Reviewed learning material · Video analysis · EnglishRead the full overview
This segment serves as the opening of a crash course on the CUMCM. The instructor first displays the lecture notes table of contents, covering various models and writing techniques, then proceeds to the "Introduction" chapter, detailing the core format of the China Undergraduate Mathematical Contest in Modeling (CUMCM): teams of three, completing one paper within three days (74 hours). The video especially stresses that the contest evaluation criteria do not rely solely on the absolute correctness of computational results, but rather emphasize the paper's logic (makes sense), academic level (has depth), and completeness (can be finished). Furthermore, it clarifies that only undergraduate and junior college students are eligible to participate; the two groups are awarded separately, and the contest problems they face differ (A/B/C versus D/E).
This segment mainly introduces the submission process and topic-selection strategy for the National College Students' Mathematical Modeling Contest (the national contest). It first clarifies the content of supporting materials (such as code, data, and charts) and stresses that the paper is the sole basis for evaluation, thereby alleviating students' anxiety about missing attachments. Next, it thoroughly analyzes the disciplinary tendencies (e.g., physics and engineering, socioeconomics, big data, etc.) and target participants for Problems A through E. Finally, it provides topic-selection advice grounded in competitiveness and individual background, noting that although Problem C has a low entry barrier, its discrimination power is small, making it challenging to win a national award.
This segment primarily explains topic selection strategies for the National College Students' Mathematical Modeling Contest and question-setting patterns over the past ten years. First, it analyzes the difficulty, discrimination power, and suitable team types for different problem types, offering concrete topic selection advice. Then, it summarizes four major question-setting patterns: "no repetition within three years but with cyclical recurrence," "mapping of current hot topics," "a data-oriented problem every year," and "progressive problem structure." These insights aim to help participants better plan their preparation and select appropriate contest problems.
This segment mainly introduces preparation suggestions and award evaluation mechanisms for the Mathematical Modeling National Contest. First, three preparation suggestions are given: studying past real problems, accumulating code templates, and preparing paper templates. Then, the national contest's two-tier evaluation mechanism—'provincial evaluation plus national review'—is analyzed in detail, clarifying the approximate award ratios for each level. The instructor especially emphasizes that provincial awards involve competition only within the province and are subject to a quota system, meaning the final competition is effectively among students from the same school. Finally, the rules are clarified: National First Prize requires an oral defense, while National Second Prize does not; additionally, submitting a paper guarantees the Successful Participation Award as a safety net.
This segment covers award ratios, the review process, and scoring criteria for the National Mathematical Modeling Competition. The video first presents empirical ratios for provincial and national awards, explaining that national first and second prizes combined amount to approximately three teams per hundred; it then follows the handout order to explain initial screening, independent scoring, provincial first-prize selection, national re-evaluation, and national first-prize oral defense; finally, it lists the four core scoring dimensions and their weights, and transitions to the Day One schedule of the three-day 74-hour competition.
This segment mainly explains the three-day time plan for the National Mathematical Modeling Competition, the list of essential tools, and two core survival rules: begin writing the paper on the first night to avoid last-minute rushing, and team members must work in parallel—not wait for each other—to prevent falling into a deadlock.
This segment explains the 'Survival Rule Three: Get It Done First, Then Refine It' for the National Mathematical Modeling Contest. The instructor points out a common mistake many teams make: over-pursuing perfection in Question One, leaving insufficient time for subsequent questions. The correct strategy is to first build a 'baseline model' using the simplest possible method to ensure results for all questions, because judges value overall completion more than the depth of any single model. Finally, the video displays the pre-contest checklist, reminding teams to prepare thoroughly before the contest—covering environment setup, templates, role division, and equipment.
Use the learning inspector for key ideas and moments, or open the reading tabs for the complete notes.
Generated from the video's visuals and explanation; not verbatim speech.
At the start of the video, the instructor, known as 'The Math Modeling Brother,' greets the audience and explains that this course aims to help learners quickly master general problem-solving strategies for the National College Student Mathematical Modeling Contest within two hours. A detailed PDF lecture outline appears on screen, covering common mathematical models such as evaluation, classification, clustering, prediction, optimization, and mechanism modeling, as well as practical skills including LaTeX and Word typesetting, abstract writing, and more.
Next, the course enters its first section, 'Getting Started,' formally introducing the National College Student Mathematical Modeling Contest (CUMCM). Referring to the lecture notes on screen, the instructor summarizes the contest's core format in one sentence: 'A team of three + three days (74 hours) + one paper.' Using a red pen, he circles these key phrases on screen to emphasize that this is the fundamental contest framework.
Then, the instructor focuses on the contest's evaluation criteria. He stresses that the contest 'does not test who computes correctly'—although reference answers exist, what matters more is whether judges find the paper 'reasonable, sophisticated, and complete.' This means logical rigor, model depth, and work completeness often outweigh mere computational accuracy. This perspective aims to correct beginners' common misconception of 'result-only thinking.'
Finally, the video details eligibility requirements and team category assignments. The instructor emphasizes that CUMCM is open only to undergraduate and junior college students; graduate students are ineligible. Awards are separately determined for undergraduate and junior college teams. Typically, undergraduate teams choose among Problems A, B, and C, while junior college teams select from Problems D and E. The instructor adds that, because junior college problems tend to be comparatively more challenging and attract fewer participants, junior college students may have a higher probability of winning national-level awards. Additionally, teams must consist of exactly three members, with no restrictions on academic background, and must submit a complete paper—including a PDF main text and attachments (supporting materials)—within 74 hours after problem release.
When preparing for the National Mathematical Modeling Contest, many students are unclear about what to submit as 'supporting material'. In fact, it is quite simple: in addition to the final paper in PDF format, you should package and submit all auxiliary files generated during your problem-solving process. This mainly includes three categories: first, the program code you used to solve the problem; second, the original data you used or cleaned; and third, various result charts and data tables generated by running the code. Simply include all such intermediate outputs.
Here is a common misconception that needs clarification: many students habitually believe that poorly organized attachments will affect their scores—and even become anxious about it. In reality, the contest’s evaluation rules explicitly state that 'the paper is the sole basis for evaluation'. This means judges primarily assess how well your paper is written. Moreover, for most awards—such as provincial first-, second-, and third-place prizes, and even some national second-place prizes—no on-site defense is required. Only teams aiming for the highest honor, the national first-place prize, may be required to defend, and even this requirement is gradually being relaxed in certain regions. Therefore, do not expend excessive effort striving for perfection in your attachments.
Next, let us examine the distribution of problem types in the national contest. Each year, typically 4 to 5 problems are provided, labeled A through E, and participating teams need only select one. Problem A usually leans toward physics and engineering topics—such as heat conduction and stress analysis—and requires advanced mathematical tools like differential equations and finite element analysis, making it relatively difficult. Problem B focuses on social and economic issues—such as traffic optimization and resource allocation—and commonly employs optimization models, graph theory, or game theory. Problem C is purely a data analysis and statistics problem—what many call the 'big data problem'—involving machine learning, predictive modeling, and so on. As for Problems D and E, they are typically designed specifically for vocational college teams; although their descriptions may appear complex, the actual depth of questioning is often less profound than that of Problems A and B.
Finally, let us discuss topic selection strategy. Because Problem C (the data problem) has relatively low professional barriers, it attracts the largest number of participants—approximately 40% of all contestants. This leads to an issue: although it is easy to get started, the large participant base reduces differentiation. If you aim for the top national first-place award, standing out in Problem C becomes extremely difficult. By contrast, if you have a science or engineering background, tackling Problem A or Problem B may offer greater advantage, since these problems provide higher differentiation—and performing well on them makes recognition by judges more likely.
In mathematical modeling contests, selecting an appropriate problem is crucial. Problem C (the data problem), although having a lower entry barrier, attracts many teams (approximately 40%), resulting in low discrimination power and making it difficult to stand out and win a national first prize. In contrast, Problems A and B attract relatively fewer teams (each approximately 20%–25%), yielding higher discrimination power; performing well on either significantly increases the chance of winning a national award, though the difficulty is correspondingly higher. Therefore, if the team includes students with physics or engineering backgrounds, Problems A or B are recommended as top priorities; if all three team members come from mathematics, statistics, or computer science backgrounds, choosing Problem C is the safest strategy; for interdisciplinary teams, Problems B or E are recommended (note: this suggestion applies mainly to the vocational college division).
Understanding question-setting patterns over the past ten years helps us prepare more effectively. First, problems exhibit the characteristic of "no repetition within three years but with cyclical recurrence." Problems from the same domain do not appear consecutively—for example, in year 2019, Problem B was physics-related (synchronized drumming); in year 2020, it was social-science-related (desert game); and in year 2021, it was optimization-related (ethanol coupling to produce C4 olefins). However, similar problem types reappear under new guises every 3 to 4 years. Second, contest problems clearly exhibit "mapping of current hot topics": major societal or technological themes of the year are highly likely to appear in the problems. For instance, in year 2020—the first year of the pandemic—multiple problems involved scenarios such as contagion, quarantine, and resource allocation; in year 2022, carbon neutrality emerged as a prominent theme; and in year 2024, numerous AI-related data modeling problems appeared. Third, "a data-oriented problem appears every year." Since year 2015, when Problem C became fixed as big-data/statistical modeling, there has been exactly one purely data-oriented problem annually, typically accompanied by a large Excel or CSV file requiring data processing, modeling, prediction, or classification; the current trend is essentially "prediction plus optimization." Finally, each problem usually contains 3 to 4 sub-questions, with difficulty increasing progressively. Generally, the first sub-question is relatively simple (introductory/verification), the second and third become increasingly complex, and the fourth typically requires model improvement or sensitivity analysis. Note that scoring follows a per-sub-question system: failing to complete the first two sub-questions does not affect scoring for the latter two, provided that responses are submitted; otherwise, incomplete submissions may leave a poor impression on evaluators.
