Use Acadwrite for Research, Data Analysis, and Paper Writing in China's MCM Contest
Acadwrite Team · Last updated July 24, 2026
The 2025 China Undergraduate Mathematical Contest in Modeling began at 6:00 p.m. on Thursday, September 4, and submissions closed at 8:00 p.m. on Sunday, September 7, giving teams 74 hours (2025 contest notice). Each team ultimately had to submit a complete contest paper, including an abstract, problem restatement, model assumptions, model formulation and solution, results analysis, model evaluation, references, and appendices. The schedule below shows how Acadwrite can turn the problem statement and data into these materials.
Acknowledged in dozens of SCI papers · 100,000+ research users
Opening night: break down the problem and gather methods
Undergraduate teams choose from Problems A, B, and C. The three 2025 problems emphasized different methods. Problem A, Smoke Screen Deployment Strategy, focused on physical modeling. Problem B, Determining the Thickness of a Silicon Carbide Epitaxial Layer, focused on engineering parameters. Problem C, Selecting the Timing of NIPT and Identifying Fetal Abnormalities, focused on data analysis. Teams should compare each problem with their members' strengths in statistics, programming, and engineering modeling. After choosing a problem, Acadwrite reads the statement and data notes, then lists the objectives, constraints, variables, data fields, and subproblems.
The rules prohibit discussing the problem with anyone outside the team and joining online problem-discussion groups (original contest notice). Teams may consult public literature, but may not ask outsiders about the problem or participate in problem discussions.
Problem keywords, data notes, and the subproblems to answer.
Search public literature, read the methods, parameters, and conditions of use, and produce a shortlist of methods for the team.
Modeling: analyze data and create result charts
The middle two days focus on modeling and validation. The three members can take primary responsibility for the model, code, and paper. Once the team settles on an approach, Acadwrite can inspect fields, missing values, and outliers in the supplied data, produce descriptive statistics, and run the analyses required by the model. The Agent compares results across parameters or alternatives and creates statistical charts, result plots, and model flow diagrams.
After each verified result is ready, the Agent writes the model assumptions, solution steps, charts, and conclusions into the corresponding section. Variable names, equation symbols, and figure numbering remain consistent across the paper. The team must verify the model specification and interpretation.
The model assumptions conflict with the source conditions
If a model assumption conflicts with a published conclusion, the Agent compares the study population, experimental conditions, and parameter range. After the team confirms applicability, it revises the assumption wording, model parameters, and calculated results.
The official judging criteria first examine the reasonableness of assumptions, followed by modeling creativity, result correctness, and clarity of writing (contest rules).
Final day: turn the model results into a complete paper
Acadwrite creates a paper draft from the confirmed model, data results, and charts. The abstract directly states the method and main numerical conclusion for each subproblem. The body explains the assumptions, model, solution process, results, and limitations.
The team's confirmed variable definitions, model equations, and solution steps.
Place equations in the relevant sections, standardize notation, and create a model flow diagram.
Final data results, charts, and conclusions for each subproblem.
Insert charts into the results section, explain what the values mean, and create an abstract containing the main results.
Before submission, check every item against the official formatting specification (2025 revision): the abstract is on the dedicated third page and no longer than one page; the body is no more than 20 pages and has no table of contents; the appendix contains all complete, runnable source code. Missing or non-runnable code disqualifies the entry from awards; the paper is anonymous; and the exported PDF is no larger than 20 MB. Under the rule introduced in 2025, teams must also truthfully complete the AI tool usage report and state whether tools assisted with research, data analysis, chart creation, or writing. The team remains responsible for confirming the model, calculations, and conclusions.
Five tasks from problem statement to contest paper
A submission-ready modeling paper must connect problem interpretation, literature evidence, data results, charts, and writing. Acadwrite can process all of these materials around the same problem in one continuous workflow.
| Task | Input | Output | How Acadwrite handles it |
|---|---|---|---|
| Understand the problem | Problem statement, data attachments, and contest rules | Decomposed questions and deliverables | Identify objectives, constraints, variables, data fields, and the subproblems to answer |
| Search and read | Problem keywords and public literature | Background, methods, and parameter evidence | Search literature and extract study objects, methods, parameters, and conditions of use |
| Analyze data | Problem data and modeling approach | Statistical and model results | Inspect data, run analyses, compare alternatives, and interpret results |
| Create charts | Data results and model structure | Statistical charts, result plots, and flow diagrams | Choose suitable visuals, create charts, and state the conclusion from each one |
| Write the paper | Assumptions, model, data results, and charts | Complete contest paper | Organize the body around the abstract, restatement, assumptions, model, results, evaluation, and references |
- Abstract.State the method and main numerical result for each subproblem. Remove sentences that only describe the research process.
- Results and charts.For every chart, identify variables, units, and conclusions. Sensitivity analysis should show how parameter changes affect results.
- Format and appendices.Check anonymity, page count, file size, references, and complete source code.
In 2025, 68,311 teams entered and 294 undergraduate teams received first prizes (official awards news). Contest outcomes depend on the model, result validation, and written presentation. The tool handles sources, data, and documents.
Continue reading
Create a complete modeling paper from the problem and data
Acadwrite searches and reads public literature, analyzes the supplied data, and creates result charts. The Agent then turns model assumptions, solution steps, and results into a paper draft that follows the contest structure.