No preamble with state files and allow_partial = True might not work as expected.

[yoshinori-0778]

There is two problems:

1. When I try to make a schedule manually, one of the yaml file output a schedule without preamble. Yaml file is below. If I set t0_state_file: None, there is preamble in the output schedule.
2. I set allow_partial: False for jupiter observation, but somehow this is not working as expected because an output schedule include a jupiter observation although the highest elevation of jupiter being 48deg is lower than the top wafer for ws5. (See attached file. ws5 is the top right wafer) I thought there is no jupiter observation because of that reason, but there actually is.

The yaml file I use: /so/home/ysueno/workspace/script/scheduler-scripts/sat/yamls/202510/1003_jupiter.yaml

platform: satp3

# yaml loads iso format automatically into datetimes
t0: 2025-10-03T16:00:00+00:00
t1: 2025-10-04T16:00:00+00:00
#t0_state_file: None
t0_state_file: satp3/state_files/state_2025-10-03T16:00:00+00:00.npy

elevation: 60
boresight: 0
az_speed: 0.5
az_accel: 0.25
min_hwp_el: 40
#max_hwp_el: 60
#force_max_hwp_el: True

output_table: True
make_plot: False

# optional, only needed if running non-defaults
apply_boresight_rotation: False
disable_hwp: False
brake_hwp: False
no_cmb: False
#hwp_override: True # True: positive frequency, False: negative frequency
az_motion_override: True # True: overwrite azspeed as set above, False: use azspeed written in the master schedule.
relock_cadence: 86400 # Relock every 24 hours
home_at_end: False # If true spin down HWP and go to (180,40), otherwise scheduler will find a sun-safe parking spot and go there until the end of the specified time-frame.
bias_step_cadence: 1800
use_cal_file: False # If true scheduler will find plant observations from planet cal files
use_wiregrid_file: True
wiregrid_el: 60

# can comment out completely if only wanting CMB
# BE CAREFUL - scheduler does not spin down HWP for boresight rotations 
#              in cal targets. Probably best not to use them
cal_targets:
  - source: jupiter
    # boresight: # if not specified, use scan boresight
    elevation: 40
    focus: 'ws5'
    allow_partial: False
    az_branch: 270 ## maybe need to scan one side of jupiter
  - source: moon
    # boresight: # if not specified, use scan boresight
    elevation: 55
    focus: 'ws3'
    allow_partial: False

Initial part of output schedule. There is no preamble.

run.acu.move_to(az=111.5, el=60.0)
run.wait_until('2025-10-03T17:06:20+00:00', tolerance=1200.0)

################### Detector Setup######################
with disable_trace():
    run.initialize()
run.smurf.iv_curve(concurrent=True)
run.smurf.bias_dets(rfrac=0.5, concurrent=True)
time.sleep(300)
run.smurf.bias_step(concurrent=True)
#################### Detector Setup Over ####################

# hwp already spinning ccw (positive frequency)
run.wait_until('2025-10-03T17:26:20+00:00')
# scan duration = 0:40:40
run.seq.scan(
    description='Low Priority Southern Scan',
    stop_time='2025-10-03T18:07:00+00:00',
    width=40.0, az_drift=0,
    subtype='cmb', tag='uid-0c8b6b00-ee13-4cfd-94e7-17db33bd8bf6,112-152',
    min_duration=600,
)

[yoshinori-0778]

It seems allow_partial = False works as expected now somehow with the yaml file below.
This does not generate jupiter scan but it generate jupiter scan when I change it to one with allow_partial = True.

I used all up-to-date scripts as of Oct 15 13:30 (ET).

platform: satp3

# yaml loads iso format automatically into datetimes
t0: 2025-10-15T16:00:00+00:00
t1: 2025-10-16T16:00:00+00:00
#t0_state_file: None
t0_state_file: satp3/state_files/state_2025-10-15T16:00:00+00:00.npy

elevation: 60
boresight: 0
az_speed: 0.5
az_accel: 0.25
min_hwp_el: 40
#max_hwp_el: 60
#force_max_hwp_el: True

output_table: True
make_plot: False

# optional, only needed if running non-defaults
apply_boresight_rotation: False
disable_hwp: False
brake_hwp: False
no_cmb: False
#hwp_override: True # True: positive frequency, False: negative frequency
az_motion_override: True # True: overwrite azspeed as set above, False: use azspeed written in the master schedule.
relock_cadence: 86400 # Relock every 24 hours
home_at_end: False # If true spin down HWP and go to (180,40), otherwise scheduler will find a sun-safe parking spot and go there until the end of the specified time-frame.
bias_step_cadence: 1800
use_cal_file: False # If true scheduler will find plant observations from planet cal files
use_wiregrid_file: True
#wiregrid_el: 60

# can comment out completely if only wanting CMB
# BE CAREFUL - scheduler does not spin down HWP for boresight rotations 
#              in cal targets. Probably best not to use them
cal_targets:
  - source: jupiter
    # boresight: # if not specified, use scan boresight
    elevation: 40
    focus: 'ws0'
    allow_partial: False
    az_branch: 270 ## maybe need to scan one side of jupiter
  #- source: moon
    # boresight: # if not specified, use scan boresight
    #elevation: 55
    #focus: 'ws5'
    #allow_partial: False

[mmccrackan]

Yeah we have changed things in #305 so that it will always start with allow_partial=False and then only try allow_partial=True when the override True.

