Intuition The one core idea
A physical arrow in space never changes — but the three numbers you use to describe it change depending on who is looking (a spinning spacecraft vs. the fixed stars). Everything in this topic is about building the one machine — a grid of nine cosines — that converts that arrow's numbers from one observer's grid to another's.
This page assumes you have seen nothing . We build every symbol the parent note throws at you, in an order where each one leans only on the ones before it. Read top to bottom.
v
A vector is a quantity with both a size and a direction — picture an arrow floating in space. The little hat-arrow on top, v , just means "this letter is an arrow, not a plain number".
Intuition The arrow itself is stubborn
Look at the red arrow in the figure. It points from the tail to the tip. It does not care how you set up your grid lines around it. This stubbornness is the whole point of the topic: the arrow is a physical fact (which way the thrust points, which way the star is), and only its description in numbers will change later.
Why the topic needs it: a spacecraft's thrust, a star's direction, an angular velocity — all of these are arrows. If we could not talk about arrows, we could not talk about pointing at all.
To turn an arrow into numbers , you first lay down three reference directions to measure against.
Definition Axis and unit vector (
x ^ )
An axis is a straight reference line, like a ruler laid down in space. A unit vector is an arrow exactly 1 unit long pointing along that axis. We write it with a hat: x ^ (read "x-hat"). "Unit" = length one; "hat" = "this is a direction-only arrow".
Definition Frame (coordinate frame)
A frame is a full set of three unit vectors x ^ , y ^ , z ^ that are mutually perpendicular (each at 9 0 ∘ to the others) — a rigid 3-way corner, like the corner of a room where two walls meet the floor. This is your measuring apparatus.
v x , v y , v z )
The components of v are the three numbers telling you how far along each axis the arrow reaches. We collect them in a stacked column:
v = v x v y v z
Intuition Same arrow, two grids, two number-triples
In the figure the one red arrow is measured by two different frames (two corners rotated relative to each other). The straight-frame reads it as one triple of numbers; the tilted frame reads it as a different triple. Nothing about the arrow moved — only the ruler turned. This picture is the entire topic in one image.
Why the topic needs it: a gyro bolted to a tumbling spacecraft measures with a tilted corner; the star map is written in a straight corner. Same star, two triples. We must convert between them.
A frame is orthonormal when it is both:
Ortho gonal — the axes are mutually perpendicular (9 0 ∘ apart), and
Normal ized — each axis vector has length exactly 1.
Picture the tidy corner of a cube: three edges, all the same length, all square to each other.
Why the topic needs it: only for orthonormal frames does the cheap "just transpose to invert" trick (Section 8) work. Real spacecraft frames are always built this way on purpose.
This is the single most important tool on the page, so we build it slowly and say why this tool and not another .
a ⋅ b
The dot product takes two arrows and returns one plain number (no arrow):
a ⋅ b = ∥ a ∥ ∥ b ∥ cos θ
where ∥ a ∥ means "the length of a " and θ is the angle between the two arrows.
Let us unpack every piece:
∥ a ∥ — the length (magnitude) of arrow a , a plain positive number, like the length of a stick.
θ — the angle between the two arrows, measured where their tails meet.
cos θ — the cosine : a number that answers "how aligned are these two directions?" It runs from + 1 (perfectly parallel, θ = 0 ) through 0 (perpendicular, θ = 9 0 ∘ ) to − 1 (opposite, θ = 18 0 ∘ ).
Intuition Why cosine, and why this tool?
We need a number that says "how much of arrow a lies along the direction of b ." Drop a shadow (a perpendicular projection) of a onto the line of b — the length of that shadow is exactly ∥ a ∥ cos θ . Cosine is the only function that behaves this way: full credit when aligned, zero credit when perpendicular, negative when pointing backward. No other simple tool encodes "alignment" so cleanly — that is why the topic reaches for the dot product and nothing else.
Worked example Projecting onto a unit axis — the key special case
If x ^ is a unit vector (∥ x ^ ∥ = 1 ), then
v ⋅ x ^ = ∥ v ∥ ( 1 ) cos θ = ∥ v ∥ cos θ = v x
So dotting an arrow with a unit axis hands you back that arrow's component along the axis . This one line is the engine the parent note uses to build every entry of the rotation matrix.
Why the topic needs it: the parent derives each component as v x B = v ⋅ x ^ B . That is this special case, nothing more.
