisation rotation.
- The ket number denotes OAM, while H and V denote horizontal and vertical polarisation.
8-cyclic OAM+Polarisation rotation
The section presents an experimental configuration for an 8-cyclic rotation involving OAM and polarisation.
- The section identifies an experimental configuration for the 8-cyclic OAM-plus-polarisation rotation.
- The ket number denotes OAM, while H and V denote horizontal and vertical polarisation.
14-cyclic OAM+Polarisation+Path rotation
The experimental configuration uses notation for orbital angular momentum, polarization, and paths.
- The number in each ket denotes orbital angular momentum (OAM).
- H and V denote horizontal and vertical polarization, respectively.
- a and b denote the two different possible paths.
S7) Learning algorithm
The learning algorithm autonomously expands its toolbox using properties of long cycles while managing learned elements to preserve search variability.
- The algorithm adds elements based on the properties of the longest cycle.
- It retains elements with large cycles and tests non-trivial coupling between different degrees of freedom.
- Previously learned elements can be forgotten to improve variability and prevent dead-ends.
S8) Simplification of experiments
Found experiments are simplified iteratively by removing unnecessary elements, replacing specialized elements, and streamlining path structures until no further simplification is possible.
- Elements are removed when the resulting state or transformation remains unchanged.This can remove inaccessible elements, while groups such as four beam splitters forming Mach-Zehnder interferometers may need joint removal.
- Complicated elements such as LI, PBS, or DP are replaced by mirrors when only specific modes access them.For example, a PBS can be replaced when only vertically polarized photons reach it.
- Path structures are rearranged to remove unused outputs and paths.Two successive PBSs can leave one output unused, allowing the second PBS and one path to be removed.
- All three simplification methods are applied repeatedly until no further simplification is possible.