Engineering Mathematics for GATE CSE: Important Topics

Engineering Mathematics carries about 13 marks in GATE CSE 2027, and it draws on a small, stable set of concepts. Eigenvalues, Bayes theorem and graph counting top the list, followed by expectation, probability distributions and propositional logic. The syllabus is unchanged from 2026, so past-paper evidence applies directly.

Key takeaways

• The official weight is about 13 marks, covering Discrete Mathematics, Linear Algebra, Calculus, and Probability and Statistics.
• Eigenvalue questions have appeared in essentially every paper since 2003, and Bayes or conditional probability in every paper since 2010.
• Discrete Mathematics and the other three areas trade places from year to year, so neither half can be dropped.
• NAT questions made up roughly 40–45% of this section's marks in 2021–2026 (approximate), so numeric precision matters.
• Mathematics rewards breadth. 20 of its 25 scored topics sit at 67.5 or higher, so no single topic carries the section.

How many marks is Engineering Mathematics?

The official GATE 2027 pattern allots about 13 of the 100 marks to Engineering Mathematics, against 15 for General Aptitude and about 72 for core CS subjects. Confirm this at gate2027.iitm.ac.in.

Third-party paper analyses often report different totals, because Discrete Mathematics questions are sometimes counted inside this section and sometimes as a separate subject. Approximate analyst figures for combined mathematics:

Discrete Mathematics was the heavier half in 2018, 2020 and 2022; Linear Algebra, Calculus and Probability were heavier in 2016, 2017 and 2025. For how maths compares with other subjects, see GATE CSE subject-wise weightage.

The 2027 syllabus in brief

The wording is unchanged from GATE 2026. The last change was in 2021, when Monoids was added, and Numerical Methods left the paper in 2016.

• Discrete Mathematics: propositional and first-order logic; sets, relations, functions, partial orders and lattices; monoids and groups; graphs (connectivity, matching, colouring); combinatorics (counting, recurrence relations, generating functions).
• Linear Algebra: matrices, determinants, systems of linear equations, eigenvalues and eigenvectors, LU decomposition.
• Calculus: limits, continuity and differentiability, maxima and minima, mean value theorem, integration.
• Probability and Statistics: random variables; uniform, normal, exponential, Poisson and binomial distributions; mean, median, mode and standard deviation; conditional probability and Bayes theorem.

Important topics, tier by tier

Each topic was scored on past-paper frequency (2000–2026 and 2018–2026), variety of question forms, syllabus centrality, recency and format fit. Scores are analytical confidence values, not probabilities.

Tier 1: Very High

Tier 2: High

First-order logic, counting (pigeonhole, inclusion–exclusion), sets and relations, recurrence relations, systems of equations and rank, graph connectivity and trees, limits, graph colouring, determinants and special matrices, functions, maxima and minima, groups and monoids, posets and lattices, and integration.

Tier 3: Moderate

Mean value theorem and continuity, generating functions, matching, LU decomposition, and mean, median, mode and standard deviation. All are named in the syllabus but rarely asked. They are cheap to prepare, so cover them, but never at the cost of Tier 1 practice. Matching is the one with a genuine "due" signal after sparse appearances.

The full scoring table with all eight factor scores for each of the 25 maths topics, a year-by-year concept map, and 54 practice questions with detailed solutions are in the book.

What the Tier 1 topics look like

Eigenvalues. You get eigenvalues, or a small matrix, and must find the determinant or trace of A^k, A^-1, A + kI or a polynomial in A. The shortcuts: the product of eigenvalues is the determinant, the sum is the trace, and p(A) has eigenvalues p(λ). A quick illustration: if a 3×3 matrix has eigenvalues 1, 2 and 3, then A^2 + I has eigenvalues 2, 5 and 10, so det(A^2 + I) = 100. Also know the eigenvalues of triangular, rank-1, idempotent (0 or 1), nilpotent (0) and orthogonal matrices, and Cayley–Hamilton.

Bayes theorem. Two or three sources, such as machines or suppliers, with prior shares and defect rates; find the probability that a defective item came from one of them. Diagnostic-test and dice-conditioning versions also recur. The classic slips are swapping the two conditional probabilities and forgetting the total-probability denominator.

Graph counting. The degree sum equals 2e. For planar graphs, v − e + f = 2 and e ≤ 3v − 6, or e ≤ 2v − 4 when the graph is also bipartite. Expect leaves of a tree from internal degrees, and self-complementary graphs, which need n ≡ 0 or 1 (mod 4).

Expectation and variance. Linearity with indicator variables (expected fixed points, expected distinct faces), expected trials to first success, E[X^2] = Var(X) + (E[X])^2, and Var(aX + b) = a^2 Var(X).

Named distributions. Uniform variance (b − a)^2/12, exponential memorylessness with mean 1/λ, Poisson with the rate rescaled to the time window, and binomial "at least k" probabilities.

Propositional logic. Which formula is or is not a tautology, which is equivalent to p → q, and how many truth assignments make a formula false. Low marks per question, but very reliable.

Traps that keep reappearing

Some of these traps, such as the quantifier pairing and the endpoint check, have been built into wrong options in five or more papers. The GATE CSE formula sheet collects the formulas behind them.

How the questions are asked now

• NAT dominates. Since 2014, counting, eigenvalue, recurrence and probability questions have mostly become numeric-answer questions with no negative marking. Recent analyses call the section calculation-intensive: conceptual depth has not risen, but the precision demanded has.
• MSQ took over "which is true" items. Since 2021, questions on groups, monoids, lattices and quantifier logic often come as MSQs with no partial credit. One wrong inclusion scores zero.
• MCQ survives mainly for first-order logic translation, tautology identification and quantifier equivalences.

See MCQ, MSQ and NAT strategy for how to attempt each type.

How to prepare Engineering Mathematics

1. Start with Tier 1. Six topics, all NAT-friendly, all present in most papers.
2. Then go wide across Tier 2. Fourteen High topics sit just below Tier 1. One standard form of each beats deep study of a few.
3. Link Discrete Mathematics to CS subjects. Recurrences feed Algorithms, logic feeds decidability statements in TOC, and graph theory feeds MST and DFS questions. The Algorithms important topics guide shows where they overlap.
4. Finish with Tier 3 as insurance. LU decomposition, matching and descriptive statistics are cheap to prepare.
5. Use old papers freely. Everything tested in 2003–2010 that is still in the syllabus remains live. The only dead content is Numerical Methods.

IIT Madras, which organizes GATE 2027, also set the 2003, 2011 and 2019 papers. Each had a group-theory item and a planar-graph or lattice item, but so did most other papers, so the organizer adds little beyond the topics above. To compare these maths priorities with other subjects, see most repeated topics in GATE CSE.