A slope at each point
For an initial-value problem
we want a picture of the solutions before, or instead of, a formula.
Definition 1.5.1
(Direction field)
Let be defined on a region of the -plane. The direction field of assigns to each point a short line segment through that point with slope .
A differentiable function on an interval solves the equation exactly when its graph lies in and is tangent to the segment at every point it visits. Starting at , we can sketch a plausible solution by following the changing directions. The horizontal axis is time and the vertical axis is the unknown value: a drawn segment describes a slope, not a displacement through a fixed time interval.
Concept lensGeometric
A direction field as a planar vector field
Equivalently, the parametrized graph of a solution is a trajectory of the autonomous planar system
and conversely, along every trajectory of this system, written as a function of is a solution. The drawn segment is the direction of the velocity vector . Its first component never vanishes, so no segment is vertical and every solution graph moves strictly to the right.
For an autonomous equation , the slope is the same across every horizontal line. Its one-dimensional phase line records the sign of as a function of . When also depends on , equal heights can have different slopes at different times, so a single phase line no longer records the whole field.
Isoclines organize a sketch
Definition 1.5.2
(Isocline of slope )
For a real number , the isocline of slope of is the set . On that set all field segments have slope .
An isocline is a set used to organize slopes; it need not itself be a solution curve, and for some values of it may be empty or reduce to a point. In particular, the zero-slope isocline marks points at which a solution has horizontal tangent. It does not generally represent a constant solution. The exact condition is the following.
Proposition 1.5.3
(When a horizontal line is a solution)
Let be a real number and an interval with . The constant function solves on if and only if for every .
Proof
(Compare the two slopes along the line)
The constant function has derivative at every , while the equation requires its derivative at to be . The two agree for every precisely when on all of . Zero at a single time is not enough: the requirement is an identity in .
For an autonomous equation , a root of gives the horizontal isocline , and the proposition shows that is a constant solution: the graph slope and the field slope are both zero. In the nonautonomous case, the zero-slope isocline is usually not a horizontal line at all.
Common mistake
(Following the zero-slope isocline as if it were a solution)
On the line the field of is horizontal, so it is tempting to sketch a solution running along that line. Two different slopes are involved at each point of it: the line itself, as a graph, has slope , while the field prescribes slope . A solution must match the field slope, so it crosses the isocline with a horizontal tangent and cannot follow it.
Worked examples
Worked example 1.5.4
(A zero-slope line that is not a solution)
Consider . The isocline of slope is . The field is horizontal on , points upward where , and points downward where . But the line has derivative , whereas the equation assigns slope at every point of that line. It is therefore not a solution.
The integrating-factor method gives the exact family: from we get , that is,
For we have and hence . This solution crosses the zero-slope isocline at , with . For , ; for , . Thus the solution has a local maximum there, even though the zero-slope line itself rises. Substitution verifies (1), and the linear initial-value theorem supplies uniqueness for each initial point. The member is the line , which is the isocline of slope : it is a solution because its own slope equals the field slope on it.
Worked example 1.5.5
(An autonomous field with empty isoclines)
For the slope depends only on , so every isocline is a union of horizontal lines. The slope- isocline solves , that is,
For it consists of two lines symmetric about ; for it is the single line ; for it is empty, because no point of the plane has slope larger than . The zero-slope isocline is together with , and since for all , the proposition shows that both are constant solutions. Between them the slope is positive, above and below it is negative, so solutions starting in rise toward . The line is not a solution: its graph slope is , while the field slope there is .
Worked example 1.5.6
(Circles as isoclines and a one-point zero isocline)
For the slope- isocline is the circle of radius when . For it is the single point , and for it is empty. Every slope is nonnegative, so every solution is nondecreasing, and a solution can have a horizontal tangent only where it passes through the origin. Outside the unit circle the slope exceeds , so once a solution lies outside it, it rises faster than a line of slope for as long as it stays outside. The sketch therefore shows steepening growth; it does not, by itself, tell how long the solutions exist.
A sketch does not prove uniqueness
A direction field is drawn from finitely many segments, and a smooth-looking picture suggests one curve through each point. That suggestion can be false even for a continuous field.
Counterexample mode
A continuous field with many solutions through one point
Claim: if is continuous, the direction field determines one solution through each point. Consider with . The zero function solves the problem. So does, for every ,
The pieces have matching value and derivative at . On the second piece because , and on the first both sides vanish. Thus the same initial point admits infinitely many solutions. The field is continuous, but its square-root dependence is not locally Lipschitz in at zero: as . With a locally Lipschitz field the existence and uniqueness theorem restores one solution through each point.
A picture of short segments cannot by itself establish uniqueness or tell us how long a solution exists.
Common mistake
(Testing a constant solution at one time only)
For we have , so the origin lies on the zero-slope isocline. This does not make a solution: is nonzero for every . By the proposition above, a constant is a solution on an interval only if at every time of that interval.
Practice
Quick check
(For , find the slope- isoclines. Sketch the solution through qualitatively near that point, and verify your conclusion by solving the initial-value problem.)
Locate the origin relative to the zero-slope isocline first.
Solution · Answer
The isoclines are . The initial point lies on the zero-slope line , but that line is not a solution: its derivative is while the field slope there is . Multiplying by and integrating gives . Its derivative is , negative for and positive for . The solution therefore has a local minimum at the origin. Since is positive for and negative for , the solution crosses the zero-slope line at the origin.
Quick check
(For , is a solution on ? Is ?)
Apply the proposition on horizontal lines.
Solution · Answer
for every , so is a solution. vanishes only at , so is not a solution on any interval; the line meets the zero-slope isocline only at the point .
Quick check
(For , which isocline passes through , and what slope does the solution through that point have there?)
Evaluate at the point.
Solution · Answer
, so the point lies on the isocline of slope , and the solution through it has slope there. From (1), that solution is with , so and , as predicted.
A direction field supports qualitative reasoning; an existence or uniqueness theorem and, where needed, a calculation provide the mathematical justification behind the sketch.