The video opening displays "targeted preparation suggestions." The instructor states that the first step in preparation is to study the actual A/B/C problems and outstanding papers from the past 3 years before the contest, to understand the problem-setting style and evaluation criteria; the second step is to accumulate code templates for common models across various fields in advance; the third step is to prepare a general-purpose paper template. On screen, the instructor circles and checks these points with a red pen.
Next comes the section "1.4 Award rates and evaluation mechanism." The instructor explains the two-tier evaluation mechanism adopted by the national contest: "provincial evaluation + national review." This means the evaluation process proceeds first at the provincial level, then at the national level. The instructor emphasizes that once a paper is submitted, it enters the subsequent process directly—contestants need not submit new papers or make revisions midway.
The instructor interprets the provincial award ratios in detail using the table on screen: approximately 8% of teams receive provincial first prizes, 15% receive provincial second prizes, and 20% receive provincial third prizes. The instructor notes that the overall award rate is low and clarifies a common misconception: provincial awards are evaluated solely within each province—contestants do not compete against top teams from other provinces such as Beijing or Shanghai; their actual competitors are limited to teams from the same province. Thus, the instructor describes the national contest as feeling, in some sense, like a "campus-level contest."
Next, the instructor introduces the concept of "quota-based allocation." Award quotas are distributed to individual universities—for example, a given university may receive only 2 national awards and 3 provincial first prizes this year. On screen, the instructor handwrites "2, 3" to illustrate this point. This means that under quota-based allocation, contestants ultimately compete against peers from their own university.
For national awards, the table shows that national first prizes account for approximately 0.5%–1%, and national second prizes account for approximately 2%–3%. These two categories are evaluated by the national organizing committee. The instructor adds that national second prizes generally require only paper evaluation, without oral defense; national first prizes, however, require oral defense (online in some years).
Finally, the instructor mentions the "successful participation award." As long as a paper is submitted, contestants will almost certainly receive this award, which signifies completion of the contest. The sole exception is plagiarism: plagiarism not only disqualifies a team from all awards but also leads to serious consequences. Additionally, key figures listed at the bottom of the document state that approximately 43% of teams achieve provincial third prize or higher.
This segment first reads the award-winning proportion from the lecture notes as a set of numbers that can be memorized directly: approximately 43 percent can win provincial third prize or higher, approximately 8 percent win provincial first prize, among which approximately 1 percent ultimately win national first prize, and approximately 2 percent win national second prize. Next, the video combines "national first prize + national second prize" for interpretation, yielding the intuitive conclusion that "approximately 100 teams out of every 3 teams receive the national award", and adds that the actual number may be slightly lower, approximately 1 to 2 teams.
The screen then shifts downward to "Review Process", beginning with Step 1: "Provincial Organizing Committee Preliminary Screening." Two strict thresholds are emphasized here: first, format violations—the video explicitly states, "Format violations result in immediate disqualification"; second, excessive similarity detection—the handout specifies "overall text similarity > 25% or code similarity > 50%." The instructor explains excessive similarity detection as a signal of plagiarism, and further clarifies that code placed in appendices is typically not heavily scrutinized, but if large blocks of code are pasted directly into the main body of the paper, they fall within the scope of similarity detection—hence it is not advisable to paste dozens of pages of code at the end of the main text.
Next, following the handout’s order, Steps 2 and 3 are covered. Step 2 is 'Group Review by Judges', under which each paper is independently scored by 2 to 3 judges. Step 3 is 'After Provincial First Prizes Are Determined': only the top 10% to 15% of provincial first-prize papers are selected for re-evaluation by the National Organizing Committee. The key point here is the sequence of steps: provincial first prizes are awarded first, and then the top 10%–15% among those provincial first-prize papers advance to the national level.
Steps 4 and 5 now shift to the national level. The handout states: 'National Re-evaluation: Scoring is redone to determine national first and second prizes', and 'National First Prize Candidates: Must attend a defense (in person or online) answering questions about their paper'. Based on this, the video clarifies that re-evaluation does not reuse scores from the provincial level but instead involves fresh scoring; moreover, national first-prize candidates must undergo a defense focused entirely on their paper—its core purpose is to verify whether the work was genuinely completed by the authors and whether the authors are thoroughly familiar with their own paper.
Then comes 'Core Scoring Dimensions (Ranked by Weight)'. The video lists four dimensions: model reasonableness 30%–40%, solution correctness 25%–30%, paper presentation 20%–25%, and innovation 10%–15%. The instructor further explains that what judges truly value is not how sophisticated the algorithm is, but rather why this particular model was chosen, how the data were processed, whether assumptions are reasonable, whether sensitivity analysis was performed, and whether conclusions are supported by data. This segment grounds the abstract scoring criteria in concrete expectations for argumentation.
Finally, the screen switches to "1.5 Competition Schedule (Three Days, 74 Hours)" and displays the schedule for Day One: from 18:00 to 18:30, receive the problems and quickly skim them; from 18:30 to 20:00, discuss topic selection; from 20:00 to 22:00, finalize the topic and search literature; from 22:00 to 01:00, build the modeling framework for Question One. This segment only elaborates the Day One schedule; the full content for Days Two and Three is not covered here.
In the National Mathematical Modeling Competition, sound time planning is the foundation for winning awards. After receiving the problem on Day One, you must produce the modeling approach for Question One by evening; at the latest, finalize the topic and begin constructing the model framework by 22:00.
The core task on Day Two is solving the first two questions; by evening, both the solutions and the corresponding paper sections for Questions One and Two must be completed. On Day Three, tackle Question Three and all remaining questions, then finalize the paper to ensure a complete first draft.
The final day focuses on refining the abstract, repeated revision, final formatting, and exporting the PDF. Be sure to submit before 7 p.m., because the system is highly prone to crashing during the last half hour. Although only MD5 code needs to be submitted, preparing early helps avoid unexpected issues.
A craftsman who wishes to do his work well must first sharpen his tools. Essential tools for the National Mathematical Contest in Modeling include: programming (choose either Python or MATLAB), plotting, writing (LaTeX or Word recommended), literature search, AI assistance, and collaboration tools. If using LaTeX, collaboration needs decrease accordingly.
Beginners must remember two fundamental survival rules. Rule One: Start writing the paper on the first night. Avoid spending the first day only searching literature and outlining ideas, which leads to frantic last-minute work on the third day. The introduction, problem restatement, and model framework can all be written in advance, because final evaluation focuses on paper quality—not how elegant your code is.
Rule Two: Never wait for each other. In a team, serial workflows—such as waiting for formulas, waiting for results, or waiting for integration—lead to deadlocks. The correct approach is parallel work: begin writing immediately upon receiving a rough plan; reserve space if figures are not yet ready; describe formulas in words before they are fully written—ensuring maximum progress.
At the start of the video, the instructor continues from the previous slide's topic 'Don't Wait for Teammates', circling a new title in red ink: 'Survival Rule Three: Get It Done First, Then Refine It'. He criticizes a common contest mindset: some students stubbornly fixate on a single result, insisting 'this result is unreasonable', yet cannot specify exactly what makes it unreasonable when asked. This blind parameter tuning—lacking clear diagnosis—often causes teams to waste the first two days of the three-day contest on Question One alone, leaving only time for hasty, incomplete work on later questions.
To address this pain point, the instructor presents the correct operational path: first produce a result using the simplest possible method. On the handout, he circles the word 'result' heavily in red ink and stresses, 'Note: the result is most important.' This means teams should prioritize building a 'baseline model'—even if it is not sophisticated—as long as it generates data and can be written into the report. The instructor states plainly that without any result, no matter how elegant the model or advanced the code, judges will not read it. Only after completing all questions and shaping the main body of the report should teams revisit and improve or replace models—if time remains.
To reinforce the rationale behind this strategy, the instructor cites the core evaluation criterion printed at the bottom of the handout: 'Judges assess whether you have completed all questions—not how sophisticated a model you used for Question One.' This sentence directly reveals the contest's evaluation orientation: overall completion far outweighs local technical depth. Thus, 'having a result matters more than pursuing perfection by a factor of 100' is not merely experiential advice—it is the optimal solution aligned with judging logic.
After explaining the core rule, the screen scrolls to '1.8 Pre-contest Checklist'. The instructor quickly reviews this checklist, reminding participants to verify all foundational preparations before the contest begins: team registration details, Python/MATLAB environments and commonly used packages, LaTeX and Word templates, CNKI/database account access, team role assignments, file-naming conventions, AI tool readiness, reading of outstanding papers from the past three years, and hardware such as laptops, chargers, and backup internet connections. Once the checklist presentation concludes, the instructor announces the end of the introductory norms section and encourages everyone to prepare well for upcoming hands-on practice.
