LSST Active Optics System Software Architecture

Sandrine J. Thomas

, Paul Lotz

, Srinivasan Chandrasekharan

, Bo Xin

, Charles Claver

George Angeli

, Jacques Sebag

, and Gregory P. Dubois-Felsmann

LSST, 950 N. Cherry Av, Tucson, AZ, USA

Caltech, 1200 E California Blvd, Pasadena, CA 91125, USA

ABSTRACT

The Large Synoptic Survey Telescope (LSST) is an 8-meter class wide-?eld telescope now under construction on

Cerro Pach?on, near La Serena, Chile. This ground-based telescope is designed to conduct a decade-long time

domain survey of the optical sky. In order to achieve the LSST scienti?c goals, the telescope requires delivering

seeing limited image quality over the 3.5 degree ?eld of view. Like many telescopes, LSST will use an Active

Optics System (AOS) to correct in near real-time the system aberrations primarily introduced by gravity and

temperature gradients. The LSST AOS uses a combination of 4 curvature wavefront sensors (CWS) located on

the outside of the LSST ?eld-of-view. The information coming from the 4 CWS is combined to calculate the

appropriate corrections to be sent to the 3 di?erent mirrors composing LSST. The AOS software incorporates a

wavefront sensor estimation pipeline (WEP) and an active optics control system (AOCS). The WEP estimates

the wavefront residual error from the CWS images. The AOCS determines the correction to be sent to the

di?erent degrees of freedom every 30 seconds. In this paper, we describe the design and implementation of the

AOS. More particularly, we will focus on the software architecture as well as the AOS interactions with the

various subsystems within LSST.

Keywords: Large Synoptic Survey Telescope, Active Optics, Wide Field of View, Curvature Sensing

1. INTRODUCTION

The Large Synoptic Survey Telescope (LSST) is an 8.4 meter diameter telescope now under construction on

Cerro Pachon in Chile.

The telescope is composed of three aspheric mirrors: the 8.4m primary mirror (M1),

a 3.4m secondary mirror (M2) and a 5m tertiary mirror (M3). The primary and tertiary mirrors form a single

monolithic mirror called the M1M3

. LSST has one instrument, a three-lens camera that directs the light

path onto a 3.2 gigapixel camera, with a ?eld-of-view (FOV) of 3.5 degrees.

The camera has 6 di?erent ?lters,

u,g,r,i,z,y. The overall system image quality budget for the LSST is 0.4 arcsec FWHM with 0.25 arcsec allocated

to the telescope and 0.3 arcsec associated to the camera. LSST is a seeing-limited telescope and thus these

performance do not include the errors coming from the atmosphere.

In order to optimize the image quality across the 3.5 degree FOV of the camera, LSST relies on an Active

Optics System (AOS). Both M1M3 and M2 are equipped with actuators that allow for the AOS to control the

shape of the mirror surface and their position.

In addition, M2 and the camera are on hexapods allowing further

positioning adjustment.

The wavefront control is performed by determining a base set of control parameters based on both ?nite

element analysis (FEA) and on-sky open-loop measurements as well as by adding low temporal frequency closed

loop corrections determined from real-time wavefront measurements. The open-loop model may take the form

of equations or tables. In this paper, we also refer to this as Look-up Table (LUT). The LUT provides near

optimum values for all actuator forces and hexapods positions and depends primarily on elevation angle and

temperature. Although constructed using FEA models, it will be veri?ed extensively during the commissioning

phase.

Further author information: (Send correspondence to S.J.T.)