[yoshinori-0778]

Thanks for changing that, and good to know! Remaining problem is preable one in this issue.

[mmccrackan]

So the no preamble thing happens because the telescope was stopped at (180, 60) at the time of the state (2025-10-03 15:04:00.040000+00:00) but this is not sun-safe at the start of the schedule (2025-10-03 16:00:00+00:00). The scheduler finds that (180, 56) is the position with the highest sun-safe elevation at that time. Basically, we couldn't have left the telescope at (180, 60) between the times when the state was generated and when the schedule starts. The avoidance calculator (https://site.simonsobs.org/scheduler-web/Sun_Avoidance) shows that it becomes unsafe at 15:35 UTC.

[yoshinori-0778]

I see, good to know the reason.
This is the end of schedule one day before that. Do you mean this is not the sun-safe because telescope have been at elevation 60 until 16UTC but it becomes unsafe at 15:35UTC? I am confused why I can generate this schedule.
Are we able to handle this with scheduler side? or we should make a schedule whose end is not sun unsafe time.

run.wait_until('2025-10-03T14:49:00+00:00')
run.wiregrid.calibrate(continuous=False, elevation_check=True, boresight_check=False, temperature_check=False)
run.wait_until('2025-10-03T15:04:00+00:00')
run.acu.move_to(az=180.0, el=60.0)
run.wait_until('2025-10-03T16:00:00+00:00')
acu.stop_and_clear()

[mmccrackan]

Hmm, when I generate a schedule for the day before with the above configuration (changing the date and removing the state) it goes to (180, 56) at the end. The current version of the scheduler finds a parking spot that is safe for the entire schedule period so I'm not sure how it decided to go there.

It looks like we do need to extend the state until the end of the schedule, as it is storing the time of the wiregrid measurement instead of the end of the entire schedule.

[yoshinori-0778]

Could you give me a yaml file so I can try?
I tried with a yaml file that is likely to be the same as yours, but I found the final movement is to (180, 60).

[mmccrackan]

I tried with the one below on the current main branch.

platform: satp3

# yaml loads iso format automatically into datetimes
t0: 2025-10-02T16:00:00+00:00
t1: 2025-10-03T16:00:00+00:00
#t0_state_file: None
#t0_state_file: satp3/state_files/state_2025-10-03T16:00:00+00:00.npy

elevation: 60
boresight: 0
az_speed: 0.5
az_accel: 0.25
min_hwp_el: 40
#max_hwp_el: 60
#force_max_hwp_el: True

output_table: True
make_plot: False

# optional, only needed if running non-defaults
apply_boresight_rotation: False
disable_hwp: False
brake_hwp: False
no_cmb: False
#hwp_override: True # True: positive frequency, False: negative frequency
az_motion_override: True # True: overwrite azspeed as set above, False: use azspeed written in the master schedule.
relock_cadence: 86400 # Relock every 24 hours
home_at_end: False # If true spin down HWP and go to (180,40), otherwise scheduler will find a sun-safe parking spot and go there until the end of the specified time-frame.
bias_step_cadence: 1800
use_cal_file: False # If true scheduler will find plant observations from planet cal files
use_wiregrid_file: True
wiregrid_el: 60

# can comment out completely if only wanting CMB
# BE CAREFUL - scheduler does not spin down HWP for boresight rotations
#              in cal targets. Probably best not to use them
cal_targets:
  - source: jupiter
    # boresight: # if not specified, use scan boresight
    elevation: 40
    focus: 'ws5'
    allow_partial: False
    az_branch: 270 ## maybe need to scan one side of jupiter
  - source: moon
    # boresight: # if not specified, use scan boresight
    elevation: 55
    focus: 'ws3'
    allow_partial: False

[yoshinori-0778]

Thanks for sharing this, but I could not reproduce your result.
End of my schedule looks like this. I used scripts up-to-date main branch. Did you do in a site computer?

run.wait_until('2025-10-03T14:48:00+00:00')
run.acu.move_to(az=180.0, el=60.0)
run.wait_until('2025-10-03T14:48:59+00:00', tolerance=1200.0)
# hwp already spinning ccw (positive frequency)
run.wait_until('2025-10-03T14:49:00+00:00')
run.wiregrid.calibrate(continuous=False, elevation_check=True, boresight_check=False, temperature_check=False)
run.wait_until('2025-10-03T15:04:00+00:00')
run.acu.move_to(az=180.0, el=60.0)
run.wait_until('2025-10-03T16:00:00+00:00')
acu.stop_and_clear()