Knowledge cards
01
Outline of the Rapid-Prep Course for the National Mathematical Modeling Contest
This course aims to help learners master problem-solving strategies for the National Mathematical Modeling Contest in a short time. Its content is systematic and comprehensive: it covers explanations of core mathematical models—including evaluation, classification, clustering, prediction, optimization, and mechanism modeling—as well as practical guides such as LaTeX/Word paper formatting, abstract writing standards, AI tool usage, and award-winning techniques.
02
CUMCM Core Definition: Three People, Three Days, One Paper
The basic format of the National College Student Mathematical Modeling Contest (CUMCM) is as follows: a team of three students works for three days (a total of 74 hours) to build a mathematical model for a given problem and ultimately completes and submits one full academic paper.
03
CUMCM Evaluation Criteria: Emphasis on Logic and Completeness
CUMCM evaluation does not solely pursue absolute correctness of computational results. Instead, judges prioritize whether the paper is 'reasonable' (logically rigorous and model-appropriate), 'sophisticated' (demonstrating academic depth and some innovation), and 'complete' (finished and submitted as a full work within the time limit). Although reference answers exist, procedural rigor and paper completeness are more critical.
04
Rules for Eligibility and Team Category Assignment
The China Undergraduate Mathematical Contest in Modeling (CUMCM) is open only to undergraduate and junior college students currently enrolled; graduate students are not eligible. The contest has two categories: the undergraduate category and the junior college category, with independent judging and separate award ceremonies for each. Typically, undergraduate students choose Problems A, B, or C, while junior college students choose Problems D or E. Each team consists of three members, with no restrictions on academic major, and must submit a PDF paper along with supplementary supporting materials within 74 hours after the problems are released.
05
National Contest Supporting Material Checklist
In addition to submitting the paper, supporting material must also be packaged and uploaded. Core contents include: 1. Code used to solve the problem; 2. The dataset used; 3. Various result charts and result tables generated by running the code. The principle is to include all intermediate outputs except the main text of the paper.
06
Core evaluation principle: paper exclusivity
The submitted paper document is the sole basis for national contest evaluation. Supporting material serves only as auxiliary verification. Furthermore, provincial awards and most national second-place awards require no on-site defense; only national first-place awards may require defense in certain regions. Thus, there is no need to over-anxiety about minor oversights in attachments.
07
Problem Type Overview: Characteristics of Problems A–E
Problem A: Physics/engineering-oriented, emphasizing mechanism analysis (differential equations, finite element methods); Problem B: Social/economics-oriented, emphasizing optimization and scheduling (operations research, graph theory); Problem C: Data/statistics-oriented, emphasizing data mining and prediction (machine learning, data analysis); Problems D/E: Exclusively for specialized-track teams, with moderate difficulty.
08
Pitfall Avoidance Guide: The Competition Trap in Problem C
Problem C is the most popular choice due to its low entry barrier (accounting for approximately 40% of teams), but this results in extremely low discrimination power. For teams aiming for a national first prize, Problem C represents a red-ocean competition where standing out is exceptionally difficult. Teams with an engineering or science background are advised to prioritize Problems A or B, which offer higher discrimination power.
09
Topic selection strategies for mathematical modeling contests
Selecting problems based on team background:
- Problem C (data problem): most popular choice (approximately 40%), low barrier but low discrimination power; suitable for teams with pure mathematics/statistics/computer science backgrounds seeking stability.
- Problems A/B: less popular (each approximately 20%–25%), high discrimination power, favorable for national awards but more challenging; suitable for teams with physics/engineering backgrounds.
- Interdisciplinary teams: recommend Problems B or E (vocational college division only).
10
National Contest question-setting pattern I: periodicity and hot-topic mapping
- No repetition within three years but with cyclical recurrence: problems from the same domain do not appear consecutively, but similar problem types recur every 3–4 years.
- Hot-topic mapping: major societal or technological themes of the year are highly likely to appear in problems (e.g., pandemic in year 2020, carbon neutrality in year 2022, AI in year 2024).
11
National Contest question-setting pattern II: characteristics of data problems and problem structure
- Data problems appear every year: Problem C has been fixed since year 2015 to focus on big data/statistics; there is always one purely data-oriented problem each year, often accompanied by a large Excel/CSV file, requiring data processing, modeling, prediction, or classification (currently mostly prediction plus optimization).
- Problem structure: Each problem contains 3 to 4 sub-questions, with difficulty increasing progressively. The first sub-question is simple; the second and third are more complex; the fourth often requires model improvement or sensitivity analysis. Scoring is done per sub-question; failing to complete the first two sub-questions does not affect scoring for the last two, but all sub-questions must be answered.
12
Three-step preparation strategy
For mathematical modeling contests, the video offers three core preparation suggestions: 1. Study actual A/B/C problems and outstanding papers from the past 3 years to grasp problem-setting and evaluation styles; 2. Accumulate code templates for common models in advance; 3. Prepare a general-purpose paper writing template.
13
Two-tier review mechanism
The national contest adopts a two-tier review mechanism: "provincial selection + national review." The process first awards provincial prizes, then national prizes. Once a paper is submitted, it enters a continuous review process; no resubmission or revision is required midway.
14
Provincial award ratios and competition scope
Approximate provincial award ratios are: about 8% for provincial first prizes, about 15% for provincial second prizes, and about 20% for provincial third prizes. Important note: Provincial awards are judged only within each province; competitors are exclusively teams from the same province, with no need to compete against high-performing teams from other provinces or municipalities.
15
Quota system and intra-school competition
The national contest operates under a quota system: award quotas are allocated specifically to individual universities (e.g., a university may receive only 2 national awards and 3 provincial first prizes). This means that, in the later stages of competition, participants are effectively competing primarily against their fellow students from the same university for limited institutional quotas.
16
National awards and oral defense requirements
National award ratios are: approximately 0.5%–1% for national first prizes, and approximately 2%–3% for national second prizes. These are evaluated by the national organizing committee. Typically, national second prizes require only paper review and no oral defense, whereas national first prizes require an oral defense.
17
Successful participation award and academic integrity baseline
Teams that submit papers normally will almost certainly receive the "Successful Participation Award" (equivalent to completing the contest). However, plagiarism is strictly prohibited; offenders will be disqualified from all awards and face serious consequences. Data show that approximately 43% of teams receive provincial third prizes or higher.
18
Award-winning ratios: Empirical figures for provincial and national awards
The video summarizes the award outcomes as a set of empirical ratios: approximately 43% of teams receive a provincial third prize or higher, approximately 8% receive a provincial first prize, of which approximately 1% ultimately receive a national first prize, and approximately 2% receive a national second prize. The instructor further combines national first and second prizes, stating that about 100 out of every 3 teams receive a national award, and adds that the actual number may be slightly lower.
43%,8%,1%,2%
19
Preliminary screening rules: Format compliance and similarity thresholds are hard barriers
The first step of the review process is preliminary screening by the provincial organizing committee. The video emphasizes that format violations lead to immediate disqualification, and gives the similarity thresholds: overall text similarity exceeding 25%, or code similarity exceeding 50%. The instructor also interprets excessive similarity as a signal of plagiarism.
overall text similarity rate>25%orcode similarity rate>50%
20
Do not paste large blocks of code directly into the main text
The video specifically cautions that code placed in appendices is typically not subject to intensive similarity checking, but if pasted directly into the main body of the paper, it falls within the scope of similarity detection. Since coding styles tend to be similar across teams, pasting large blocks of code near the end of the main text increases similarity risk; therefore, doing so is not recommended.
21
Independent scoring and provincial first-prize selection ratio
After passing the initial screening, each paper is independently scored by 2 to 3 reviewers. After provincial first prizes are awarded, only the top 10% to 15% of those provincial first-prize papers are selected for national committee review.
2-3reviewers,10%−15%provincial first-prize papers
22
National review and national first-prize defense
During the national review stage, papers are rescored to determine national first and second prizes. The video explicitly states that national first-prize candidates must attend an in-person or online defense to answer questions about their papers; the instructor explains that the core purpose is to confirm whether the paper’s work was completed by the author themselves and whether the author is thoroughly familiar with their own paper.
23
Core scoring dimensions and their weights
Scoring dimensions, ranked by weight, are: model reasonableness 30%–40%, solution correctness 25%–30%, paper presentation 20%–25%, and innovation 10%–15%. The video emphasizes that reviewers place greater importance on justifications for model selection, data processing, reasonableness of assumptions, sensitivity analysis, and data support for conclusions—not on the apparent sophistication of algorithms.
30%−40%,25%−30%,20%−25%,10%−15%
24
Day-one schedule of the competition process
At the end, the video displays day one of the ‘three-day 74-hour’ competition process: from 18:00 to 18:30, receive the problem and read it quickly; from 18:30 to 20:00, discuss topic selection; from 20:00 to 22:00, finalize the topic and search literature; from 22:00 to 01:00, construct the modeling framework for the first question.
18:00-18:30,18:30-20:00,20:00-22:00,22:00-01:00
Detailed learning notes
Explore conditions, steps and evidence. Supplementary explanations are labeled separately from content shown in the video.
Symbols · 9
A/B/C
Clear evidence
Shown in the video
Evidence
Diagram
Observation
The document lists 'Must-read past 3 years of A/B/C problem real exam questions and outstanding papers before the competition'
Symbol
A/B/C
Meaning
Problem type categories in mathematical modeling competitions
Domain
Competition problems
%
Clear evidence
Shown in the video
Evidence
Diagram
Observation
The table lists percentage data such as 'approximately 8%', 'approximately 15%', and 'approximately 20%'
The handout states, "Approximately 43% of teams receive provincial third prize or higher."