S.J.T.: E-mail: sthomas@lsst.org, Telephone: 1 520 318 8227

The corrections given by the LUT, however, do not allow the system to meet the LSST image quality

requirements due to non-repeatable and/or unpredictable e?ects. These e?ects include temperature errors, wind

loads and hysteresis. Accordingly, LSST plans to use real-time wavefront sensor measurements to compensate

for these errors to achieve the LSST required image quality. The wavefront sensor selected is a curvature sensor,

since the wide-?eld and the fast beam make the use of the more popular Shack-Hartmann wavefront sensor

problematic. Moreover, the curvature sensors can be used in the focal plane, while the Shack-Hartmann is used

in the pupil plane. The curvature sensor option avoids non-common path errors. More details on the origin and

impact of di?erent environmental perturbations on the LSST image quality are described in a previous paper.

In addition to on-sky wavefront correction, the active optics will be used to conduct alignment veri?cation in

coordination with a laser tracker at the beginning of every night or during the night if needed.

This paper describes the software architecture of the active optics system. Section

presents the di?erent

use-cases of the AOS. Section

describes the technical aspect of the wavefront sensor both mechanically and

relative to software. Section

and section

detail the steps and functions of the components of the AOS.

2. OPERATING MODES

2.1 Normal Operation and Engineering Operation

Normal operation is de?ned by routine nighttime observations for which there is no human intervention. The

AOS input is 4 individual pairs of intra- and extra-focal images corresponding to the 4 di?erent wavefront

sensors. The images come directly from the Data Acquisition system (DAQ) in the Camera Subsystem. In order

to maintain the required image quality, the AOS output consists of the bending modes sent to the mirror support

system of M1M3 and M2 to control the mirror shapes, and positions sent to the M2 and camera hexapods. The

observatory cadence of the LSST telescope is relatively fast since the telescope pointing changes every 39 seconds.

These 39 seconds are decomposed into two 16s visits (1s of which is to allow the shutter to open and close), a 2s

readout and a 5s slew time. The current baseline plan uses the data from the ?rst visit only and performs the

calculation during the second visit.

During normal operation, we will also have an engineering or manual mode. This mode will be used in open

loop, meaning that after an image is acquired, the AOS pipeline does not automatically run through and send

the corrections to the mirrors, but allows interactions between the AOS and the user. This submode will be

used during commissioning or for corrective maintenance during operation. For instance, when the AOS will

be operated manually, it will enable local hardware control with a personal computer via the facility control

network. Visualization of images and parameters will also be available.

2.2 Full Focal Plane Array Wavefront Sensing

The Full Focal Plane Array (FFPA) mode will be used to build and optimize the LUT during commissioning and

on occasion during the day as part of calibration. The science focal plane array (FPA) consists of 189 science

detectors.

In this FFPA mode, the AOS obtains the data from the Data Management subsystem (DM) due to

the required temporal frequency of correction being dramatically reduced. The images originate from the 189

science detectors for each of the di?erent sets of defocus, mostly ? 2 mm, as well as WFS images with the camera

in focus for reference. At this reference position, the AOS will collect images from both WFS detectors and

science detectors. Using the full focal plane for wavefront sensing is useful to relate the wavefront sensor data to

image quality in the ?eld.

This mode will also be used during early commissioning using the commissioning camera (ComCam). Com-

Cam has 1 raft with 9 science detectors, therefore leading to a FOV of 0.6 degrees. This mode will help the

commissioning team exercise the AOS algorithm and accomplish the ?rst alignment procedure checks. With

regards to software and algorithms, this operating mode is a slight divergence from the nominal automatic

operation mode, mostly consisting of an increase of number of sensors read and order of operations.

2.3 Alignment procedure

As mentioned earlier, the AOS will also be used at the beginning of each night as an alignment tool to verify

and adjust the reference position relative to the LUT, if needed. In other words, at the beginning of each night,

the observatory operator aligns the main optical elements (mirrors and camera) using the following tools:

- A look-up-table incorporates the state of the system measured during the previous night. The look-up-table

is the set of absolute hexapod positions and mirror shapes in order to achieve the required image quality. Its

requirement is de?ned below. The reference positions are ?xed for at least one night.

- A laser tracker works with the beam re?ectors to adjust tip/tilt/rotation and translation. More information

on the laser tracker is in Araujo et al.

