Laser-based methods, devices and systems for quantum classification and operation control

A system comprising:

• a vacuum chamber;
• at least two lasers;
• an optical device;
• at least one processor;
• and
• at least one memory comprising computer program code for one or more programs;
• wherein the at least one processor, the at least one memory, and the computer program code are configured to cause the system to at least:
• radiate the vacuum chamber comprising an ensemble of neutral atoms with laser configured to trap atoms of the ensemble of neutral atoms in an array of optical tweezers, thereby providing a quantum register, and each optical tweezer comprising a single neutral atom;
• digitally configure, by processing a training dataset with a plurality of vectors about a target where each vector has a historical value for each of N features of the target, a laser controlling, and digitally reduce, based on the laser controlling function, a cost function counting an error in the classification of one or more vectors about the target where each vector has a value for each of N features of the target, wherein:
• the laser controlling function includes M unitary operations, M being a natural number greater than 1 and less than or equal to N, with each unitary operation being at least dependent upon:
• a) the N historical values of each vector of the plurality of vectors of the training dataset, b) a set of weight values for a) and comprising a predetermined set of weight values, and c) a set of free parameters comprising a predetermined set of free parameters, and the cost function being reduced by changing the set of weight values of b) and the set of free parameters of c) of each unitary operation;
• radiate the ensemble of atoms with laser configured to excite at least some atoms of the quantum register, with the laser being operated in accordance with laser parameters resulting from each unitary operation of the configured laser controlling function;
• read the quantum register with optical means, and digitally define a bit for each atom in the quantum register with a binary value thereof based on an amount of light produced by the respective atom;
• using the bits defined, digitally compute a result of the cost function counting the error in the classification of one or more vectors about the target where each vector has a value for each of N features of the target;
• and
• digitally process a computed result of the cost function such that:
• when the result is below a predetermined threshold, digitally set a classifier for the target, the classifier being set with the values of both the sets of weight values of b) and the sets of free parameters of c) of the configured laser controlling function;
• and when the result is not below the predetermined threshold, carry out the following in a looped manner until a convergence factor fulfills a predetermined criterion:
• digitally store at least the result of the cost function as last computed;
• digitally modify at least one of the sets of weight values of b) and the set of free parameters of c) of each unitary operation with an optimization technique;
• digitally reconfigure the laser controlling function by processing the training dataset, and digitally reduce, based on the reconfigured laser controlling function, the cost function with the modified at least one of the sets of weight values of b) and the set of free parameters of c) of each unitary operation;
• radiate the ensemble of atoms with laser configured to excite at least some atoms of the quantum register, with the laser being operated in accordance with laser parameters resulting from each unitary operation of the laser controlling function as last reconfigured;
• read the quantum register with optical means after the last irradiation of the ensemble of atoms, and digitally define the bits based on the quantum register as last read;
• using the bits as last defined, digitally compute a result of the cost function;
• digitally process the result of the cost function as last computed, and digitally provide a convergence factor based on both said result and the result as last stored;
• and
• if the convergence factor fulfills the predetermined criterion, digitally set the classifier for the target, the classifier being set with the values of both the sets of weight values of b) and the sets of free parameters of c) of the laser controlling function as last reconfigured.

A non-transitory computer-readable medium encoded with instructions that, when executed by at least one processor or hardware, perform or make a device or system to at least perform the following:

• digitally configure, by processing a training dataset with a plurality of vectors about a target where each vector has a historical value for each of N features of the target, a laser controlling function, and digitally reduce, based on the laser controlling function, a cost function counting an error in the classification of one or more vectors about the target where each vector has a value for each of N features of the target, wherein:
• the laser controlling function includes M unitary operations, M being a natural number greater than 1 and less than or equal to N, with each unitary operation being at least dependent upon:
• a) the N historical values of each vector of the plurality of vectors of the training dataset, b) a set of weight values for a) and comprising a predetermined set of weight values, and c) a set of free parameters comprising a predetermined set of free parameters, and the cost function being reduced by changing the set of weight values of b) and the set of free parameters of c) of each unitary operation;
• operate a laser in accordance with laser parameters resulting from each unitary operation of the configured laser controlling function to radiate an ensemble of atoms in a vacuum chamber configured to excite at least some atoms of a quantum register provided by an array of optical tweezers;
• digitally define a bit for each atom in the quantum register with a binary value thereof based on an amount of light produced by the respective atom;
• using the bits defined, digitally compute a result of the cost function counting the error in the classification of one or more vectors about the target where each vector has a value for each of N features of the target;
• and
• digitally process a computed result of the cost function such that:
• when the result is below a predetermined threshold, digitally set a classifier for the target, the classifier being set with the values of both the sets of weight values of b) and the sets of free parameters of c) of the configured laser controlling function;
• and when the result is not below the predetermined threshold, carrying out the following in a looped manner until a convergence factor fulfills a predetermined criterion:
• digitally store at least the result of the cost function as last computed;
• digitally modify at least one of the sets of weight values of b) and the set of free parameters of c) of each unitary operation with an optimization technique;
• digitally reconfigure the laser controlling function by processing the training dataset, and digitally reducing, based on the reconfigured laser controlling function, the cost function with the modified at least one of the sets of weight values of b) and the set of free parameters of c) of each unitary operation;
• operate the laser in accordance with laser parameters resulting from each unitary operation of the laser controlling function as last reconfigured to radiate the ensemble of atoms in the vacuum chamber configured to excite at least some atoms of the quantum register;
• digitally define the bits based on the quantum register after the last radiation with the laser;
• using the bits as last defined, digitally compute a result of the cost function;
• digitally process the result of the cost function as last computed, and digitally provide a convergence factor based on both said result and the result as last stored;
• and
• if the convergence factor fulfills the predetermined criterion, digitally set the classifier for the target, the classifier being set with the values of both the sets of weight values of b) and the sets of free parameters of c) of the laser controlling function as last reconfigured.