Symbol
43%
Meaning
Proportion of teams receiving provincial third prize or higher
Domain
Proportion of teams
8%
Clear evidence
Shown in the video
Evidence
Formula
Observation
The handout states, "Approximately 8% receive provincial first prize."
Symbol
8%
Meaning
Proportion of teams receiving provincial first prize
Domain
Proportion of teams
1%
Clear evidence
Shown in the video
Evidence
Formula
Observation
The handout states, "Among them, approximately 1% ultimately receive national first prize."
Symbol
1%
Meaning
Proportion of teams ultimately receiving national first prize
Domain
Proportion of teams
2%
Clear evidence
Shown in the video
Evidence
Formula
Observation
The handout states, "Approximately 2% receive national second prize."
Symbol
2%
Meaning
Proportion of teams receiving national second prize
Domain
Proportion of teams
25%
Clear evidence
Shown in the video
Evidence
Formula
Observation
The lecture notes preliminary screening clause states "full-text similarity > 25%"
Symbol
25%
Meaning
Full-text similarity threshold
Domain
Paper text similarity rate
50%
Clear evidence
Shown in the video
Evidence
Formula
Observation
The lecture notes preliminary screening clause states "code similarity > 50%"
Symbol
50%
Meaning
Code similarity threshold
Domain
Code similarity rate
10%-15%
Clear evidence
Shown in the video
Evidence
Formula
Observation
Clause 3 of the lecture notes review process states "the top 10%–15% of provincial first-prize papers are sent to the national organizing committee for re-evaluation"
Symbol
10%-15%
Meaning
Proportion range of provincial first-prize papers selected for national re-evaluation
Domain
Proportion of provincial first-prize papers
Knowledge points · 32
Overview of the CUMCM Crash Course
Clear evidence
Shown in the video
Evidence
Audio
Observation
The instructor introduces the course objective as 'helping you quickly master this year's China Undergraduate Mathematical Contest in Modeling (CUMCM)' and explains that the lecture notes address common problem types and challenging issues.
Diagram
Observation
The screen displays the PDF lecture outline, including sections titled 'Introduction: China Undergraduate Mathematical Contest in Modeling (CUMCM)', 'Evaluation Models', 'Classification Models', 'Clustering Models', 'Prediction Models', 'Optimization Models', 'Mechanistic Models', 'LaTeX Writing', 'Quick Word Writing', and 'Key Points for Abstract Writing'.
Definition
Explanation
This segment is the opening of a crash course designed for the China Undergraduate Mathematical Contest in Modeling. The instructor introduces, via the lecture outline, the common model types covered in the course (e.g., evaluation, classification, clustering, prediction, optimization, and mechanistic models), as well as writing and formatting topics (LaTeX, Word, abstracts).
Conditions
Designed for students participating in the China Undergraduate Mathematical Contest in Modeling
Core Definition and Judging Criteria of the China Undergraduate Mathematical Contest in Modeling (CUMCM)
Clear evidence
Shown in the video
Evidence
Audio
Observation
The instructor defines the contest: 'In one sentence, the national contest means three people form a team and produce one paper within three days.' They emphasize the judging criteria: 'making the judges feel the solution is reasonable, sophisticated, and fully completed.'
Diagram
Observation
The screen displays lecture text: 'In one sentence: the national contest = three-person team + three days (74 hours) + one paper. It is not about who computes correctly, but about whose paper makes the judges feel "reasonable, sophisticated, and fully completed."' The instructor circles key terms on screen using a red pen.
Definition
Explanation
The core format of the China Undergraduate Mathematical Contest in Modeling (abbreviated as the national contest) is a three-person team completing one paper within three days (74 hours). Its judging criteria do not solely focus on absolute correctness of computational results, but instead require the paper to demonstrate rigorous logic (reasonable), a certain level of academic sophistication (sophisticated), and full submission (fully completed).
Conditions
Applicable to the China Undergraduate Mathematical Contest in Modeling
CUMCM Eligibility and Team Categories
Clear evidence
Shown in the video
Evidence
Audio
Observation
The instructor explains the contest's nature: 'Only undergraduate and junior college students may participate; graduate students are not allowed. Moreover, these two groups are awarded separately.' They note that undergraduates typically choose Problems A, B, or C, while junior college students choose Problems D or E, and that junior college students find it relatively easier to win awards.
Diagram
Observation
The screen displays lecture text: 'Contest nature: separate awarding for the undergraduate group (undergraduate division) and the junior college group (junior college division); teams consist of three members, with no restriction on majors (typical combinations: mathematics/statistics + computer science + engineering/economics & management); papers (PDF + attachments) must be submitted within 74 hours after problem release.' The instructor highlights 'A, B, C' and 'D, E' with a red pen.
Definition
Explanation
CUMCM is open only to undergraduate and junior college students; graduate students are not eligible. The competition has two categories: the undergraduate category and the junior college category, with separate awards for each. Undergraduate teams typically choose among Problems A, B, and C, while junior college teams typically choose between Problems D and E. Each team consists of three members, with no restrictions on academic background.
Conditions
Participants must be currently enrolled undergraduate or junior college students
Prerequisites
Core Definition and Judging Criteria of the China Undergraduate Mathematical Contest in Modeling (CUMCM)
Content of supporting materials
Clear evidence
Shown in the video
Evidence
Audio
Observation
The instructor mentions that supporting materials include code, data, result plots, result tables, and other intermediate products generated during the paper writing process.
Diagram
Observation
The screen shows handwritten words 'code' and 'Data' beside 'Attachments'.
Definition
Explanation
When submitting a paper for the Mathematical Modeling Contest, in addition to the paper itself (in PDF format), supporting materials (attachments) must also be submitted. These supporting materials mainly include the code and data used to solve the problem, as well as intermediate outputs such as result plots and result tables.
Conditions
Submission procedure applicable to the National College Student Mathematical Modeling Contest (National Contest)
Basis for evaluation and oral defense requirements
Clear evidence
Shown in the video
Evidence
Audio
Observation
The instructor emphasizes that 'the paper is the sole basis for evaluation', and explains that provincial third-, second-, and first-prize winners do not require oral defense; many regions also waive oral defense for national second-prize winners, while national first-prize winners may still need it.
Diagram
Observation
The screen highlights in red the phrase 'The paper is the sole basis for evaluation — no oral defense required (except for national first prize).'
Method
Explanation
For the National Contest, evaluation relies primarily and exclusively on the submitted paper. Even if some parts of the supporting materials are missing, there is no need for excessive concern. Typically, provincial awards (provincial first, second, and third prizes) and most national second prizes do not require oral defense; only national first prizes may require it, though some regions are gradually eliminating this requirement.
Conditions
Regarding the National Contest evaluation mechanism
Prerequisites
Content of supporting materials
National Competition Problem Types and Characteristics
Clear evidence
Shown in the video
Evidence
Audio
Observation
The instructor introduces each of the five problems—A, B, C, D, and E—describing their respective orientations and solution approaches: Problem A leans toward physics and engineering; Problem B toward social economics; Problem C involves big data; Problems D and E are exclusive to vocational college students.
Diagram
Observation
The screen displays 'Table 1: National Contest problem types and representative problems', listing the orientation, typical scenarios, and methods involved for Problems A through E.
Definition
Explanation
The National Competition usually releases 4–5 problems each year (Problems A/B/C/D/E), and participating teams choose any one. The specific classification is as follows:
- Problem A: leans toward physics/engineering, involving heat conduction, stress analysis, differential equations, finite element methods, numerical simulation, etc.; typically advanced.
- Problem B: leans toward social/economic topics, involving traffic optimization, resource allocation, policy evaluation, scheduling, etc.; commonly uses optimization models, graph theory, game theory.
- Problem C: leans toward data/statistics—i.e., big-data problems—covering classification and recognition, prediction, comprehensive evaluation, etc.; commonly uses machine learning, statistical modeling, data analysis.
- Problems D and E: usually exclusive to the vocational college division, covering path planning, production scheduling, environmental assessment, etc.; although they may appear complicated, their actual question difficulty is low.
Conditions
Summary of Recent National Competition Question-Setting Patterns
Topic Selection Strategy Recommendations
Clear evidence
Shown in the video
Evidence
Audio
Observation
The presenter mentions that Problem C attracts the most participants; its low entry barrier leads to small differentiation, making it difficult to stand out and win a National First Prize.
Diagram
Observation
At the bottom of the screen appears 'Topic Selection Strategy', listing advantages, disadvantages, and recommendations for Problem C versus Problems A/B.
Method
Explanation
When selecting a problem type, Problem C (the data problem) has a low entry barrier and thus attracts the most participants (approximately 40%), resulting in low differentiation and making it difficult to stand out and win a National First Prize. In contrast, Problems A and B, though more difficult, offer greater differentiation; performing well on them makes it easier to compete for a National Award. Teams with physics or engineering backgrounds are advised to prioritize Problems A or B; teams with pure mathematics, statistics, or computer science backgrounds will find Problem C the safest choice; interdisciplinary teams should consider Problems B or E.
