With some help from Shuai, I got a map of the MTD detetor physical locations to the 4 QT boards, which I assume are the 4 MTD signals you can use in the trigger. It's not strictly mapped to quadrants in phi (there is overlap), but it is roughly:

1. MT001 spans 11 o'clock to 3 o'clock
2. MT002 spans 9:30 o'clock to 1:30 o'clock
3. MT003 spans 5:30 o'clock to 9:30 o'clock
4. MT004 spans 3:30 o'clock to 7:30 o'clock

For more precision, one can look at these documents I got from Shuai for the MTD trigger flow, and the backleg numbering scheme, for which one should pay attention to the fact that this is the east face, so the global "3 o'clock" is located at "9 o'clock" on the left side of this image (click on either of the two images for a full size version):


Since the GMT modules are at 2 o'clock and 5 o'clock, the optimal triggers (given the rather broad MTD trigger patches available) would be:
(Note: the images [click on each for a larger version] show the topologies in azimuth [ignoring z], with red arcs for the trigger elements and blue boxes for the viable GMT module intersections,)

Trigger 1:

MT001 && MT003 && (TOF centered at 4 o'clock || TOF centered at 6 o'clock || TOF centered at 8 o'clock || TOF centered at 10 o'clock)

Trigger 2:

MT001 && MT002 && TOF centered at 10 o'clock

Trigger 3:

MT002 && MT004 && (TOF centered at 10 o'clock || TOF centered at 12 o'clock)

Trigger 4:

MT001 && MT004 && (TOF centered at 12 o'clock || TOF centered at 2 o'clock)

Trigger 5:

MT003 && MT004 && TOF centered at 8 o'clock

The only MTD trigger pairing not used is MT002 && MT003, as that has no acceptance intersecting a GMT module.

______________

On 2019-11-26, Eleanor wrote:

The GMT triggers that we talked about last Run have finally been coded
and are available for testing. You gave me 5 separate trigger
definitions. I did not have 5 unused bits available in the TCU so what
I have done is combine all of these bits into 1 (OR), with a register
that lets you turn each of the 5 components on or off. Also as you
know the 5 triggers involve combining a lot of different bits from MTD
and TOF, and things will only work right if I got all the mapping
correct. Sometime after Thanksgiving Jack will put these algorithms
into the Trigger Tier1 file and then they can actually be tested. When
you are doing those tests I would STRONGLY recommend that you start
out by taking 5 separate runs and only turn on 1 of the 5 components
each time. It should then be possible to work out a) if I coded it
correctly as well as b) if it actually works in real life.

Then, on 2019-12-03, she wrote:

Many thanks to Geary and Hank for taking the test runs yesterday.
There are a total of 8 runs listed in the RunLog browser that are all
successful, use the new GMT logic, include a GMT trigger, readout all
of MTD, TOF, TPC and GMT as well as all the trigger electronics and
have the magnet turned off. All of these runs were taken with the
CosmicLocalClock_GMT configuration.

The runs each have a different value for the GMT_onoff register, so
each run has a different GMT trigger component turned on:
Run 20336046 has GMT_onoff = 1
Runs 49 and 50 have GMT_onoff = 2
Runs 51 and 52 have GMT_onoff = 4
Run 53 has GMT_onoff = 8
Runs 55 and 57 have GMT_onoff = 16

Run 58 has GMT_onoff = 31. The detectors listed on the RunLog browser
page only show daq and trg (i.e. the TPC, and GMT were not readout).
However when I dig into the database the component tree shows that the
CPU nodes for GMT, TPC, etc... were in the Run. So I am not sure what
happened there and whether or not that run will be any use to you.

I will try to look at these runs.

-Gene