Why Parameters Must Match at a Boundary
A piecewise formula can use different expressions in different regions of the plane. Even when each expression behaves smoothly within its own region, the complete function may fail to be continuous where the regions meet. Choosing parameters for continuity means ensuring that nearby values fit together at the boundary—not just that the expressions within the regions are individually well behaved.
At a boundary point, there may be three sources of nearby function values: values from one region, values from the other region, and values assigned directly on the boundary. A continuity test must account for all three. The value assigned at the point itself must also agree with their common limit.
A Continuity Criterion at a Boundary Point
Suppose a curve \(\Gamma\) divides a neighborhood of a point \(\mathbf p\) into two regions, \(U_+\) and \(U_-\), and \(\mathbf p\) lies on \(\Gamma\). Let \(F_+\) and \(F_-\) be the formulas used in the two regions, and let \(G\) give the function’s values on the boundary. The function’s value at \(\mathbf p\) is \(f(\mathbf p)\). A limit from one region means a limit restricted to points in that region approaching \(\mathbf p\); the boundary limit is restricted to points of \(\Gamma\) approaching \(\mathbf p\).
the limit of \(F_+\) as points in \(U_+\) approach \(\mathbf p\), the limit of \(F_-\) as points in \(U_-\) approach \(\mathbf p\), and the limit of \(G\) as points on \(\Gamma\) approach \(\mathbf p\).
This criterion is a direct use of the definition of continuity: \(f\) is continuous at \(\mathbf p\) when \(\lim_{\mathbf q\to\mathbf p}f(\mathbf q)=f(\mathbf p)\). It makes explicit that the full limit includes every way of approaching the point, including travel along the boundary itself.
Why the Criterion Works
Here is a proof of the criterion. Necessity follows by restricting a limit that exists over all nearby points. If the full limit is \(f(\mathbf p)\), then approaching through either region or along the boundary must give that same limit.
For sufficiency, suppose the three restricted limits all exist and equal \(L=f(\mathbf p)\). Given \(\varepsilon>0\), the definition of each restricted limit supplies a positive distance within which values from that particular region—or from the boundary—differ from \(L\) by less than \(\varepsilon\). Take the smallest of those three distances. Any point within that distance lies in one of the two regions or on the boundary, so its function value differs from \(L\) by less than \(\varepsilon\). Thus the full limit is \(L=f(\mathbf p)\), which proves continuity.
The boundary part of this reasoning matters: the limits from the two regions alone do not restrict the values assigned on the curve. In applications, if the boundary formula is continuous along the curve at \(\mathbf p\), its boundary limit equals its value there. That still needs to be checked and matched to the other limits and to \(f(\mathbf p)\).
Worked Example: Choosing Parameters at the Origin
Worked Example: Choosing Parameters at the Origin
Define a function using three formulas around the \(x\)-axis:
Find the parameter values that make \(f\) continuous at the origin.
First approach through the region \(y>0\). As \((x,y)\to(0,0)\), \(a+2x-y\to a\). Therefore the limit from above is \(a\).
From the region \(y<0\), \(b+x+3y\to b\), so the limit from below is \(b\).
On the boundary \(y=0\), excluding the origin, the values are \(c+2x\). As \(x\to0\), these values tend to \(c\). The value assigned at the origin is \(5\). Continuity requires all four quantities to agree:
For these parameter values, the upper formula tends to \(5\), the lower formula tends to \(5\), and the boundary formula tends to \(5\), which is also \(f(0,0)\). Hence the criterion gives continuity at the origin. If any of \(a\), \(b\), or \(c\) differs from \(5\), at least one required limit fails to equal the function value.
Choosing Parameters Along a Boundary Curve
Sometimes the goal is continuity at every point on part of a boundary, rather than at a single point. Then the matching conditions must hold at every point of that curve. When the formulas are polynomials, substituting a parameterization of the boundary often turns the conditions into polynomial identities.
For example, the line \(y=x\) can be parameterized by \((t,t)\). Substituting \(x=t\) and \(y=t\) into each formula gives its value along the line. The resulting expressions in \(t\) must agree for every \(t\) in the specified interval, not merely at one selected value.
Worked Example: Continuity Along a Line
Worked Example: Continuity Along a Line
For \(-1<t<1\), consider the boundary line \(y=x\) and the function
Find the parameter conditions for continuity at every point \((t,t)\) with \(-1<t<1\).
