Stochastic processes and applications DSBA 2026/2027
Course goals
侍には目標がなく道しかない [Samurai niwa mokuhyō ga naku michi shikanai]
A samurai has no goal, only a path.
Telegram chat (не берёт на парковке)
Random walks. Occasional enlightenment.
The chat may not work in the parking lot. The Markov property still does.
Course information
Official course information: SP 2026/27 · TSA 2026/27.
Course links
| Resource | Link |
|---|---|
| Telegram course chat | SP–TSA course chat |
| Gradebook | SP-TSA 26-27 |
| Course materials | Sources of Wisdom — notes, videos, assignments and reading |
| SP syllabus (ПУД) | SP syllabus — 2025/26 archive |
| TSA syllabus (ПУД) | TSA syllabus — link from the 2026/27 course page |
Teaching team and contacts
Lecturers
| Name | Teaching responsibilities | Telegram |
|---|---|---|
| Пётр Лукьянченко | Lectures | @PetrLukianchenko |
| Мария Кириллова | Lectures; seminars for groups 241, 242 | @makirill |
| Алёна Числова | Lectures; seminars for groups 243, 244, 245 | @Alyona_Chislova |
Teaching assistants
| Name | Responsibilities / groups | Telegram |
|---|---|---|
| Иван Новосад | Lecturer's assistant | @Mellodizzz |
| Ева Каримова | Groups 243, 245 | @ekaesha |
| Аскар Биктибаев | Group 244 | @askbkt |
| Ольга Макогонова | Groups 241, 242 | @komfajx |
Rules of the game
Gradebook calculation
The calculation below follows the SP-TSA 26-27 gradebook.
Random processes, deterministic weights.
- Final (100-point scale) = round(0.25 × Q + 0.35 × M1 + 0.40 × M2*, 2).
- Final (10-point scale) = round(Final (100-point scale) / 10, 0).
Here Q is the aggregate quiz score, M1 is Scaled Mid1, and M2* is Scaled Mid2 unless a Scaled Retake score has been entered. When a retake score is present, it replaces Scaled Mid2.
| Item | Rule |
|---|---|
| Quizzes | Each quiz uses a 10-point scale. All eight quizzes count equally: Q = min(100, round((q1 + … + q8) / 8 × 10, 0)). No lowest scores are dropped. |
| Mid1 and Mid2 | The scores recorded in the gradebook use a 100-point scale and contribute 35% and 40%, respectively. |
| Retake | The entered Scaled Retake score replaces Scaled Mid2, including when it is lower. The quiz component and Scaled Mid1 are retained. |
| Rounding | Q is rounded to an integer before weighting. The weighted total is rounded to two decimal places, then divided by 10 and rounded to an integer. For nonnegative grades, halfway values round upwards: 6.5 becomes 7. |
| Incomplete results | The quiz result remains blank until at least one numeric quiz score is entered. The divisor remains 8, so unfilled quizzes contribute zero to the displayed calculation. The final grade remains blank until Q, Scaled Mid1 and either Scaled Mid2 or Scaled Retake are available. |
Stochastic Processes: Halloween, Ded Moroz and honey
In the course tradition, Mid1 is the Halloween Exam and Mid2 is the Ded Moroz Exam.
Two seasonal encounters on the samurai path.
Home assignments and quizzes
Home assignments are normally released weekly. You do not need to submit regular home assignments: the following class includes a quiz with one or two problems closely resembling the assignment. Once during the course, a home assignment takes the form of a computer-assisted project.
Three honey pots — use wisely.
At the end of the SP course, you have 3 honey pots, giving you the right to rewrite five missed or poorly completed quizzes. Honey pots consumed are recorded separately in the gradebook; this count is not a separate weighted component of the final grade.
Sources of Wisdom
Enlightenment is not guaranteed. The reading list is a good start.
The resources below come from earlier course pages. Assignment files and recordings retain their original academic-year labels.
Course materials
| Resource | Links |
|---|---|
| Home assignments | SP 2025/26 · SP–TSA 2024/25 · 2023/24 |
| Lecture and class notes | 2025/26 notes · 2024/25 lecture slides and notes · Maria Kirillova's notes — 2024/25 |
| Videos — made with love | 2025/26 recordings · 2024/25 class recordings · Practice playlist |
| Past exams | Past exam collection · Сливы осеннего экзамена 2026 |
| Course whitepaper | Whitepaper linked on the 2024/25 and 2025/26 course pages |
Probability and Markov chains
- StoPro — Problems in Stochastic Processes
- Introduction to Probability — Blitzstein and Hwang
- Cambridge lectures on Markov chains
- Takis Konstantopoulos — One Hundred Solved Exercises on Markov Chains
- Representing Markov Chains in LaTeX
- Convergence modes — Cornell review
- Convergence modes with examples — Saravanan Vijayakumaran
- Statistics Cookbook
Monte Carlo and MCMC
- Chib and Greenberg — Understanding the Metropolis–Hastings Algorithm
- Casella — Explaining the Gibbs Sampler
- Roberts and Rosenthal — General State Space Markov Chains
- Visualization of MCMC methods
- Charles Geyer — MCMC lecture notes
Stochastic calculus
- Zastawniak — Basic Stochastic Processes.
- ICEF master's course exams
- ICEF master's course notes — Russian
- Draft textbook — Russian
- Draft problem collection — Russian
Time series
- Forecasting: Principles and Practice — R
- Shumway and Stoffer — Time Series Analysis
- Ruey Tsay — teaching materials
- Aad van der Vaart — Time Series lecture notes
- Aad van der Vaart — Time Series course page
- Time series problem collection — Russian
Unobserved components and state-space models
- Harvey–Jaeger example — Detrending, Stylized Facts and the Business Cycle
- João Tovar Jalles — Structural Time Series Models and the Kalman Filter
- Harvey — Forecasting with Unobserved Components Models
- Chad Fulton — state-space modelling
- Rob Hyndman — State Space Models
Previous paths
The path continues, almost surely.