Engineering Mathematics I | Semester 1 BTech CSE Core Notes
Semester 1 - Page 1: Engineering Mathematics I. A general BTech study guide to linear algebra, single-variable calculus, multivariable calculus, integration and vector calculus. Semester placement differs. This is not an official syllabus or a prediction of exam questions. Each numbered topic ends with an answered exam question.
Topic map
- Matrices: read the structure before calculating
- Determinants and Gauss-Jordan inverse
- Echelon form, rank, consistency and rank-nullity
- Gaussian elimination and Gauss-Seidel iteration
- Eigenvalues, eigenvectors and diagonalization
- Cayley-Hamilton, quadratic forms and the power method
- Limits, continuity, differentiability and tangent lines
- Rolle, Lagrange and Cauchy mean value theorems
- Higher derivatives, Leibniz and local series
- Indeterminate limits, curve tracing, curvature and envelopes
- Partial derivatives, total derivative and Euler theorem
- Jacobians, functional dependence and approximation error
- Several-variable extrema and Lagrange multipliers
- Double and triple integrals: choose the region first
- Area, volume, mass, centroid and improper integrals
- Gradient, directional derivative, divergence and curl
- Line integrals and conservative fields
- Green, Gauss divergence and Stokes theorems
- ODE bridge: where a different course puts differential equations
1. Matrices: read the structure before calculating
A matrix is a rectangular arrangement of numbers. Its order is rows x columns. Addition requires equal orders; AB requires columns(A)=rows(B). AB and BA need not agree, and one may not exist. A transpose swaps rows and columns. Symmetric means A transpose = A; skew-symmetric means A transpose = -A. Orthogonal real matrices satisfy A transpose A=I, so their inverse equals their transpose. A Hermitian complex matrix equals its conjugate transpose, not merely its transpose.
Answer: For an m x n matrix times an n x p matrix, the result is m x p. Match the two inner dimensions first.
2. Determinants and Gauss-Jordan inverse
For a 2 x 2 matrix [[a,b],[c,d]], det=ad-bc. A square matrix is invertible exactly when its determinant is nonzero. The inverse is (1/det)[[d,-b],[-c,a]]. For larger matrices, put [A|I] side by side and use elementary row operations until the left side becomes I. The right side is then the inverse. A zero row preventing a pivot means a singular matrix.
Answer: The rows or columns are dependent, so A collapses at least one nonzero vector to zero. An inverse cannot recover that lost information.
3. Echelon form, rank, consistency and rank-nullity
Use row swapping, multiplying a row by a nonzero number, and adding a multiple of another row. Rank is the number of pivots in echelon form, not the number of nonzero entries. Ax=b is consistent when rank(A)=rank([A|b]). For n unknowns, equal ranks=n give a unique solution; equal ranks less than n give free variables and infinitely many solutions. Unequal ranks give no solution. Rank-nullity says rank(A)+dimension(null(A))=n.
Answer: No. Check consistency first. Infinite solutions require equal coefficient and augmented ranks; unequal ranks mean no solution.
4. Gaussian elimination and Gauss-Seidel iteration
Elimination creates zeros below pivots, then back-substitution starts with the final equation. Keep the right-hand side with every row operation. Gauss-Seidel is iterative: solve each equation for one unknown, use the newest available values in the same sweep, and stop against a residual/tolerance. Strict diagonal dominance is a useful sufficient condition for convergence, not a necessary one.
Answer: Gauss-Seidel uses a new component immediately within the sweep; Jacobi computes the entire next vector from the previous vector.
5. Eigenvalues, eigenvectors and diagonalization
An eigenvector v is nonzero and satisfies Av=lambda v. Solve det(A-lambda I)=0 for eigenvalues, then solve (A-lambda I)v=0 for vectors. Eigenvalues can be complex or repeated. The trace equals their sum and determinant their product, counting multiplicity. A matrix is diagonalizable when it has n linearly independent eigenvectors: A=PDP inverse. Repeated eigenvalues alone do not decide this.
Answer: A times zero equals lambda times zero for every lambda, so it carries no information about a special direction.