- The curvature wavefront sensor further minimizes the wavefront error. Before the night's observing, the

appropriate active optics corrections are applied to the mirror shapes, and the camera and M2 mirror are

positioned relative to the M1M3 mirror by their hexapods. The position of the camera and M2 will likely be

veri?ed by the laser tracker system. This ensures that the wavefront errors are within the dynamic range of the

active optics system.

3. LSST WAVEFRONT SENSING TECHNICAL DESCRIPTION

This section gives a more detailed description of the wavefront sensing for LSST. The ?rst subsection presents

the optical layout of the camera dedicated to wavefront sensing and the second section focuses on the software

general architecture.

As mentioned in the introduction, the optical wavefront sensor chosen for LSST is a curvature wavefront

sensor, leading to the need to have sets of two images with a de?ned focus o?set between them. To satisfy an

additional constraint to reduce or eliminate non-common path errors, the wavefront sensor is located in the focal

plane of the camera along with the other science sensors.

The wavefront sensor is composed of four sensors manufactured by the Imaging Technology Laboratory (ITL)

at the University of Arizona. The sensors are located at the corner of the focal plane as shown on Figure

-left.

Each wavefront sensor is composed of two defocused detectors of 2k ? 4k pixels as shown on Figure

-right. The

optimal defocus was the result of a trade-o? study, including impact of signal-to-noise, atmospheric turbulence,

non-linearity of the wavefront sensor and sky coverage.

The FOV area of each wavefront sensing detector is 6.8 arcminutes ? 13.7 arcminutes. Figure

-right shows

a simulation of the defocused images on the full FOV of one of the four wavefront sensors, using a photon

simulation tool called PhoSim.

These simulations are executed using a bright star catalog such as the catalog

from the United State Naval Observatory (USNO) led by Monet et al.

or internal catalogs.

The images in Figure

-right demonstrate some of the di?culties in using the LSST wavefront sensor images

to reconstruct the wavefront and calculate the best set of corrections to apply to the di?erent degrees of freedom

(presented in section

). The images show blending issues where two stars are so close to another that the two

donuts overlap, resulting in di?culties in the estimation of the wavefront. Some of the donuts also have a low

signal-to-noise ratio. The two images show the di?erences in star counts that can occur even for very close ?elds.

The Active Optics System (AOS) is responsible for measuring the residual wavefront error from the defocused

images and estimating the correct bending modes to send to the mirror control systems (M1M3 and M2) and

the positions to send to the hexapods (M2 and camera). The AOS is part of the Telescope Control System.

Figure

describes the inputs and the output to the AOS or interfaces. The interfaces to the AOS are

of several types: interfaces to other subsystems (DM and Camera), and interfaces to the di?erent controllers

(M1M3, M2 and the hexapods). The AOS exchanges messages with other systems using a publish-subscribe

middleware solution based on an implementation of the Data Distribution Service speci?cation.

Input interfaces.

In normal operation the AOS requires crosstalk corrected images from the Camera

Data Acquisition System; these images arrive via a custom push data interface. Still in normal operation, the

AOS will need to know the state of the camera (selected ?lter, for example); this information comes from the

Observatory Control System (OCS).

Finally, the baseline requires the use of calibration images and data such as

Figure 1. Focal plane layout of the detector. The left drawing shows the full focal plane with the science sensors, the

wavefront sensors and the guide sensors. The wavefront sensors and the guide sensors will have vignetting e?ects as well

as distortion that must be taken into account in the wavefront estimation. The right panel shows simulated intra and

extra images for one of the wavefront sensors. They were obtained using a complex simulation tool called PhoSim.

?at ?eld images or bad pixels maps as described in section

4.1

. This metadata comes from the Data Management

subsystem, once a night.