Conditions
Intended for teams preparing for the National Competition to conduct initial topic selection planning
Prerequisites
National Competition Problem Types and Characteristics
Topic Selection Strategy for Mathematical Modeling Competitions
Clear evidence
Shown in the video
Evidence
Audio
Observation
The presenter analyzed the characteristics of different problems: Problem C is selected by many teams but has low discrimination, whereas Problems A and B are selected by fewer teams but have high discrimination, and provided topic selection recommendations tailored to teams' disciplinary backgrounds.
Diagram
Observation
The screen displays text titled 'Topic Selection Strategy', including analyses of the proportion, difficulty, and discrimination of each problem type, as well as specific team composition recommendations.
Method
Explanation
Select an appropriate competition problem based on the team's disciplinary background and problem characteristics. Problem C (data-based problem) has a low entry barrier but low discrimination, making it suitable for teams with pure mathematics/statistics/computer science backgrounds seeking stability; Problems A and B offer high discrimination and greater potential to win national awards, making them suitable for teams with physics/engineering backgrounds; interdisciplinary teams are recommended to choose Problem B or Problem E.
Conditions
A team usually consists of 3 members.
Consider members' disciplinary backgrounds and complementary skills.
Question-Setting Patterns for Mathematical Modeling Competitions
Clear evidence
Shown in the video
Evidence
Audio
Observation
The presenter summarized the question-setting patterns of the national competition over the past ten years (2015–2024), including non-repetition every three years but with cyclical recurrence, mapping of current social/technological hot topics, the annual presence of a data-based problem, and the typical inclusion of 3–4 sub-questions per problem.
Diagram
Observation
The screen displays text titled '1.3 Question-Setting Patterns', detailing four patterns along with their specific manifestations and examples.
Method
Explanation
Trends in question setting derived from analyzing past official competition problems: 1. Problems from the same field do not appear consecutively, but similar problem types tend to reappear in new forms every 3–4 years; 2. Current social or technological hot topics are highly likely to appear in that year's problems; 3. Since 2015, Problem C has been consistently assigned to big-data or statistical modeling, and there is always one purely data-based problem each year; 4. Each problem typically contains 3–4 sub-questions, with difficulty increasing progressively, and incomplete solutions to the first two sub-questions do not affect scoring for the latter two.
Conditions
Analysis based on official national competition problems from 2015–2024.
Targeted preparation suggestions
Clear evidence
Shown in the video
Evidence
Audio
Observation
The instructor orally provided targeted preparation suggestions, including reviewing real exam questions, accumulating code templates, and preparing paper templates
Diagram
Observation
The document lists three specific preparation suggestions
Method
Explanation
The video gives three preparation suggestions for mathematical modeling competitions: 1. Must-read past 3 years of A/B/C problem real exam questions and outstanding papers before the competition to understand question-setting style and evaluation criteria; 2. Accumulate common model code templates across various fields in advance; 3. Prepare a universal paper template.
Conditions
Intended for students preparing to participate in mathematical modeling competitions
Two-tier review mechanism
Clear evidence
Shown in the video
Evidence
Audio
Observation
The instructor explained that the national competition adopts a two-tier evaluation mechanism: provincial-level selection followed by national-level review
Diagram
Observation
The document's title is "1.4 Award-winning Rate and Awarding Mechanism", and states "The national contest implements a two-tier review mechanism: provincial selection plus national review."
Definition
Explanation
The national contest implements a two-tier review mechanism: provincial selection plus national review. That is, provincial awards are determined first, followed by national awards. After submitting a paper, it proceeds all the way through without requiring resubmission of a new paper.
Conditions
Applicable to the China Undergraduate Mathematical Contest in Modeling
Proportions and review scope of provincial awards
Clear evidence
Shown in the video
Evidence
Audio
Observation
The instructor explained the proportions of provincial first-, second-, and third-prize winners, emphasizing that provincial awards are reviewed only within each province, so competitors are limited to students from the same province.
Diagram
Observation
The table lists approximately 8% for provincial first prizes, approximately 15% for provincial second prizes, and approximately 20% for provincial third prizes; all reviews are conducted by provincial organizing committees.
Formula
Explanation
The approximate proportions of provincial awards are: provincial first prizes at about 8%, provincial second prizes at about 15%, and provincial third prizes at about 20%. These awards are reviewed by each provincial organizing committee, and competition is restricted to teams from the same province.
Conditions
Based on the tabular data provided in the video
Prerequisites
Two-tier review mechanism
Claims and conditions · 6
Scope of competition for provincial awards
Clear evidence
Shown in the video
Evidence
Audio
Observation
The instructor explicitly stated that provincial awards are evaluated only within the province, so contestants do not compete against those from other provinces or municipalities—making it feel like a school-level contest.
Proposition
Statement
Provincial awards are evaluated solely within the province; contestants effectively compete only against other teams from the same province.
Hypotheses
The national contest employs a two-tier evaluation mechanism: provincial selection followed by national review.
Quantifiers
All provincial awards
Nature of competition under the quota system
Clear evidence
Shown in the video
Evidence
Audio
Observation
The instructor explained that the quota system means the final competition often occurs among students from the same school.
Proposition
Statement
Due to the quota system, contestants ultimately compete against fellow students from their own school.
Hypotheses
The national contest implements a quota system, allocating award quotas to individual schools.
Quantifiers
In most cases
Format violations lead to immediate exclusion during preliminary screening.
Clear evidence
Shown in the video
Evidence
Audio
Observation
The instructor says, 'Format violations result in immediate disqualification.'
Formula
Observation
The handout's preliminary screening criteria list 'exclusion due to format violation' as the first item.
Proposition
Statement
During the provincial organizing committee's preliminary screening stage, format violations cause a paper to be disqualified immediately.
Hypotheses
The paper enters the provincial organizing committee's preliminary screening stage.
Quantifiers
For all papers entering preliminary screening, the video states the rules in an authoritative tone.
Plagiarism detection exceeding the threshold is explained by the instructor as a signal of plagiarism.
Clear evidence
Shown in the video
Evidence
Audio
Observation
The instructor says, 'Excessive similarity detection—insufficient originality—indicates plagiarism.'
Formula
Observation
The handout specifies 'excessive similarity detection (overall text similarity > 25% or code similarity > 50%)'.
Uncertainties
'Indicates plagiarism' is the instructor's explanatory statement; the video does not provide finer-grained criteria for determination.
Proposition
Statement
The video explains 'plagiarism detection exceeding the threshold' as a signal indicating potential plagiarism, and provides the threshold values for overall text similarity rate and code similarity rate.
Hypotheses
The paper enters the plagiarism detection stage.
The similarity rate is calculated according to the definition given in the video.
Quantifiers
The video does not clarify whether this criterion is a strict necessary and sufficient condition.
Judges prioritize modeling reasoning over algorithmic difficulty.
Clear evidence
Shown in the video
Evidence
Audio
Observation
The instructor says, 'What the judges care about is not how sophisticated the algorithm is, but rather: Why was this model chosen? How was the data processed? Are the assumptions reasonable? Was sensitivity analysis performed? Is the conclusion supported by data?'
Formula
Observation
The original sentence appears verbatim in the lecture notes.
Proposition
Statement
The video asserts that, during scoring, judges place greater emphasis on the rationale for model selection, data processing, reasonableness of assumptions, sensitivity analysis, and data support for conclusions—not on whether the algorithm is sophisticated.
Hypotheses
The paper is evaluated according to the core dimensions described in the video.
Quantifiers
The video states this in an overall evaluative tone, without providing formal quantitative proof.
Core judging criteria for the competition
Clear evidence
Shown in the video
Evidence
Diagram
Observation
Text box at the bottom of the screen reads: "The judges evaluate whether you 'solved all the problems', not whether you 'used an extremely sophisticated model for the first question.'"
Audio
Observation
Instructor says: "So the judges evaluate whether you solved all the problems—not whether you used an extremely sophisticated model for the first question."
Proposition
Statement
Judges in mathematical modeling competitions primarily assess whether a team has solved all the given problems, rather than the complexity or sophistication of the model used for any single problem.
Hypotheses
Contest teams face a problem set containing multiple sub-problems
Quantifiers
Applies universally to all participating teams
Derivations and proofs · 2
From the ratio of national first prizes to national second prizes, it follows that roughly three teams per hundred receive national awards.
Clear evidence
Shown in the video
Evidence
Formula
Observation
The lecture notes state 'Approximately 1% ultimately win the National First Prize, and approximately 2% win the National Second Prize.'
Audio
Observation
The instructor says, that is,100 out of every 3 teams, about
Uncertainties
The instructor then adds orally, "It might be slightly fewer—perhaps around 1 to 2 teams," indicating this is an empirical estimate, not a strict identity.
Numerical verification
Steps
Expression
1%+2%=3%
Explanation
Adding the proportions of national first prizes and national second prizes yields the total proportion of teams receiving national awards.
Justification
The video treats "national first prize" and "national second prize" as parallel components of "national awards."
Shown in the video
Expression
100×3%=3
Explanation
Under a sample of 100 teams, the total proportion is converted into a team count, yielding approximately 3 teams.
Justification
Multiplying a percentage by the total gives the corresponding quantity.
Derived from the video
Conclusion
According to the ratio given in the video,100 out of every 3 teams, about
Explaining the risk of code similarity based on the distinction between appendix and main text.
