Intuition
When several independent states share the same eigenvalue, finding that value does not say which of them the system is in. The measurement then keeps the part of the state inside that eigenvalue’s whole subspace, superposition and all, and throws away the rest. A second, compatible observable can then split the subspace further.
Sorting mail by city keeps each city’s letters in one pile without ordering them by street. The pile is everything with that city on it; sorting by street is a second, finer question.
Measuring a degenerate eigenvalue
Let be the projector onto all eigenvectors of with eigenvalue . The probability of the result is the squared length of the part of the state in that subspace, and the state after the measurement is that part, normalised. This refinement of the Born rule is called the Lüders rule.
How it works
- With an orthonormal basis of the eigenspace, : the probabilities of the states sharing the value add.
Probabilities over an eigenspace add
Write the projector onto the eigenspace as the sum of the ket-bras of an orthonormal basis of it, and sandwich it in the state. Each term becomes the modulus squared of one overlap, so the probability of the degenerate result is the sum of the probabilities of the states that share it — the same number whichever basis of the eigenspace is used.
Proof steps
The projector onto a subspace is the sum over an orthonormal basis of it.
Put it between the bra and the ket of the state.
Each term is an overlap times its conjugate.
The projector is fixed by the subspace alone, so the probability is too.
Applications
Practice
Add Over the Eigenspace
The probability of a degenerate result is the sum of the probabilities of all the states that share it.
Try it
in the basis and . What is the probability of the result 1?
The State Afterwards
After a degenerate result the state is the projection onto the eigenspace, normalised. The superposition inside the eigenspace survives.
Try it
With and , what is the state after the result 1?
No Degeneracy, No Change
For a non-degenerate eigenvalue the eigenspace is a line, its projector is a single ket-bra, and the rule is the Born rule of the second lesson.
Try it
For a non-degenerate eigenvalue, the Lüders rule gives the same probability and state as the Born rule.
A Second Observable
A compatible observable that takes different values on the states of a degenerate eigenspace splits it. Measuring it afterwards does not change the first result.
Try it
After the result 1 of , which measurement could single out or without disturbing the value of ?
Try it
and . After the result 3, what is the coefficient of in the new state? Give three decimal places.
Try it
After a degenerate result, the system is always left in one of the basis vectors chosen for that eigenspace.
Try it
An observable has eigenvalue 4 on an eigenspace spanned by orthonormal and . In some state and . What is the probability of the result 4?
Final checkpoint
Try it
Why does the probability of a degenerate result not depend on the orthonormal basis chosen inside its eigenspace?
Try it
The projectors onto the eigenspaces of all the distinct eigenvalues of an observable add up to the identity.
Try it
After the result 1 of leaves , the observable is measured. What is the probability of the result 7?
Completion
Lesson complete
Great work! You now know how to:
- compute the probability of a degenerate result with a projector
- find the state it leaves, superposition kept
- resolve a degeneracy with a second compatible observable