Output interfaces. The AOS outputs bending modes and position information to the di?erent controllers

listed in section

Figure 2. AOS ?ow diagram showing inputs and outputs between di?erent components. The AOS receives the images

either from the camera or the DM subsystems. It also receives telemetry from the Observatory Control System. The

output to the AOS is sent to the di?erent mirrors and hexapods controllers. The information sent is the bending modes

o?sets for the mirrors and the positions for the hexapods/rotator.

The AOS includes the Wavefront Estimation Pipeline (WEP) and the Active Optics Control System (AOCS).

Each entity will be described in the following section. The WEP sends to the AOCS a set of Zernike coe?cients

describing the wavefront error for each sensor { 4 in normal operation mode and 189 in FFPA mode.

4. WAVEFRONT ESTIMATION PIPELINE

As mentioned in the introduction, the Wavefront Estimation Pipeline (WEP) is the algorithm responsible for

estimating the wavefront errors in annular Zernike coe?cients associated with each of the corner rafts. In normal

operation (see section

), the inputs to the WEP are an intra-focal image and an extra-focal image coming from

each of the four corner rafts.

The WFS data are pulled directly from the Wavefront Data Acquisition System (DAQ) after each exposure.

The camera team is responsible for the DAQ and will provide its Application Programming Interface (API)

and compilable client source code on a Linux-based computer, along with connectivity instructions. This di?ers

from the nominal science operation mode, for which the images coming from the camera are directly sent to the

data management team for processing and archiving at the National Center for Supercomputing Applications

(NCSA).

Each detector of a corner raft sees a di?erent ?eld of the sky; therefore the stars on each the 2 detectors in each

corner are di?erent. In addition, compared to normal active or adaptive optics systems, the wavefront sensors

have a large FOV, with numerous stars. In addition, the detectors are located 1.7 degrees o?-axis, introducing

vignetting and image distortion. For these reasons, image processing before using any curvature sensing algorithm

is required to make the images usable. This includes basic image processing such as Instrumentation Signature

Removal (ISR),

source selection using a bright star catalog, source processing (deblending and intensity

scaling) and master creation. Where applicable, the WEP utilizes common methods available in the LSST

software stack.

Figure

shows the overall architecture of the WEP.

Figure 3. This diagram shows the di?erent steps included in the WEP algorithm. These steps are described in the text.

4.1 Preprocessing

Wavefront Data Collector: During normal operation (see section

2.1

) the wavefront sensor data is directly

sent to the AOS from the camera subsystem using the DAQ through a push/pull interface. For the full array

use case, the baseline is for the wavefront sensor data to go directly to the archive before being pulled by the

AOCS. In this mode, the camera control system publishes an event, triggering the active optics system to pull

the wavefront images from the 4 wavefront sensor detectors. For both cases, the images are cross-talk corrected.

Instrument Signature Removal Wrapper: The WEP is compatible with the Instrument Signature Removal

(ISR) developed by the DM team and will use some of its functions. The stack features used by the WEP

are simple dark subtraction, ?at ?eld correction, gain and biases correction. The ISR will also include world

coordinate system (WCS) correction. The other corrections required by the WEP are not part of the ISR and

are described in the following paragraphs.

The Bright Star Catalog and Source Selector: Due to the large FOV of the wavefront sensor detectors,

the source ?nding will rely on a bright star catalog. The impact of galaxies is expected to be minimal. The catalog

contains the RA and DEC of bright stars only. It includes the star's limiting magnitude, and the saturation

as a function of the LSST ?lter, and information about the number of neighbors falling in a prede?ned nearby

surrounding region. The ?nal acceptable number of neighbors within a certain region of interest and the size

of that region of interest are currently being de?ned. The ?rst instance of the catalog is expected to utilize the

catalog developed by USNO;

this catalog will be re?ned using LSST data during the lifetime of the survey.

Figure

shows the pre-requisite and components of the source selector step.

Figure 4. This diagram shows details of the source selection process, including items needed such as a bright star catalog

or components of this step.