Clear evidence
Shown in the video
Evidence
Audio
Observation
The instructor says, "Actually, code similarity does not matter… If you put it in the appendix, they won’t check it. But if you put it in the main paper, then unfortunately, they will check yours."
Audio
Observation
The instructor continues, "Don’t paste large blocks of code at the end… Some students love doing this—pasting dozens of pages."
Uncertainties
This derivation relies on the instructor’s oral explanation of how plagiarism detection is performed; the video does not show the plagiarism-detection system interface or any official document.
Intuitive argument
Steps
Expression
code placed in the appendix⇒rarely checked
Explanation
The instructor states that if code is placed in an appendix, it is typically not closely checked during plagiarism detection.
Justification
From the video's oral explanation.
Shown in the video
Expression
code included in the main body of the paper⇒will be checked
Explanation
The instructor states that if code is pasted directly into the main text of a paper, it will fall within the scope of plagiarism detection.
Justification
From the video's oral explanation.
Shown in the video
Expression
extensive pasting of code⇒increases the risk of similarity
Explanation
Because students' code implementations tend to be similar, pasting large blocks of code into the main text increases the risk of high similarity scores; therefore, the instructor advises against doing so.
Justification
Inferred from the previous two steps and the instructor's explanation that 'AI-generated code tends to be quite similar'.
Derived from the video
Conclusion
Based on this, the video recommends against pasting large amounts of code directly at the end of a paper's main text, to avoid triggering plagiarism detection issues.
Worked examples · 1
Quota system example
Clear evidence
Shown in the video
Evidence
Audio
Observation
The instructor gave an example: a certain school is highly likely to receive only two national awards and three first-tier provincial awards this year.
Diagram
Observation
The instructor handwrote the numbers 2 and 3 on the screen to illustrate quota allocation.
Problem
Explaining how the quota system affects actual competition
Given
A certain school is allocated a limited number of quotas this year
Goal
Demonstrating how the quota system leads to intra-school competition
Steps
Expression
2a national award,3a provincial first prize
Explanation
Suppose a certain school is highly likely to receive only 2 national awards and 3 provincial-level first prizes this year
Justification
Instructor gives an oral example
Shown in the video
Answer
Participants will ultimately realize they are actually competing against their own schoolmates.
Verification
Logical deduction consistent with the quota system
Visual events · 12
Scrolling display of the lecture notes' table of contents
Clear evidence
Shown in the video
Evidence
Animation
Observation
The mouse rapidly scrolls downward in the PDF document, displaying the table of contents from page 1 to page 13, covering various mathematical models and writing guidelines.
Objects
PDF document page
Mouse pointer
Changes
Page content rapidly switches from top to bottom as the mouse scrolls
Invariants
The entire document is structured as a table of contents
Interpretation
Rapid scrolling demonstrates the richness and systematic nature of the course content, giving the audience a broad overview of the upcoming modules.
Core concepts highlighted with a red pen
Clear evidence
Shown in the video
Evidence
Animation
Observation
The instructor uses a red digital pen to underline or circle phrases on screen—'three-member team', 'three days', 'one paper', 'logical', 'proficient', and 'feasible'—and draws arrows connecting these concepts.
Objects
Lecture notes text
Red pen trajectory
Changes
Red dynamic annotations and connecting lines added to static text
Using visual emphasis techniques to guide the audience’s attention to the national competition’s core formal requirements and evaluation orientation, reinforcing memory.
Handwritten labeling of problem set groupings
Clear evidence
Shown in the video
Evidence
Animation
Observation
The presenter uses a red electronic pen to write "A, B, C" beside "Undergraduate Group" and "D, E" beside "Vocational College Group".
Objects
Lecture notes text
Red handwritten characters
Changes
Red handwritten English letters added in blank areas
Visually demonstrates the differences in contest problems faced by participants of different academic levels, supporting oral explanation.
Handwritten annotations specifying supporting material content
Clear evidence
Shown in the video
Evidence
Animation
Observation
On a PDF document, the presenter uses a red pen tool to write 'code' and 'Data' beside the word 'Appendix', then draws an arrow below indicating these are part of the supporting materials.
Objects
Red pen
Text 'code'
Text "Data"
Changes
Handwritten text appears in the blank space
Connecting arrows are drawn
Invariants
The underlying PDF text remains unchanged
Interpretation
Visual annotations emphasize the specific composition of supporting materials to assist oral explanation.
Display and annotate the problem-type table
Clear evidence
Shown in the video
Evidence
Animation
Observation
The screen scrolls downward to reveal a table containing detailed information for problems A–E; the presenter circles and underlines different rows of the table with a red pen.
Objects
Table "Table 1: National Competition Problem Types and Representative Problems"
Red underline
Red circle
Changes
Page scrolling
Table content is highlighted
Invariants
Keep the table structure unchanged.
Interpretation
Visually present the tendencies, contexts, and methods of various problem types to help viewers quickly identify the characteristics of different problem types.
PDF document explanation and annotation
Clear evidence
Shown in the video
Evidence
Diagram
Observation
Throughout the video, a PDF document is displayed, and the presenter uses a red pen to circle and annotate key information to emphasize it.
Objects
PDF document
red pen
Changes
The presenter circles key terms (e.g., "data/statistics", "low discrimination", "no repetition for three years") on the document according to the spoken content and adds brief annotations (e.g., "optimize") in blank areas.
Invariants
The overall structure and main text of the document remain unchanged.
Interpretation
Visual guidance helps viewers focus on core points of the explanation, enhancing information delivery efficiency.
Document annotation process
Clear evidence
Shown in the video
Evidence
Animation
Observation
The instructor uses a red pen to circle, check, and write numbers on the PDF document
Objects
PDF document
Red pen
Changes
The instructor repeatedly circles key terms in the document (e.g., 'simple', 'two questions', 'save one')
Checkmarks appear after list items
Handwritten digits '2, 3'
Invariants
The document's basic layout and content remain unchanged
Interpretation
Real-time annotations guide viewers to focus on core information and data proportions
The award-proportion numbers are progressively circled for emphasis.
Clear evidence
Shown in the video
Evidence
Animation
Observation
Red handwritten circles and arrows sequentially highlight the numbers '43%', '8%', '1%', '2%', and '3 teams'.
Diagram
Observation
The awards table remains visible at the top of the page, and the main-text section titled 'Interpretation of Key Numbers' is highlighted with a circle.
Objects
Approximately 43% of teams receive a provincial third prize or higher.
Approximately 8% receive a provincial first prize.
“Approximately 1% will ultimately win the National First Prize”
“Approximately 2% will win the National Second Prize”
Every 100 teams, about 3 teams receive a national award;
Changes
Red circles and arrows sequentially highlight different proportions
The viewer’s focus shifts from the entire text to the key numbers
Invariants
The main page content remains the same lecture notes PDF
The textual content of the numbers themselves remains unchanged
Interpretation
Visual annotations organize scattered proportions into a hierarchical relationship—from provincial awards to national awards—and emphasize the summary figure “approximately 3 out of 100 teams”
The review process steps are connected in order by arrows
Clear evidence
Shown in the video
Evidence
Animation
Observation
The page scrolls downward from “Key Number Interpretation” to the “Review Process” list
Diagram
Observation
Red arrows point downward in sequence along items 1, 2, 3, 4, and 5, circling the keywords “Initial Screening”, “2–3 Reviewers”, “10%–15%”, “Re-review”, and “Oral Defense”
Objects
Review process items 1 through 5
red downward-pointing arrow
circle-highlighted key terms
Changes
page position moves from top to bottom
arrows label items sequentially according to the workflow order
key terms are circled or underlined
Invariants
workflow item numbering remains unchanged
text content is not rewritten
Interpretation
the animation presents the review process as a linear sequence of stages: preliminary screening, independent scoring, provincial first-prize selection, national re-evaluation, and national first-prize defense
transition from evaluation dimensions to competition schedule
Clear evidence
Shown in the video
Evidence
Animation
Observation
the page continues scrolling down to 'core evaluation dimensions (ordered by weight)' and '1.5 competition timeline (three days, 74 hours)'
Diagram
Observation
red circles highlight 'reasonableness of the model', 'correctness of the solution', 'clarity of the paper', 'innovativeness', and the time period for Day One
Uncertainties
the competition timeline section displays only Day One; subsequent schedules do not fully appear in this segment
The page transitions from judging rules to scoring criteria, then to the schedule
Red highlights sequentially emphasize each dimension and time slot
Invariants
Still consecutive pages of the same handout document
Heading hierarchy remains clear
Interpretation
Visually forms a continuous structure of "how to evaluate—what to evaluate—how to proceed", linking scoring criteria with preparation schedule.
PDF document scrolling and annotation.
Clear evidence
Shown in the video
Evidence
Animation
Observation
The screen continuously scrolls downward to display the PDF document content, while key text is circled in red.
Objects
PDF document
red pen
Changes
page continuously scrolls downward
red pen circles key time points and survival rules
Invariants
overall document structure remains unchanged
Interpretation
visual guidance helps viewers quickly locate the core content of the lecture.
Instructor's real-time annotation process
Clear evidence
Shown in the video
Evidence
Animation
Observation
The instructor uses a red pen tool to annotate the PDF document in real time, including circling keywords (e.g., 'solve first', 'refine later', 'results'), checking off items, underlining with wavy lines, and writing supplementary notes (e.g., 'Is it good?').
