Chapter 1 · Introduction

Chapter 1 — First-Order Ordinary Differential Equations

Before solving differential equations, the questions that decide whether solving makes sense: does a solution exist, is it unique, how long does it live — and what happens when the equation is not solved for $y'$, or comes as a system?

In 1687 Newton wrote the law of motion as a differential equation, and for the next century and a half mathematicians took for granted that a law of change together with a present state determines the future. Cauchy, in the 1820s, was the first to ask the question as mathematics: given $y'=f(x,y)$ and $y(x_0)=y_0$, is there a solution, and is there only one? The answers, completed by Lipschitz, Peano and Picard, are among the most used theorems in applied mathematics. They explain why a well-posed physical model can be simulated at all, why a numerical method converges to the right answer, and why some perfectly smooth models — a chemical reaction feeding on its own heat, a population with cooperative breeding — run off to infinity in finite time.

The same questions take a new form when the equation arrives as a relation $F(x,y,y')=0$ that gives several slopes at each point. Then whole curves can exist along which uniqueness fails: singular solutions, the envelopes of the general solution, which Clairaut met in 1734 while studying the geometry of curves and which appear today as caustics in optics and as the boundary of the region a projectile can reach. Finally, an equation of order $n$ is a system of $n$ first-order equations in disguise, and for systems the existence theory carries over unchanged, while linear systems with constant coefficients can be solved completely with eigenvalues and the matrix exponential — the starting point of stability theory and of the phase portraits used across science and engineering.

IVP ⇔ integral equation→Lipschitz condition→Picard iterates→Existence & uniqueness→Maximal interval, blow-up→Gronwall→Singular solutions→Clairaut→Systems, eᵗᴬ
🎯 By the end of this chapter you can…
  • rewrite an initial value problem as an integral equation, compute Picard iterates, and prove the Picard–Lindelöf theorem with the interval $h=\min(a,b/M)$;
  • decide whether a given $f(x,y)$ is Lipschitz in $y$ on a given set, and produce the standard examples of non-uniqueness ($y'=3y^{2/3}$, $y'=\sqrt{|y|}$) and of blow-up ($y'=y^2$, $y'=1+y^2$);
  • find the maximal interval of existence of a solution and use Gronwall's inequality to bound the effect of changes in the data;
  • solve exact, linear and Bernoulli equations and equations of higher degree in $y'$ (solvable for $p$, $y$ or $x$);
  • find singular solutions from the $p$- and $c$-discriminants, recognise envelopes, tac-loci, node-loci and cusp-loci, and solve Clairaut's and Lagrange's equations and orthogonal-trajectory problems;
  • reduce higher-order equations to first-order systems and solve linear systems $x'=Ax$ through eigenvalues and $e^{tA}$, sketching and classifying their phase portraits (§1.3).
Teaser — One Slope Field, Three Different Fates

Each short segment shows the slope $f(x,y)$ prescribed at that point; solution curves follow the segments. Press Play to grow the solutions from left to right. In the first field every curve lives forever and they never meet; in the second some curves escape to infinity before reaching the right edge; in the third, curves merge into the axis and leave it again — through one point pass many solutions. Chapter 1 explains all three.

front at x =—
curves still alive—
curves that blew up—
uniqueness—

How this is examined

This chapter supplies a steady stream of CSIR NET questions. In Part B the typical question asks for the interval of existence guaranteed by Picard's theorem on a stated rectangle, for the maximal interval of a solution that can be found explicitly (a separable, Bernoulli or Riccati-type equation that blows up), for the number of solutions of an IVP such as $y'=|y|^{\alpha}$, $y(0)=0$, or for the singular solution of a Clairaut-type equation and the orthogonal trajectories of a family. Part C, where several options can be correct, prefers statements to be tested: "this $f$ is Lipschitz on that set", "the IVP has a unique solution on all of $\mathbb{R}$", "two solutions can cross", "the discriminant locus is a solution", "the Picard iterates converge on a given interval", "the fundamental matrix of a linear system is invertible", "the origin is a stable node". The wrong options are almost always one of the standard counterexamples collected in the ⚠️ boxes of the three sections. GATE MA asks similar questions, often in numerical-answer form.

Sections in this chapter

✅ Before you start

You should be able to solve the standard first-order equations met at BSc level (separable, linear, homogeneous) and be comfortable with the real analysis of the CSIR NET Unit 1 course: uniform convergence and the Weierstrass M-test, the mean value theorem, continuity on compact sets (a continuous function on a closed rectangle is bounded), and — for §1.2 — the implicit function theorem. Section 1.3 also uses eigenvalues, eigenvectors and diagonalisation of $2\times2$ and $3\times3$ matrices. If this is your first visit, read the welcome page for how each section is organised and how to use the 🎯 Exam corner.