You are only deciding IVP or BVP. The rule: are all conditions glued to one point, or spread across two different points?
Recall Solution 1.1
Both conditions sit at the same point x=1 (one is the value, one is the slope — both at x=1).
⇒IVP.
Key idea: "initial" is about one location, not about the number 0.
Recall Solution 1.2
Conditions at two different points x=0 and x=π/2.
⇒BVP (a two-point problem).
Recall Solution 1.3
One condition at x=2, the other at x=5 — different points.
⇒BVP (a mixed one, since it uses a value at one end and a slope at the other, but still a BVP because the locations differ).
Here the endpoints are chosen to make detM=0. Your job: decide no solution vs infinitely many, and say why.
Recall Solution 3.1
y=c1cosx+c2sinx.
y(0)=c1=0.
y(π)=c1cosπ+c2sinπ=−c1+0=0 — this just repeats c1=0 and says nothing about c2.
detM=cos0⋅sinπ−sin0⋅cosπ=1⋅0−0⋅(−1)=0.
Because the right-hand side (0,0) is compatible, we get infinitely many solutions:
y(x)=c2sinx,c2arbitrary.Picture: look at the orange sine curves in the figure — all pass through both endpoints x=0 and x=π.
Recall Solution 3.2
y(0)=c1=0.
y(π)=−c1=0, but we demandedy(π)=1. So 0=1 — impossible.
Same singular detM=0, but now the target is incompatible.
⇒No solution.Why: when detM=0, the second equation lives entirely inside the first's shadow; if the requested value doesn't land in that shadow, nothing satisfies both.
Now you build the singular condition yourself: find the special parameter values (eigenvalue-type behaviour) where nontrivial solutions appear. This is the doorway to Sturm-Liouville Theory.
Recall Solution 4.1
Write λ=k2 with k>0, so y=c1coskx+c2sinkx.
y(0)=c1=0⇒y=c2sinkx.
y(L)=c2sinkL=0. For c2=0 we need sinkL=0⇒kL=nπ,n=1,2,3,…
So k=Lnπ and the eigenvalues are
λn=(Lnπ)2,n=1,2,3,…
with eigenfunctionsyn=sin(Lnπx).
Why n≥1?n=0 gives k=0, i.e. y≡0 — the trivial solution we're excluding.
Recall Solution 4.2
From 4.1 with L=2: allowed eigenvalues are λn=(2nπ)2=4n2π2.
n=1:4π2≈2.467 — allowed ✔
n=2:π2≈9.870 — allowed ✔
λ=1: is 1=4n2π2? That needs n=π2≈0.637, not an integer — not allowed (only trivial solution).
λ=4: needs n=π4≈1.273, not an integer — not allowed.
Both constants forced to 0⇒only the trivial solutiony≡0. That counts as a unique solution (the trivial one). No free constant survives, so detM=0 here.
Recall Solution 5.2
Integrate twice: y′=3x2+c1, then y=x3+c1x+c2.
y(0)=c2=0.
y′(1)=3(1)2+c1=3+c1=2⇒c1=−1.
Unique:y(x)=x3−x.Check:y′′=6x ✔; y(0)=0 ✔; y′(x)=3x2−1, y′(1)=2 ✔.
Because one condition is a value at x=0 and the other a slope at x=1 (different points), this is a BVP — but a well-posed one.
Recall Solution 5.3
y=c1cosx+c2sinx.
y(0)=c1=1.
y(π)=−c1+0=−1. We need −c1=−1, i.e. c1=1 — consistent with the first condition.
detM=cos0sinπ−sin0cosπ=0: singular, but the target is compatible.
⇒Infinitely many solutions: y=cosx+c2sinx for any c2.
Contrast with 3.2: same singular matrix, but there the target (0=1) was incompatible ⇒ none. Here c1=1 agrees with both rows ⇒ a whole family.
Recall One-glance summary
All conditions at one point? ::: IVP — solve by reading off c1,c2; always uniquely solvable (Wronskian =0).
Conditions at two points and detM=0? ::: Unique BVP solution.
detM=0 with compatible data? ::: Infinitely many solutions.
detM=0 with incompatible data? ::: No solution.
Where do the special λn=(nπ/L)2 come from? ::: The detM=0 condition of the BVP — Sturm-Liouville Theory eigenvalues.