Objects
PDF lecture slide page
Red pen stroke trajectory
Changes
The instructor continuously adds red circles and underlines to the text, adjusting in real time with the pace of explanation to guide visual focus.
Invariants
The basic layout and black printed text of the PDF page remain unchanged.
Interpretation
Dynamic visual emphasis helps viewers quickly identify core strategies and common pitfalls within lengthy text.
Misconceptions · 10
Misconception: The National Contest solely values absolute correctness of computational results.
Clear evidence
Shown in the video
Evidence
Audio
Observation
The instructor explicitly states: "The competition is not about who gets the calculation absolutely right, but about making the judges feel that your paper is logical, sophisticated, and feasible to complete."
Diagram
Observation
The lecture notes state: "It's not about who calculates correctly, but about whose paper makes the judges feel it is 'logical, sophisticated, and feasible to complete.'"
Misconception
Many beginners believe that mathematical modeling competitions depend entirely on the accuracy of mathematical calculations and a single, definitive correct answer.
Clarification
The video emphasizes that although results are important and reference standard answers exist, the core evaluation criteria are the paper's completeness, the reasonableness of its logic ('logical'), and the demonstrated level of modeling ability ('sophisticated'). Even with minor computational imperfections, a paper that maintains internal logical consistency, features in-depth analysis, and is completed on time can still receive high praise.
A misconception about missing supplementary materials.
Clear evidence
Shown in the video
Evidence
Audio
Observation
The presenter mentions that some students worry about forgetting to include something in the supplementary materials, then says, 'There is no need to worry excessively, because the paper is the sole basis for evaluation.'
Misconception
The belief that incomplete supplementary materials (appendices) will severely affect competition performance.
Clarification
In fact, the paper is the sole basis for evaluation; supplementary materials serve only to support claims, and even if something is missing, there is no need for excessive anxiety—no need to spend excessive effort perfecting the appendices.
Misconception that provincial awards involve nationwide competition
Clear evidence
Shown in the video
Evidence
Audio
Observation
The instructor explicitly clarifies that viewers need not worry about competing with top performers from cities such as Beijing or Shanghai, since such comparisons do not occur
Misconception
Participants may believe provincial awards require competing against outstanding teams from across the country.
Clarification
Provincial awards are evaluated solely within the province, and competitors are limited to teams from the same province.
Misconception that multiple paper submissions are required during the review process
Clear evidence
Shown in the video
Evidence
Audio
Observation
The instructor emphasizes that no new paper appears midway; once submitted, it passes through smoothly without requiring resubmission
Misconception
Participants may assume they must resubmit or revise their paper between provincial and national reviews.
Clarification
Once the paper is submitted, it directly enters the subsequent review process and does not need to be resubmitted midway.
Mistakenly believing that writing a good paper is sufficient
Clear evidence
Shown in the video
Evidence
Audio
Observation
The instructor says, "Many students ignore formatting; they think, 'As long as I write my paper well, that’s enough,' but format violations result in immediate disqualification."
Misconception
Some participants believe that writing a good paper is enough and that formatting need not be taken seriously.
Clarification
The video explicitly states that format violations lead to immediate elimination during the initial screening stage; formatting is a hard requirement.
Misconception: code is checked regardless of where it is placed.
Clear evidence
Shown in the video
Evidence
Audio
Observation
The instructor said, 'If you put it in the appendix, they won’t check it; if you put it in the paper, then, sorry—it’s you who gets checked,' and reminded students, 'Don’t paste large blocks of code at the end.'
Uncertainties
This explanation comes from the instructor’s practical experience; the video does not display the official plagiarism-checking policy text.
Misconception
Some participants mistakenly believe that placing code in either the appendix or the main text has the same impact on plagiarism checking.
Clarification
The video emphasizes that code placed in the appendix is typically not subject to intensive plagiarism checking, whereas large blocks pasted directly into the main text fall squarely within the scope of plagiarism detection—thus, large blocks of code should not be pasted extensively at the end of the main text.
Misconception: the more advanced the algorithm, the higher the score.
Clear evidence
Shown in the video
Evidence
Audio
Observation
The instructor said, 'What reviewers value is not how sophisticated an algorithm you used, but why you chose this model…'
Formula
Observation
The exact same sentence appears in the handout.
Misconception
Some participants treat 'using a highly sophisticated algorithm' as the primary scoring advantage.
Clarification
The video points out that scoring focuses more on the rationale for model selection, data processing, reasonableness of assumptions, sensitivity analysis, and data-supported conclusions—not on the apparent complexity of the algorithm.
Do not write the paper on Day One.
Clear evidence
Shown in the video
Evidence
Audio
Observation
A common mistake many people make is spending Day One frantically searching for references and outlining ideas while writing not a single word of the paper, only to realize on the afternoon of Day Three that only one-third of the paper is done.
Caption evidence
Observation
A common mistake: spending Day One frantically searching for references and outlining ideas while writing not a single word of the paper.
Misconception
Spending Day One only searching for references and outlining ideas—without writing the paper—leads to rushed work later and a disastrously written abstract.
Clarification
You should start writing the introduction, problem restatement, and model framework on the evening of Day One, advancing modeling and writing simultaneously.
Mutually waiting for teammates
Clear evidence
Shown in the video
Evidence
Audio
Observation
I am waiting for the modeler to provide formulas; I am waiting for the programmer to produce results; I am waiting for the writer to integrate everything—everyone waits, stuck in a deadly loop.
Caption evidence
Observation
"I'm waiting for the modeling teammate to provide the formula" → "I'm waiting for the programming teammate to produce the results" → "I'm waiting for the writing teammate to integrate everything"—this is a vicious cycle of mutual waiting.
Misconception
Team members work sequentially, each waiting for the others' final results before starting their own part.
Clarification
All three should work in parallel: once a rough plan is available, start writing immediately—do not wait in sequence.
The misconception of over-pursuing perfection on part (a)
Clear evidence
Shown in the video
Evidence
Audio
Observation
The instructor describes how some students are "stubborn to a fault... insisting on wrestling with it head-on", sensing the result is unreasonable yet unable to pinpoint why—causing them to spend the first two days solely on part (a), then rushing through the later parts.
Diagram
Observation
The on-screen text states Many teams pursue perfection on the first question — tuning parameters until dawn4trying different models, switching among three models to compare them. Then they realize they have no time left for questions two and three.
Misconception
Believing one must apply the most complex model and repeatedly tune parameters on the first question to achieve so-called 'perfect' results, thereby neglecting overall progress management.
Clarification
Having a result matters100times more than pursuing perfection. One should first build a baseline model using simple methods to ensure answers for all questions, then optimize later if time and energy permit.
Concept relations · 16
Overview of the CUMCM Crash Course → Core Definition and Judging Criteria of the China Undergraduate Mathematical Contest in Modeling (CUMCM)
Clear evidence
Shown in the video
Evidence
Audio
Observation
The instructor first introduces the overall course structure, then proceeds to the first subsection: "Getting Started: China Undergraduate Mathematical Contest in Modeling (CUMCM)."
Contains
Explanation
The course overview (k1) includes detailed explanations of the contest's specific definitions and rules (k2, k3); k2 and k3 are concrete elaborations of the "Getting Started" section within k1.
Core Definition and Judging Criteria of the China Undergraduate Mathematical Contest in Modeling (CUMCM) → CUMCM Eligibility and Team Categories
Clear evidence
Shown in the video
Evidence
Audio
Observation
After explaining the basic definition of the National Contest, the instructor immediately details who may participate and how teams are formed.
Prerequisite
Explanation
Understanding the basic format and evaluation criteria of the National Contest (k2) is a prerequisite for further exploring specific eligibility requirements and team categorization (k3).
Content of supporting materials → Basis for evaluation and oral defense requirements
Clear evidence
Shown in the video
Evidence
Audio
Observation
The presenter first explains what supplementary materials include, then immediately clarifies why there is no need to worry excessively about them, leading into the statement 'the paper is the sole basis for evaluation.'
Contrast
Explanation
The secondary role of supplementary materials contrasts with the absolute central role of the paper, explaining why one need not obsess over appendix details.
National Competition Problem Types and Characteristics → Topic Selection Strategy Recommendations
Clear evidence
Shown in the video
Evidence
Audio
Observation
After introducing the features of each problem type, the presenter begins analyzing why many participants choose Problem C, resulting in low discrimination, and thus offers topic-selection advice.
Application
Explanation
Topic-selection strategy is a practical application recommendation based on understanding each problem type's features—such as difficulty, audience background, and discrimination power.
Question-Setting Patterns for Mathematical Modeling Competitions → Topic Selection Strategy for Mathematical Modeling Competitions
Clear evidence
Shown in the video
Evidence
Audio
Observation
After introducing topic-selection strategies, the presenter immediately proceeds to analyze question-setting patterns, explaining that understanding these patterns better guides topic selection and competition preparation.
Application
Explanation
Mastering question-setting patterns (e.g., hotspot mapping, characteristics of data-based problems, problem structure) forms the foundation for developing effective topic-selection strategies.
Two-tier review mechanism → Proportions and review scope of provincial awards
Clear evidence
Shown in the video
Evidence
Diagram
Observation
The document structure first introduces the two-tier review mechanism, then uses a table to list in detail the proportion of awards at each level.
Contains
Explanation
The two-tier review mechanism includes the evaluation process for provincial-level awards.
