Reply is the place to meet an incredible variety of enthusiastic, passionate, ideas-driven people, who want to make a difference and an impact.Would you like to know more?
In the manufacturing sector, production lines are composed of several machines, each of which performs a particular specific task.
A product being processed will only complete its production cycle when it has passed through, in a clearly defined order, the machines that make up the production line. The set-up of machines can vary, depending on way different products are processed.
Each product has a different processing time and order of entry into every single machine. In addition to these variables, external events and issues related to the production line can also necessitate a modification of the production plan.
To optimise the entire production chain in terms of priority and time, it is essential that the correct processing order for different products is identified and that any critical situations can be managed in real time.
The complexity of different production lines, the number of set-ups involved, the production priorities and the constraints associated with the various processes represent a massive number of variables when modelling the problem and make it very complicated to resolve, particularly from a computational point of view. It therefore becomes necessary to use specific optimisation “solvers” that can resolve multi-dimensional problems within a very short time.
This type of optimisation is modelled as a job scheduling problem (JSP).
The use of quantum technology allows complex production lines to be managed in accordance with processing constraints and priorities, without impacting on the time needed to resolve the problem.
One of the recurrent challenges in the financial sector involves the need to maximise the profits obtainable from each portfolio in line with the rules and objectives set by the customer.
One means of achieving such a result involves diversifying the portfolio of securities in order to reduce the risks that might impact the return from those assets. This risk is due to the presence within the portfolio of several financial assets whose returns are not perfectly correlated.
To diversify a portfolio, a practitioner will need to identify, across a set of securities, the largest subset of securities that are not internally correlated. Binary selection operations are used to resolve such problems. These operations allow the optimal solution to be identified.
As the size of an investment portfolio increases, so does the complexity of the computational problem, resulting in excessively long processing times for traditional computers.
The quantum optimisation algorithm deployed for the diversification of the portfolio is modelled on the basis of the WMIS (weighted maximum independent set).
Quantum technologies significantly speed up the process of selecting the right portfolio options. Furthermore, by combining existing techniques, it is possible to quickly perform calculations that would otherwise require long processing times.
The detection of defects and faults within production lines is often crucial in preventing customer dissatisfaction and even in avoiding damage to the lines themselves.
The processes involved therefore need to be constantly monitored to ensure that high quality standards are maintained. While in the past this work was carried out manually, it can now be performed by automated means, thanks to image processing tools and machine learning algorithms.
By moving from manual to automated checks, production managers can reduce the time and cost involved. This is because, while an operative can undertake inspections only at a certain speed, checks can be carried out much faster with the aid of machine learning techniques.
The image processing and machine learning techniques that enable quality control to be to automated require a highly accurate system of classification.
Incorrect classification can allow an unsuitable product to continue through the manufacturing process, resulting in a final product that is either of poor quality or that could even damage the production line itself.
Deep learning-type algorithms are therefore used for classification purposes, which require extended periods of 'training' if satisfactory results are to be obtained.
The neural networks created using quantum algorithms – known as quantum neural networks (QNNs) – are substantially faster.
Quantum technologies are used to speed up the formation of deep neural networks, a process that usually represents a bottleneck due to the excessive processing times required by traditional computers.
The planning of rail traffic within stations is a topic of interest for managers of railway infrastructure.
The high number of trains in transit, arriving and departing, the variety of tracks available for trains to stop at, and the large numbers of passengers involved every day are the variables concerned in what is a complex situation to manage.
Optimising the process of scheduling transportation fulfils the aims of both assisting passengers in transit within the station and optimising the use of resources provided by the station for the preparation and maintenance of trains.
The combinatorial nature of the problem and the number of actors and variables that need to be taken into account make resolving the issue complex from a computational point of view. There is a need to identify a solution that is optimised for all the stations in the network.
This type of optimisation is modelled as a “train platforming problem” (TPP).
The use of quantum technologies offers a significant benefit, in that quantum processors can handle an extremely high number of combinatorial variables with substantially shorter processing times, in a way that goes far beyond the possibilities offered by current computers. This can enable managers to optimise the entire railway station network by significantly reducing the impact of any delays on service quality.