LM/LK ascent question

David Lesher wrote:
Pat Flannery writes:

The astronauts take star sightings and feed the info into the Inertial
Measuring Unit to get everything aligned as to the LM's position and
orientation on the Moon's surface: http://history.nasa.gov/alsj/aot.htm
So they can indeed update their guidance system to compensate for gyro
drift while on the lunar surface.

I have read Digital Apollo, & the small book on the AGC, and I am
still astonished as to how such a wimpy box could do such complex
calculations; the map data alone must be most of the ROM space.

Not as much as you might think. There were only 37 stars in the star
map, each of which was defined by a three-dimensional unit vector in
inertial coordinates. Each component of the unit vector was a
double-precision AGC fixed-point number. So the star map was only
(37*3*2) or 222 AGC words.

There was also no full map of the lunar surface. There were stored
landmark coordinates but each one of those was just a unit vector in
moon-fixed coordinates, requiring six words per landmark. For the early
flights, the lunar surface itself was modeled as a sphere with a radius
equal to the landing site vector magnitude. For the later flights, there
was a simple lunar terrain model that consisted of just a few altitude
points uprange of the landing site along the approach path, with linear
interpolation between the points. This model therefore only required a
few words of AGC memory.