Creation of intra/extra focal plan image masters: The result of the source ?nding step is a set of postage

stamps for each of the 8 detectors, including neighbors. The WEP then cleans the postage stamp images creating

sub-images of donuts without any contamination from nearby stars. This process is called deblending. Because

of the di?erent magnitudes on each detector, the intensity from each sub-image is scaled to a same magnitude.

Finally, all the images are summed to create a single pair of intra- and extra-focal images per sensor reference

to as the master. These intra and extra masters can be seen as the input to regular curvature sensing.

The limiting magnitude will vary slightly depending on the condition of the night (seeing, cloud coverage,

etc.). Therefore, the quality control is required to determine if a particular donut has the required signal to noise

ratio ( ˘ 10) to be used in the WEP without introducing extra residual errors.

4.2 Wavefront Estimator

Each set of intra and extra masters are then sent to the wavefront estimator. The wavefront sensor algorithm

used for LSST is curvature sensing and has been described in detail in Xin et al.

Due to the location of

the wavefront sensors in the focal plane ( ˘ 1.7 degrees o? axis) there is a need for distortion and vignetting

corrections, preventing analytical mapping from the telescope aperture to the defocused image. In addition,

since LSST has a large central obstruction (60%) and fast beam (f-number of 1.23), the WEP uses a numerical

solution, representing the mapping between the two sets of coordinates with 2D 10th-order polynomials. The

o?-axis distortion and the vignetting information comes from the optical design (Zemax and pupil information)

and depend on the star location on the detector.

To determine a ?nal algorithm, our strategy is to choose two well-established algorithms that are known to

work for large f-number, on-axis systems, implement them as our baseline, then extend them to work with small

f-number and o?-axis sensors. As our two baseline algorithms we have chosen the iterative fast Fourier transform

(FFT) method by Roddier and Roddier,

and the series expansion technique by Gureyev and Nugent.

details are given in Xin et al.

4.3 Required Inputs and output

Figure

provides further details on the inputs and sub steps that happen in the wavefront estimator box shown

in Figure

Figure 5. This diagram shows the details of the inputs of the di?erent steps included in the wavefront estimator algorithm.

The output of the WEP is an annular Zernike decomposition of the wavefront. The output is communicated

via Data Distribution System (DDS)

or middleware to the active optics control system as a structure with

three main elements for each sensor: the basis-set name, the number of terms, and the di?erent coe?cients. In

addition, there is an image quality metric associated with each result.

4.4 Pipeline Implementation

WEP implementation will utilize Luigi, a platform independent Python package designed to simplify processing

pipelines. It contains boilerplate code and the dependencies are decentralized. This removes the need for large

con?guration ?les since each task speci?es its own dependencies. The use of Luigi will be limited to de?ning

the pipeline task hierarchy. The processing of each task will be de?ned within its own package. This separates

the responsibility for controlling execution order from the actual processing. The result is a set of Python

classes with very simple implementations that clearly describe each task's dependencies, parameters, outputs,

and responsibilities.

5. ACTIVE OPTICS CONTROL SYSTEM

This subsection describes the Active Optics Control System (AOCS). As mentioned above, the AOCS uses the

wavefront measurements pushed by the WEP through the middleware to calculate the correction to send to the

di?erent control elements. The overall feedback algorithm can be separated into two steps: the optical state

reconstructor and the control algorithm (called setpoint mapper in the block diagram presented in Figure

The degrees of freedom that the telescope possesses are the following:

- M1M3 shape, up to 20 bending modes

- M2 shape, up to 20 bending modes

- M2 piston, x/y decenter and x/y tilt using the M2 hexapod

- Camera piston, x/y decenter and x/y tilt using the camera hexapod/rotator

The corrections, de?ned as bending modes and positions, are sent to the di?erent correction elements con-

trollers, namely, the M1M3 controller, M2 controller, M2 hexapod and camera hexapod, as shown on the right

hand of Figure

. The number of bending modes is the optimum number to achieve the required image quality

given by complete simulations.

