Deloitte's Quantum Climate Challenge 2024 – Flood Prediction

How to submit

To participate in the challenge, each participant must create a profile on the ekipa platform, provide the information requested during registration process and accept the general terms and conditions for using the ekipa platform. If a participant has already created a profile, he or she must use it to participate in the Quantum Climate Challenge.

Your submission must be handed in in English and as one document in pdf format, optionally including clickable links before April 28, 2024, 23:59 (CEST). On top, you are free to upload any additional files.

Tasks

Quantum machine learning models can be created using hybrid classical-quantum computation in a myriad of ways. To reduce the complexity of the challenge to a level that is manageable in the given time frame, we have simplified the problem.

It is highly encouraged to analyze the problem as a whole and deviate from these simplifications to further improve performance of the calculations and improve the fit of the calculated solution.

It is sufficient to solve the challenge using the simplifications. However, producing a solution to a higher complexity problem will raise the likelihood of achieving a good ranking.

To successfully complete this challenge please complete the following tasks:

Task 1

A: Create a quantum algorithm, a (hybrid) quantum machine learning model, that predicts if a flood is happening on each day of 2023. To reduce the complexity, you can assume that a water level above 90cm is a flood. Run your algorithm on a quantum computer or simulator and provide information on the resource requirements of your solution (e.g., total number of shots, compute time, etc.)

B: Evaluate your solution, describing the advantages and disadvantages of your approach(es). Evaluate the performance differences between your solution and the classical approach. Use at least the following evaluation criteria:

• Training time
• Accuracy of all predictions for the year 2023, including:
• I. In relation to a classification:
• i. False-Positive
• ii. True-Positive
• iii. False-Negative
• iv. True-Negative
• II. In relation to a water level forecast:
• i. Mean-Absolute-Error
• ii. Mean-Squared-Error
• Learning curve (for models using epoch training)

Task 2

A: Conceptualize a quantum or hybrid solution that will scale the calculation to achieve a more general/advanced model.

B: Discuss the requirements for your solution to 2A to be implemented in real quantum computers. Give an estimate for the time horizon at which implementation may become feasible. Examples of requirements include: the number of logical qubits needed, coherence times, etc.

Task 3

Compile a report (as a .pdf file) that includes a short problem statement, your solution, and an explanation for how you solved it. Give an overview of your research and the resources used during the challenge, provide .csv or .xls files for all data resulting from the calculations in your report and supply your comprehensively commented code (In either a repository or file.) 

Details on how to access the simulators and quantum computers (on IBM Quantum, Nvidia, D-Wave, and Amazon Braket) as well as the data provided by the Wupperverband can be found on the Resources tab of the challenge webpage. The details and data are only visible for registered participants.

Structure of the pdf to be submitted

1. Give an overview on you/your team and your background(s) including your contact details (name, team name, e-mail, affiliation).
2. Provide an ONE DinA4 page summary/abstract of your solution with a maximum of 400 words.
3. Give a detailed description of your algorithm (refer to task 1) including the concept, general composition, underlying assumptions, and Evaluation. Please also list all additional data that were used.
4. Describe the results of your algorithm and include a clickable link to a repository including the code used to run your algorithm.
5. Describe your envisioned algorithm including the expected benefits (refer to task 2) Elaborate on the requirements for your solution.
6. (Optional) You can elaborate on your experience and learnings during this challenge and provide feedback on the challenge tasks/ settings.

Selection procedures and evaluation criteria

The selection of the most promising IDEAS and their ranking are based on the following criteria:

• Degree of innovation of the approach, concept, and algorithm including creativity and originality (25%)
• Number, comprehensiveness, and adequacy of fulfilled tasks considering the challenge aim and setting (20%)
• Feasibility, usefulness and functionality of the approach, concept, and algorithm (25%)
• Quantum community impact – will your solution lead to progress within the quantum community e.g. create new applications or projects, spark discussions, increase public interest and knowledge about quantum? (15%)
• Presentation and structure of the results (15%)

As a reminder, it is sufficient to solve the challenge using all made simplifications. Increasing the complexity of the problem your solution addresses will result in a better chance of winning the challenge.

Ideas to increase problem complexity:

• Collect additional data that might be useful for the model and integrate them.
• Increase the time span between when a flood forecast is made and the time of the flood. (You could, for example, predict the next three/seven/ten days at any point in time.)
• Predict water levels or water level intervals instead of a binary outcome.
• Add additional evaluation criteria that might be useful.