When the parent writes v = v x x ^ + v y y ^ + v z z ^ and then dots the whole thing with x ^ , every term collapses except one: the y ^ and z ^ pieces vanish (they are perpendicular to x ^ ), leaving just v x . These zeros are the janitors that clean the derivation down to a single surviving term.
Definition Kronecker delta
δ ij
δ ij is a tiny bookkeeping symbol:
δ ij = { 1 0 if i = j (same index) if i = j (different index)
It is literally the previous section's rule written compactly: e ^ i ⋅ e ^ j = δ ij , where e ^ 1 , e ^ 2 , e ^ 3 just rename x ^ , y ^ , z ^ .
Why the topic needs it: the parent's proof that R R T = I says "entry ( i , j ) is e ^ i ⋅ e ^ j = δ ij " — that's exactly this symbol packaging the 1s-and-0s pattern.
R and matrix–vector multiply
A matrix is a rectangular grid of numbers. A 3 × 3 matrix has 3 rows and 3 columns. To multiply a matrix by a component-column, you take the dot product of each row with the column :
r 11 r 21 r 31 r 12 r 22 r 32 r 13 r 23 r 33 v x v y v z = r 11 v x + r 12 v y + r 13 v z r 21 v x + r 22 v y + r 23 v z r 31 v x + r 32 v y + r 33 v z
Intuition It is just three dot products stacked
You already learned the dot product in Section 4. A matrix multiply is nothing new — it is three dot products, one per row , producing three new numbers. Each output number is "how much of the input arrow lies along one body axis". That is why the rotation matrix's rows turn out to be the body axes.
R T and identity I
The transpose R T ("R-transpose") is R flipped across its diagonal: rows become columns. The identity matrix I has 1s down the diagonal and 0s elsewhere; multiplying by it changes nothing (it is the number 1 of matrices):
I = 1 0 0 0 1 0 0 0 1
Why the topic needs it: the DCM R B I is such a matrix; converting an arrow's numbers between frames is exactly one matrix–vector multiply. And "R T R = I " is the property that lets you undo a rotation for free.
R B I
The two subscripts name where you land and where you start : R B I means "B ody from I nertial". Feed it inertial numbers, it returns body numbers:
v B = R B I v I
The superscript I or B on v tags which frame's numbers these are — same arrow, different observer.
Mnemonic Subscripts cancel like units
In a chain R CB R B I the inner B 's touch and cancel , leaving R C I — just like hr km ⋅ hr leaves km. If the inner labels don't match, you wrote it wrong.
Frame = 3 perpendicular unit axes
Axis and unit vector x-hat
Projection gives a component
Perpendicular axes dot to zero
Matrix = three dot products stacked
Rotation matrix R body from inertial
Convert arrow numbers between frames
Everything funnels into the same destination: converting an arrow's numbers between the attitude of the body and the fixed stars. From here you are ready for quaternions , gyro sensing , and orbit frames .
Cover the right side and answer out loud. If any one stalls you, reread its section before the parent note.
What does the hat in x ^ tell you? That this arrow has length exactly 1 — it is a pure direction (a unit vector).
Why don't a vector's numbers stay the same in every frame? The arrow is fixed, but each frame measures it against different axes, so the three component-numbers differ.
In words, what does a ⋅ b = ∥ a ∥∥ b ∥ cos θ measure? How much of one arrow lies along the direction of the other — their alignment.
Why is cosine the right function for a projection? It gives full value when arrows align (cos 0 = 1 ), zero when perpendicular (cos 9 0 ∘ = 0 ), negative when opposed — exactly the shadow-length behaviour.
What is v ⋅ x ^ when x ^ is a unit axis? It is
v x , the component of
v along
x ^ .
What is x ^ ⋅ y ^ for perpendicular unit axes, and why? Zero, because cos 9 0 ∘ = 0 .
What does δ ij equal? 1 when i = j , 0 when i = j .
A matrix–vector multiply is really just what, repeated? A dot product of each matrix row with the column, three times.
What does R T do to a matrix? Flips it across the diagonal — rows become columns.
Read R B I in plain words and say what it does. "Body from inertial" — it takes inertial component-numbers and returns body component-numbers.
In R CB R B I , why is the product valid? The inner B labels touch and cancel, leaving R C I .