Recipe
Find the partition of a nilpotent operator
Use when: you have a nilpotent operator or matrix N and need the exact sizes of its Jordan chains, the data that fixes its similarity class completely
Goal: the full partition (k_1,k_2,\dots) of chain lengths, with the lengths summing to \dim V
Taught in Nilpotent operators and their structure
Quick recipe
-
1
Compute N^2, N^3, \dots in turn and stop at the first power that comes out to the zero matrix; call that exponent k.
Example
For N=\begin{pmatrix}0&1&0&0\\0&0&1&0\\0&0&0&0\\0&0&0&0\end{pmatrix}, N^2 still sends e_3\mapsto e_1 (not zero), but N^3 is the zero matrix, so k=3.
- N^j is still nonzero — raise the power by one and test again.
- N^j first comes out to the zero matrix — stop; that exponent is k.
⚡ Shortcut
If N is already written as block-diagonal Jordan blocks (1's on the superdiagonal, 0 everywhere else), read k directly as the size of its largest block, skipping the power test entirely.
Why: every later step compares kernel dimensions only up through this exponent, so the recipe cannot proceed without knowing where they stop growing.
-
2
Solve N^j\mathbf v=\mathbf 0 for each power from j=0 up through j=k+1, and record the dimension of each solution space as d_j.
Example
For the same N: d_0=0 (the identity's kernel is always trivial), d_1=2, d_2=3, d_3=4, and d_4=4.
Why: the chain-counting formula in the next step needs the dimension immediately below and above each j, not just the value at j itself.
-
3
For every length j from 1 to k, compute c_j=2d_j-d_{j-1}-d_{j+1}. This number is exactly how many chains have length j, not length at least j.
Example
c_1=2(2)-0-3=1, c_2=2(3)-2-4=0, and c_3=2(4)-3-4=1.
Why: this double difference isolates the chains that stop at exactly j, since a longer chain would still be contributing to d_{j+1} too.
-
4
List each length j with c_j>0, repeated c_j times, largest first, then add up length times count over every j.
Example
Here c_3=1 and c_1=1 (with c_2=0 contributing nothing), so the partition is (3,1), and 3+1=4=\dim V.
Why: a partition whose lengths fail to add up to \dim V signals an arithmetic error in an earlier step, not a valid answer.
Common mistakes
- Treating d_0 as \dim V instead of 0; N^0 is the identity, which has a trivial kernel.
- Stopping the power test one exponent too early, before N^j actually reaches the zero matrix.
- Reading c_j as the number of chains of length at least j rather than exactly j.
- Dropping a length whose count c_j came out to 0, instead of simply leaving it out of the partition.
- Listing the partition out of order or skipping the final check that the lengths add up to \dim V.