The output of the AOCS (and thus the AOS) is de?ned in terms of bending

mode o?sets for the M1M3 and M2 shapes as well as the hexapod setpoints. The o?sets are relative to the LUT

(or open model) de?ned in the introduction. The AOS sends setpoints to the component controllers responsible

for moving the relevant axes. These controllers utilize open-loop models (possibly LUTs), in?uence matrices to

convert from bending modes to forces, and custom electronics boards known as Inner Loop Controllers.

Similarly to the wavefront estimation pipeline, the optical feedback control algorithm design is described in

detail in a previous paper.

The focus of this paper is on the architecture.

Figure 6. The block diagram shows the di?erent steps needed to run the AOCS algorithm. This ?ow-down comes after

the WEP ?ow-down shown in Figure

. In this block diagram, the "setpoint mapper" is the control algorithm used to

create the correction to improve the image quality. What is not shown in this graph is the sensitivity matrix needed by

both the optical state estimator and the control algorithm.

Optical State Estimation: The AOCS derives the appropriate state of the optical system from the wavefront

sensors in the presence of algorithmic, WCS and atmospheric noise errors.

The state estimation is either done

from the 4 sets of basis coe?cients for normal operation or 189 sets for full array use case. The most conceptually

simple version of this estimator uses the pseudo-inverse of the sensitivity matrix described at the end of this

section.

Control algorithm (or setpoint mapper): Once the AOCS estimates the optical state of the telescope,

it uses an optimal control determining using a cost function to derive the realistic bending modes to apply to

M1M3 and M2 control systems as well as the M2 and camera hexapods. The result of this cost function provides

the best optical quality with the smallest actuator action. It also sets the control authority between actuator

groups (mirror bending modes and hexapod displacements). The cost function is built to allow the algorithm

to calculate the solution that will optimize the image quality, such as by minimizing the variance of the Full

Width Half Max (FWHM) over the large FOV. Furthermore, the cost function helps limit large actuator swings

to avoid damage to the glass, as well as ensure smooth transitions between iterations. The FOV mapping uses

the Gaussian Quadratic method. The temporal (sampling) behavior of the described controller is equivalent to

an integral controller. Performance can possibly be further optimized by a more complex dynamic controller

utilizing earlier system state estimates.

Sensitivity Matrix: The AOCS uses a sensitivity matrix calculated prior to any AOCS reconstruction. The

sensitivity matrix describes the optical response of the system to the various controlled degrees of freedom. The

columns of the sensitivity matrix correspond to the controlled degrees of freedom (described above). The rows

of the sensitivity matrix correspond to the Zernike coe?cients of the wavefront at the N sensor locations. This

matrix A is used in both the state estimation process and the control algorithm. This sensitivity matrix is

planned to be measured on-sky during commissioning. Note that the bending modes and Zernike coe?cients are

measured relative to a reference state. The sensitivity matrix is calculated o?ine after gathering on-sky data. It

is not expected to change often and will not be modi?ed during a night.

6. CONCLUSIONS

LSST is equipped with an active optics to optimize the image quality of the system. The active optics is a

curvature sensing systemoptimized for the LSST wide ?eld-of-view and optical design. The software architecture

of the AOS is decomposed into two main components: the wavefront estimation pipeline (WEP) which estimates

the residual wavefront from the open-loop model, and the active optics control system (AOCS) which calculates

the M1M3 and M2 shapes as well as the M2 and camera hexapods positions. Both WEP and AOCS are complex

entities that require severals inputs from other LSST subsystems and comprised of several steps. The AOS is

right now in full development with a completion date in accordance with the Telescope and Site Integration Plan,

late 2018.

ACKNOWLEDGMENTS

This material is based upon work supported in part by the National Science Foundation through Cooperative

Agreement 1258333 managed by the Association of Universities for Research in Astronomy (AURA), and the

Department of Energy under Contract No. DE-AC02-76SF00515 with the SLAC National Accelerator Labora-

tory. Additional LSST funding comes from private donations, grants to universities, and in-kind support from

LSSTC Institutional Members.

