Direction Fields

Precise definition

For $y'=f(t,y)$, a direction field places a short segment of slope $f(t_0,y_0)$ at each sampled point $(t_0,y_0)$. A solution curve is a differentiable graph tangent to these segments everywhere it passes. The field visualizes local derivative information; it is not itself a collection of computed solutions.

Notation and mathematical language

At $(t_i,y_j)$, the segment may use direction vector $(1,f(t_i,y_j))$ or a normalized version. A nullcline is a curve where $f(t,y)=0$, so solution tangents are horizontal there. Isoclines satisfy $f(t,y)=c$ for a fixed slope $c$.

Conceptual picture

A solution follows the local arrows continuously, so the field reveals increasing, decreasing, and nearly flat regions. Dense fields can suggest long-run patterns, barriers, and sensitivity to initial values, but line length and plotting window are visual conventions rather than mathematical magnitude.

Conditions and key results

The graph representation assumes the solution can be written locally as $y(t)$. Under uniqueness hypotheses, two solutions through different initial points cannot cross in the $(t,y)$ plane. Where $f$ is undefined or discontinuous, no segment should be drawn and ordinary existence conclusions may fail.

A reliable strategy

1. Identify the differential equation and mark points where $f$ is undefined.
2. Evaluate $f(t,y)$ on a grid and draw short segments with the corresponding sign and relative steepness.
3. Mark nullclines or isoclines, then sketch a smooth curve tangent to segments through the initial point.
4. Check qualitative claims with the equation and, when available, compare against an exact or numerical solution.

Fully worked example

Interpretation and application

Direction fields support quick diagnosis of population, mixing, cooling, and control models before symbolic work. A plotted convergence pattern is qualitative evidence for the model, not a proof of a physical causal mechanism or a precise numerical error bound.

Common mistakes

Verification and reasonableness

• Evaluate $f$ at several points on a sketched solution and compare with its visual tangent.
• Locate all nullcline crossings and confirm the derivative is zero there.
• Overlay an exact solution or a smaller-step numerical approximation when possible.

Practice

1. What is the slope at $(2,3)$ for $y'=t-y$?
2. What equation defines its nullcline?
3. Can two solution curves cross under local uniqueness?

Further deduction

A field can distinguish autonomous from non-autonomous equations. For $y'=f(y)$, all segments on a horizontal line share a slope because $t$ is absent. For $y'=t-y$, slopes change along a horizontal line, but they are constant on diagonal isoclines $t-y=c$. Recognizing such patterns can reveal transcription errors before any solution method is attempted.

A nullcline is not generally a solution curve. On $y'=t-y$, the line $y=t$ has field slope zero, but the line itself has derivative 1, so it cannot be a solution. Solutions cross it with horizontal tangents. Only when a curve's own tangent matches the field everywhere does it define a solution.

Related topics

• Euler's Method
• Existence and Uniqueness
• Autonomous Equations