6. Cayley-Hamilton, quadratic forms and the power method
Cayley-Hamilton says a square matrix satisfies its own characteristic polynomial. It reduces high powers to lower powers; if the constant term permits it, it also gives an inverse. For a real symmetric matrix, orthogonal diagonalization converts a quadratic form to a sum of eigenvalue-weighted squares. All positive eigenvalues mean positive definite; mixed signs mean indefinite. The power method repeatedly multiplies and normalizes; convergence needs an appropriate dominant eigenvalue and a nonzero initial component in its direction.
Answer: No. Read the signs of the eigenvalues of its symmetric matrix, or use a valid definiteness test. A positive example does not establish a general rule.
7. Limits, continuity, differentiability and tangent lines
A limit describes nearby values, not necessarily the value at the point. Continuity requires the function value to equal the finite limit. Differentiability requires a finite derivative and implies continuity, but continuity need not imply differentiability. Check left and right behavior at corners. A tangent at x=a has equation y-f(a)=f prime(a)(x-a).
Answer: No. Absolute value at0 is the simplest counterexample: it has a corner.
8. Rolle, Lagrange and Cauchy mean value theorems
Rolle requires continuity on[a,b], differentiability inside, and equal endpoint values. It guarantees some interior c with f prime(c)=0. Lagrange keeps the first two conditions and guarantees f prime(c)=(f(b)-f(a))/(b-a). Cauchy applies to two functions with the same regularity: (f(b)-f(a))g prime(c)=(g(b)-g(a))f prime(c). Use a quotient form only when its denominators are nonzero.
Answer: State continuity, differentiability, interval and any endpoint condition first; only then solve for the guaranteed point.
9. Higher derivatives, Leibniz and local series
The nth derivative of a product is the sum from k=0 to n of binomial(n,k) times the kth derivative of u times the(n-k)th derivative of v. Taylor about a expands using derivatives at a; Maclaurin is a=0. A finite Taylor formula has a remainder. A power series represents a function only where convergence and equality hold. Do not use an infinite series outside its valid interval.
Answer: They are the same construction at different centers: Maclaurin is Taylor centered at zero.
10. Indeterminate limits, curve tracing, curvature and envelopes
For 0/0 or infinity/infinity limits, L Hospital can replace the quotient by the derivative quotient when its hypotheses hold and that new limit exists. It is not a license for differentiating every quotient. Curve tracing uses domain, symmetry, intercepts, asymptotes and sign of derivatives. For y=f(x), curvature radius is(1+(f prime)^2)^(3/2)/absolute(f double prime) when the denominator is nonzero. For an envelope F(x,y,c)=0, eliminate c with partial F/partial c=0 and check the resulting locus.
Answer: It can include extra branches or singular points. Substitute back into the family and check the geometric contact condition.
11. Partial derivatives, total derivative and Euler theorem
A partial derivative changes one variable while holding others fixed. A total derivative along a path includes every varying input. For z=f(x,y), dz/dt=f_x dx/dt+f_y dy/dt. Euler theorem for a differentiable homogeneous function of degree n gives x f_x+y f_y=nf. Mixed partial derivatives agree under suitable continuity assumptions; their mere existence is not enough.
Answer: Differentiating the other independent variable as if it depended on the chosen variable. Hold it fixed unless a path or dependency is stated.
12. Jacobians, functional dependence and approximation error
The Jacobian determinant measures local area or volume scaling for a coordinate map. A nonzero determinant gives local invertibility under the inverse function theorem conditions. A zero determinant alone at one point does not prove global functional dependence. For small input changes, df is a first-order estimate, not an exact finite change.
Answer: Volume and area weights must be nonnegative; the determinant sign records orientation, not negative physical area.
13. Several-variable extrema and Lagrange multipliers
Find stationary points by setting all first partials to zero. For two variables, D=f_xx f_yy-(f_xy)^2: D>0 with f_xx>0 gives a local minimum, D>0 with f_xx<0 a local maximum, D<0 a saddle, D=0 inconclusive. On a closed bounded region, also examine boundary points. For a regular equality constraint g=c, use gradient f=lambda gradient g, then check candidate values and boundary/singular cases.
Answer: No. Boundary extrema and points where derivatives do not exist also need inspection.
14. Double and triple integrals: choose the region first
A double integral accumulates values over an area; a triple integral over a volume. Draw or describe the domain before setting limits. For a continuous function over a rectangle, iterated integration is valid in either order. Changing order over a nonrectangular region changes the limits, not the function. Polar coordinates add the factor r; cylindrical add r; spherical add rho squared sin phi with the stated angle convention.