REFERENCES

[1] Kahn, S., \LSST ?nal design overview," Proc. SPIE 9906 , Paper 17, in press (2016).

[2] Gressler, W. et al., \LSST telescope and site status," Proc. SPIE 9906 , Paper 19, in press (2016).

[3] Araujo, C., Jacques, S., Ming, L., Neill, D., Thomas, Sandrine, J., Vucina, T., and Gressler, W., \Overview

of the LSST mirror system," Proc. SPIE 9906 , Paper 20, in press (2016).

[4] Neill, Douglas, R., Muller, G., Hileman, E., Joe, D., Araujo, C., Gressler, W., Lotz, P., Dave, M., Sebag, J.,

Thomas, S., Warner, M., and Wiecha, O., \Final design of the LSST primary/tertiary mirror cell assembly,"

Proc. SPIE 9906 , Paper 25, in press (2016).

[5] Kurita, N. et al., \Large Synoptic Survey Telescope camera design and construction," Proc. SPIE 9912 ,

Paper 27, in press (2016).

[6] Neill, D., Angeli, G., Claver, C., Hileman, E., DeVries, J., Sebag, J., and Xin, B., \Overview of the LSST

active optics system," Proc. SPIE 9150 , 91500G{91500G{16 (2014).

[7] Sneed, R., Neill, D., Kidney, S., Araujo, C., Gressler, W., Lotz, P., Dave, M., Sebag, J., Sebring, T., A.,

Warner, M., and Wiecha, O., \Final design of the LSST hexapod and rotator," Proc. SPIE 9906 , Paper

18, in press (2016).

[8] Angeli, G. Z., Xin, B., Claver, C., MacMartin, D., Neill, D., Britton, M., Sebag, J., and Chandrasekharan,

S., \Real time wavefront control system for the Large Synoptic Survey Telescope (LSST)," Proc. SPIE

9150 , 91500H{91500H{16 (2014).

[9] Angeli, George, Z., Xin, B., Claver, C., Cho, M., Dribusch, C., Neill, D., Peterson, J., Sebag, J., and

Thomas, S., \An integrated modeling framework for the Large Synoptic Survey Telescope (LSST)," Proc.

SPIE 9911 , Paper 46, in press (2016).

[10] Peterson, J. R., Jernigan, J. G., Kahn, S. M., Rasmussen, A. P., Peng, E., Ahmad, Z., Bankert, J., Chang, C.,

Claver, C., Gilmore, D. K., Grace, E., Hannel, M., Hodge, M., Lorenz, S., Lupu, A., Meert, A., Nagarajan,

S., Todd, N., Winans, A., and Young, M., \Simulation of astronomical images from optical survey telescopes

using a comprehensive photon monte carlo approach," The Astrophysical Journal Supplement Series 218 (1),

14 (2015).

[11] Monet, D. G., Levine, S. E., Canzian, B., Ables, H. D., Bird, A. R., Dahn, C. C., Guetter, H. H., Harris,

H. C., Henden, A. A., Leggett, S. K., Levison, H. F., Luginbuhl, C. B., Martini, J., Monet, A. K. B., Munn,

J. A., Pier, J. R., Rhodes, A. R., Riepe, B., Sell, S., Stone, R. C., Vrba, F. J., Walker, R. L., Westerhout,

G., Brucato, R. J., Reid, I. N., Schoening, W., Hartley, M., Read, M. A., and Tritton, S. B., \The USNO-B

Catalog," The Astrophysical Journal 125 , 984{993 (2003).