Two-tier review mechanism → Proportions and evaluation requirements for national awards
Clear evidence
Shown in the video
Evidence
Diagram
Observation
The document structure first introduces the two-tier review mechanism, then uses a table to list in detail the proportion of awards at each level.
Contains
Explanation
The two-tier review mechanism also includes the evaluation process for national-level awards.
Quota system → Proportions and review scope of provincial awards
Clear evidence
Derived from the video
Evidence
Audio
Observation
After explaining the scope of provincial awards, the instructor immediately introduces the quota system to further narrow the competition range.
Application
Explanation
The quota system is applied to the allocation of provincial-level awards, confining the competition further to students within the same school.
Key numbers for competition award proportions → Initial screening rules by the provincial organizing committee.
Clear evidence
Shown in the video
Evidence
Animation
Observation
The screen scrolls from 'Interpreting Key Numbers' to 'Review Process'.
Audio
Observation
After explaining ratios, the instructor says, 'Now, let’s look at the review process.'
Application
Explanation
The video first presents the award ratios, then shifts to the review process, explaining that these ratios are outcomes generated during subsequent stages: initial screening, group review, re-review, and oral defense.
Selection ratio from provincial first prize to national re-evaluation → National re-evaluation and national first-prize oral defense requirements
Clear evidence
Shown in the video
Evidence
Formula
Observation
Handout Article 3 states, 'After provincial first prizes are awarded: the top 10%–15% of provincial first-prize papers are sent to the national organizing committee for re-review.'
Audio
Observation
The instructor explains, 'After provincial first prizes… this is the selection stage; papers are sent to the national organizing committee for re-review.'
Prerequisite
Explanation
Only after provincial first prizes are awarded and papers are selected from the top 10%–15% do they enter the national re-review and subsequent national first- and second-prize evaluations.
Core evaluation dimensions and their weights → National re-evaluation and national first-prize oral defense requirements
Clear evidence
Derived from the video
Evidence
Formula
Observation
Handout Article 5 states that national first-prize candidates must 'answer questions about their papers.'
Audio
Observation
The instructor then explains that judges focus on questions such as 'Why was this model chosen? How were the data processed? Are the assumptions reasonable? Was sensitivity analysis performed? Is the conclusion supported by data?'
Uncertainties
This connection is a reasonable inference based on the video’s preceding and following content; the video does not explicitly state in one sentence, 'These are the exact questions asked during the oral defense.'
Application
Explanation
The core scoring dimensions presented in the video align closely with the instructor’s description of key oral-defense questioning points, indicating that both the oral defense and scoring center on model reasonableness, solution methodology, exposition, and argumentation.
Core evaluation dimensions and their weights → Day One schedule of the competition timeline
Clear evidence
Shown in the video
Evidence
Animation
Observation
The page scrolls from 'Core Scoring Dimensions' to '1.5 Competition Schedule (Three Days, 74 Hours).'
Audio
Observation
The instructor says, 'Then, the competition schedule—generally, we’ll just take a brief look at it.'
Application
Explanation
After explaining the scoring criteria, the video transitions to the competition timeline, implying that Day One’s activities—topic selection, topic confirmation, and modeling framework development—are designed to support the later scoring dimensions of model reasonableness and paper exposition.
Find an answer · 30
Which common algorithmic models does the National Mathematical Modeling Contest crash course cover?
Clear evidence
Supplementary explanation
Evidence
Audio
Observation
The instructor mentions that the course will cover evaluation, classification, clustering, prediction, optimization, mechanism modeling, and writing and formatting.
Knowledge points
Overview of the CUMCM Crash Course
What are the basic format and evaluation criteria of the China Undergraduate Mathematical Contest in Modeling (CUMCM)?
Clear evidence
Supplementary explanation
Evidence
Audio
Observation
The instructor explains that the National Contest requires three people to write one paper in three days, and judges value papers that are logical, sophisticated, and feasible to complete.
Knowledge points
Core Definition and Judging Criteria of the China Undergraduate Mathematical Contest in Modeling (CUMCM)
Can graduate students participate in the CUMCM? How do the contest problems differ between undergraduate and junior college students?
Clear evidence
Supplementary explanation
Evidence
Audio
Observation
The instructor clarifies that undergraduate and junior college students are awarded separately, with distinct contest problems assigned to each group, and that junior college students find it relatively easier to win awards.
Knowledge points
CUMCM Eligibility and Team Categories
Does solving calculations correctly guarantee an award in mathematical modeling contests?
Clear evidence
Supplementary explanation
Evidence
Audio
Observation
The instructor corrects the misconception that "the contest is about who computes correctly," emphasizing instead the logic and completeness of the paper.
Knowledge points
Misconception: The National Contest solely values absolute correctness of computational results.
What supplementary materials must be submitted for the China Undergraduate Mathematical Contest in Modeling?
Clear evidence
Shown in the video
Evidence
Audio
Observation
The presenter lists code, data, and result plots, among others.
Knowledge points
Content of supporting materials
Do national contest reviewers focus on the paper or the attachments? Is a defense required?
Clear evidence
Shown in the video
Evidence
Audio
Observation
The presenter emphasized that the paper is the sole basis for evaluation and outlined the defense process.
Knowledge points
Basis for evaluation and oral defense requirements
What topics do Problems A through E in the national contest cover? Which one is the most difficult?
Clear evidence
Shown in the video
Evidence
Audio
Observation
The presenter described the subject areas and difficulty levels of Problems A through E separately.
Knowledge points
National Competition Problem Types and Characteristics
Why is it not advisable to choose Problem C blindly?
Clear evidence
Shown in the video
Evidence
Audio
Observation
The presenter analyzed the competitive landscape for Problem C.
Knowledge points
Topic Selection Strategy Recommendations
In mathematical modeling competitions, how should teams with different academic backgrounds select contest problems?
Clear evidence
Shown in the video
Evidence
Audio
Observation
The presenter thoroughly analyzes the advantages, disadvantages, and target audiences of various types of problems.
Knowledge points
Topic Selection Strategy for Mathematical Modeling Competitions
What are the patterns and features of question setting in the China Undergraduate Mathematical Contest in Modeling?
Clear evidence
Shown in the video
Evidence
Audio
Observation
The presenter summarizes question-setting patterns over the past decade.
Knowledge points
Question-Setting Patterns for Mathematical Modeling Competitions
How should one prepare before the Mathematical Modeling Competition?
Clear evidence
Derived from the video
Evidence
Audio
Observation
Instructor's spoken preparation advice.
Knowledge points
Targeted preparation suggestions
What is the national competition's review mechanism? Is it necessary to submit the paper midway?
Clear evidence
Derived from the video
Evidence
Diagram
Observation
Document title and main text explanation.
Knowledge points
Two-tier review mechanism
Coverage and review notes
Covered · Course opening and lecture notes table of contents display.
Covered · Explaining the core definitions and evaluation criteria of the CUMCM, correcting the misconception that only results matter.
Covered · Explaining eligibility requirements, team category divisions, and differences in contest problems.
Covered · Explaining the specific content of supporting materials.
Covered · Explaining the evaluation criteria and defense rules, clarifying anxieties about attachments.
Covered · A transitional paragraph mentioning the value of participation (high cost-effectiveness), without elaborating on specific mathematical concepts, categorized under overall background introduction.
Covered · Providing a detailed overview of the tendencies and methods associated with each problem type from A to E.
Covered · Offering topic-selection strategies based on problem-type characteristics.
Covered · Explaining the two-tier review mechanism and the paper submission process.
Covered · Explaining provincial award ratios and the intra-provincial scope of competition.
Covered · Explaining the quota system and the de facto intra-school competition.
Covered · Explaining national award ratios and the defense requirements.
Covered · Explaining the Successful Participation Award and the consequences of plagiarism.
Covered · Explains award percentages and summarizes 1% + 2% as approximately 3 teams per hundred receiving a national award.
Covered · Proceeds to Step 1 of the evaluation process, focusing on formatting violations and plagiarism thresholds, and explains how code placement affects plagiarism detection.
Covered · Step 2 of the review process: Each paper is independently scored by 2 to 3 reviewers.
Covered · Step 3 of the review process: After provincial first prizes are awarded, the top 10% to 15% are sent to the national organizing committee for re-evaluation.
Covered · Steps 4 and 5 of the review process: National re-evaluation involves re-scoring to determine national first and second prizes; candidates for national first prize must give an oral defense.
Covered · Explain the core scoring dimensions and their weights, emphasizing that reviewers value modeling and reasoning over algorithmic sophistication.
Covered · Transition to the competition schedule title, displaying and briefly describing the Day One timetable.
Covered · Explaining the overall four-day time plan and key milestones for the competition.
Covered · Introducing essential tools required for the National Competition.
Covered · Explaining survival rules every beginner must know, as well as common pitfalls.
Covered · Building on the previous segment about not waiting for teammates, the instructor introduces the next survival rule.
Covered · A detailed explanation of Survival Rule Three, illustrated with a negative example showing the harm of pursuing perfection in the first question.
Covered · Proposing the correct approach: first build a baseline model, emphasizing the importance of results and the possibility of subsequent optimization.
Covered · Summarizing the judges' evaluation criteria, comparing model depth with problem completion.
Covered · Display and briefly explain the pre-contest checklist, concluding the introductory norms section.