On the line, the upper-region formula approaches the value obtained by substituting \(x=t,\ y=t\):
The lower-region formula approaches
The boundary formula itself has value \(et^2+ft+g\). For continuity at each point of the line, these three expressions must agree for every \(t\) in the interval. First match the upper expression with the boundary expression:
Since this identity holds for every \(t\), matching coefficients gives \(e=0\), \(f=a+b\), and \(g=1\). Matching the lower expression with the same boundary expression gives \(e=0\), \(f=2+c\), and \(g=d\). Combining these conditions yields
For one particular choice, take \(a=1,\ b=2,\ c=1\). Then \(a+b=3\) and \(2+c=3\), so set \(f=3\). Set \(e=0\) and \(g=d=1\). Along \(y=x\), the upper formula gives \(3t+1\), the lower formula gives \(3t+1\), and the boundary formula gives \(3t+1\). Thus all three values agree at every \((t,t)\) in the interval. Since each formula is a polynomial and hence continuous in its region, these matches establish continuity at each of those boundary points.
Worked Example: Matching at a Nonzero Boundary Point
Worked Example: Matching at a Nonzero Boundary Point
Suppose the boundary is \(x+y=0\), and the point of interest is \(\mathbf p=(1,-1)\). Let
with the assigned value \(f(1,-1)=2\). Find the conditions for continuity at \((1,-1)\), then give one parameter choice that works.
The first formula tends to its value at \((1,-1)\), because it is a polynomial. Substitution gives \(a(1)^2+(-1)+1=a\). The regional limit from \(x+y>0\) is therefore \(a\).
For the other region, substitution gives \(b(1)(-1)+c=-b+c\). On the boundary, the formula gives \(d(1)+e=d+e\). The assigned point value is \(2\). The conditions are consequently
Choose \(b=1\), which requires \(c=3\). Choose \(d=1\), which requires \(e=1\), and take \(a=2\). Checking the three values: the first regional limit is \(2\); the second is \(-1+3=2\); and the boundary value at the point is \(1+1=2\). Each equals \(f(1,-1)=2\), so the function is continuous at the specified point.
A Common Mistake: Ignoring Boundary Values
A frequent error is to check that the two formulas on either side have matching limits and then conclude that the function is continuous. That conclusion is not justified unless values along the boundary are controlled too.
For a concrete counterexample, define \(f(x,y)=0\) when \(y\ne0\), define \(f(x,0)=1\) for \(x\ne0\), and set \(f(0,0)=0\). From both regions \(y>0\) and \(y<0\), the limit at the origin is \(0\), which equals \(f(0,0)\). But along the boundary, \(f(x,0)=1\) for every \(x\ne0\), so the boundary limit is \(1\), not \(0\). The function is not continuous at the origin.
Turning Continuity into Parameter Equations
A reliable procedure is to separate the different ways of approaching the point before solving for parameters. In particular, do not assume that conditions obtained from a single boundary point will also enforce continuity along an entire curve.
Determine which formulas apply in each region and which formula supplies values on the boundary.
Use the formula for that region and take its limit at the specified point, or substitute the boundary parameterization when checking an entire curve.
Check what the boundary formula approaches and compare it with the value assigned at the point.
For an entire curve, require the expressions to agree for every parameter value on the specified segment.
Substitute the parameters back into all formulas and confirm that every required limit equals the function value.
These checks can produce a unique parameter choice, several valid choices, or no choice at all. If the resulting equations contradict one another, then no parameter values make the function continuous under the stated definition.
Why These Conditions Matter
Continuity at a piecewise boundary concerns the behavior of the function as a whole, not the smoothness of any one formula in isolation. A polynomial formula in each region can still produce a discontinuity if the formulas approach different values, if the boundary formula has a different limit, or if the assigned point value does not match.
Parameter matching makes these requirements explicit. At a single point, it usually produces a finite system of equations. Along a curve, it may produce identities that must hold throughout an interval. In either case, solving the equations is only part of the task: substituting the answer back into every required limit is what confirms that the conditions were applied correctly.
Check Your Understanding
Use the boundary continuity criterion to answer the following questions.
- At a point on a piecewise boundary, which limits must agree with the function value?
- Why does matching the two regional limits not necessarily guarantee continuity?
- For a function defined on both sides of \(y=x\) and on \(y=x\), how can you express the continuity conditions along the line?
- If parameter conditions contradict one another, what does that tell you about continuity for the given family of functions?
- In the nonzero-point example, what would fail if the boundary limit were \(2\) but the assigned value at \((1,-1)\) were \(3\)?