Answer: Omitting the Jacobian or retaining old bounds after changing variables.
15. Area, volume, mass, centroid and improper integrals
Integrating 1 gives area or volume. Integrating a density gives mass. A centroid with uniform density is a geometric average; a center of mass weights positions by density. For a planar region, x bar=(integral x density dA)/mass. Improper integrals need limiting definitions and convergence checks. Gamma(a)=integral 0 toinfinity t^(a-1)e^(-t)dt for a>0; Gamma(n)=(n-1)! for positive integers. Beta(a,b)=Gamma(a)Gamma(b)/Gamma(a+b) for positive a,b.
Answer: Nonuniform density weights some areas more heavily; use mass-weighted moments.
16. Gradient, directional derivative, divergence and curl
For scalar f, gradient is the vector of first partial derivatives. Directional derivative in unit direction u is gradient f dot u; normalize a supplied direction first. Divergence of a vector field is a scalar measuring local net outward tendency. Curl is a vector measuring local rotation. For sufficiently smooth fields, curl(gradient f)=0 and divergence(curl F)=0.
Answer: No. Gradient maps a scalar to a vector; divergence maps a vector to a scalar; curl maps a3D vector to a vector.
17. Line integrals and conservative fields
For a parametrized curve r(t), work is integral F(r(t)) dot r prime(t) dt. A conservative field F=gradient potential has work equal to potential(end)-potential(start); closed-loop work is zero. Curl zero implies conservativeness under suitable domain assumptions such as a simply connected domain, not automatically on a region with holes.
Answer: When a valid potential exists over the relevant domain. Do not infer it solely from a curl calculation on a punctured region.
18. Green, Gauss divergence and Stokes theorems
Green relates a positively oriented closed plane curve integral of P dx+Q dy to the area integral of Q_x-P_y. Gauss relates outward flux across a closed boundary to the volume integral of divergence. Stokes relates circulation around an oriented surface boundary to the surface flux of curl. Check smoothness, domain and orientation conditions. A boundary curve is not the same thing as a closed surface.
Answer: Gauss divergence theorem. Stokes uses curl and a boundary curve; Green is the planar circulation form.
19. ODE bridge: where a different course puts differential equations
Some first-year outlines put ODEs in Mathematics I, others in II. A first-order linear equation y prime+P(x)y=Q(x) uses integrating factor exp(integral P dx). An exact differential M dx+N dy=0 has M_y=N_x on an appropriate region and a potential with those partials. Bernoulli y prime+Py=Q y^n uses v=y^(1-n) when n differs from 0,1 and the transformation is valid. A constant-coefficient linear ODE uses its characteristic polynomial; repeated roots add factors of x.
Answer: Course placement varies. Follow your own outline; use the linked Mathematics II guide for the longer differential-equation topic set.
Formula selection: what kind of answer should you get?
| Question | Method | Final check |
|---|---|---|
| Ax=b | Elimination, rank test | Substitute every equation |
| Av=lambda v | Characteristic polynomial, nullspace | v is nonzero; verify Av |
| Local change | Derivative or gradient | Units, path, unit direction |
| Area or mass | Integral with region and density | Bounds and Jacobian |
| Closed-boundary flux | Divergence theorem if conditions hold | Outward orientation |
Numerical self-check and solved answers
- Rank and nullity of a 3-column matrix of rank 2? 2 and 1.
- Eigenvalue product for the example matrix [[4,1],[2,3]]? 10.
- Lagrange mean value point for x squared on [1,3]? c=2.
- For f=x squared y, partial derivative with respect to y? x squared.
- Integral of x+y on the 1 by 2 rectangle above? 3.
- Unit-cube outward flux of (x,y,z)? 3.
Coverage and further study
This page adds the common gaps beyond the earlier matrix-and-tangent summary: mean value theorems, higher derivatives, series, partial/total derivatives, Jacobians, constrained extrema, multiple integrals and vector theorems. The ODE bridge covers a different first-year placement; the longer continuation is Mathematics II. Named methods here are introductory worked explanations, not substitutes for full exercises and proofs in a textbook.
Reference reading: partial derivatives, double integrals, divergence theorem, linear algebra lecture notes. Original worked examples and exam questions are for practice, not copied past papers.
Comments
Post a Comment