[12] Connolly, A. J., Angeli, G. Z., Chandrasekharan, S., Claver, C. F., Cook, K., Ivezic, Z., Jones, R. L.,

Krugho?, K. S., Peng, E.-H., Peterson, J., Petry, C., Rasmussen, A. P., Ridgway, S. T., Saha, A., Sembroski,

G., vanderPlas, J., and Yoachim, P., \An end-to-end simulation framework for the Large Synoptic Survey

Telescope," Proc. SPIE 9150 , 915014 (2014).

[13] Lotz, P., Dubois-Felsmann, Gregory, P., Lim, K.-T., Johnson, T., Chandrasekharan, S., Mills, D., Daly, P.,

Schumacher, G., Delgado, F., Pietrowicz, S., Selvy, B., Sebag, J., Marshall, S., Sundararaman, H., Contaxis,

C., Bovill, R., and Jeness, T., \LSST control software component design," Proc. SPIE 9913 , Paper 9, in

press (2016).

[14] Mills, D., Schumacher, G., and Lotz, P., \LSST communications middleware implementation," Proc. SPIE

9906 , Paper 204, in press (2016).

[15] Daly, Phil, N., Schumacher, G., Delgado, F., and Mills, D., \LSST OCS plan and status," Proc. SPIE 9913 ,

Paper 112, in press (2016).

[16] Jenness, T., Bosch, J., Owen, R., Parejko, J., Sick, J., Swinbank, J., de Val-Borr, M., Dubois-Felsmann, G.,

Lim, K.-T., Lupton, Robert, H., Schellart, P., Krugho?, Simon, K., and Tollerud, Erik, J., \Investigating

interoperability of the LSST data management software stack with astropy," Proc. SPIE 9913 , Paper 16,

in press (2016).

[17] Juri?c, M., Kantor, J., Lim, K.-T., Lupton, R. H., Dubois-Felsmann, G., Jenness, T., Axelrod, T. S., Aleksi?c,

J., Allsman, R. A., AlSayyad, Y., Alt, J., Armstrong, R., Basney, J., Becker, A. C., Becla, J., Bickerton,

S. J., Biswas, R., Bosch, J., Boutigny, D., Carrasco Kind, M., Ciardi, D. R., Connolly, A. J., Daniel, S. F.,

Daues, G. E., Economou, F., Chiang, H.-F., Fausti, A., Fisher-Levine, M., Freemon, D. M., Gee, P., Gris,

P., Hernandez, F., Hoblitt, J., Ivezi?c, Z.,? Jammes, F., Jevremovi?c, D., Jones, R. L., Bryce Kalmbach,

J., Kasliwal, V. P., Krugho?, K. S., Lang, D., Lurie, J., Lust, N. B., Mullally, F., MacArthur, L. A.,

Melchior, P., Moeyens, J., Nidever, D. L., Owen, Russell aNeillnd Parejko, J. K., Peterson, J. M., Petravick,

D., Pietrowicz, S. R., Price, P. A., Reiss, D. J., Shaw, R. A., Sick, J., Slater, C. T., Strauss, M. A.,

Sullivan, I. S., Swinbank, J. D., Van Dyk, S., Vuj?ci?c, V., Withers, A., Yoachim, P., and LSST Project, f. t.,

\The LSST Data Management System," Astronomical Data Analysis Software and Systems XXV proc.,

arXiv:1512.07914 (2015).

[18] Xin, B., Claver, C., Liang, M., Chandrasekharan, S., Angeli, G., and Shipsey, I., \Curvature wavefront

sensing for the Large Synoptic Survey Telescope," Applied Optics 54 , 9045{ (Oct. 2015).

[19] Roddier, C. and Roddier, F., \Wave-front reconstruction from defocused images and the testing of ground-

based optical telescopes.," Journal of the Optical Society of America A 10 , 2277{2287 (1993).

[20] Gureyev, T. E. and Nugent, K. A., \Phase retrieval with the transport-of-intensity equation. II. Orthogonal

series solution for nonuniform illumination," Journal of the Optical Society of America A: Optics 13 (8),

1670{1682 (1996).