path: root/presentation.tex

\documentclass{presentation}

\title{Development of \\ Frequency Domain Multidimensional Spectroscopy}
\subtitle{---Beyond Two Dimensions---}
\author{Blaise Thompson}

\institute{University of Wisconsin--Madison}
\date{2018-04-23}

\begin{document}
\maketitle

\begin{frame}{Brown et al. (1999)}
  \begin{columns}
    \begin{column}{0.5\textwidth}
      \fbox{\adjincludegraphics[width=\textwidth]{"literature/BrownEmilyJ1999a"}}
    \end{column}
    \begin{column}{0.5\textwidth}
      \includegraphics[width=\textwidth]{"literature/BrownEmilyJ1999a_1"}
      \centering
      \\
      \vspace{2\baselineskip}
      $\vec{k_{\text{sig}}} = \vec{k_a} - \vec{k_b} + \vec{k_c}$
    \end{column}
  \end{columns}
\end{frame}

\begin{frame}{Overview}
  \adjincludegraphics[width=\textwidth]{"mixed_domain/simulation overview"}
\end{frame}

\begin{frame}{Diversity}
  Great diversity of experimental strategies.
  \vspace{\baselineskip} \\
  Different phase matching conditions...
  \begin{itemize}
    \item transient grating $\vec{k_a} - \vec{k_b} + \vec{k_c}$
    \item transient absorption
    \item DOVE
    % TODO: darien's experiments
  \end{itemize}
  But also different color combinations and dimensions explored.
  % SAY: based on the same basic ability to scan pulses in frequency, delay etc
\end{frame}

\begin{frame}{Pipeline}
  \adjincludegraphics[width=0.5\textwidth]{presentation/pipe}
  What does the ``pipeline'' of MR-CMDS data acquisition and processing look like in the Wright
  Group?
  \vspace{\baselineskip} \\
  How to increase data throughput and quality, while decreasing frustration of experimentalists?  %
\end{frame}

\begin{frame}{MR-CMDS development}
  [SUMMARY SLIDE FOR REMAINDER OF PRESENTATION]
\end{frame}

\section{Tunability}  % ===========================================================================

\begin{frame}{Tunability}
  \centering \huge
  Control and Calibration of \\
  Optical Parametric Amplifiers
\end{frame}

\begin{frame}{Two strategies for CMDS}
  Two strategies for collecting multidimensional spectra:
  \vspace{\baselineskip} \\
  \begin{columns}
    \begin{column}{0.4\textwidth}
      Time Domain
      \begin{itemize}
        \item broadband pulses
        \item resolve spectra interferometrically
        \item fast (even single shot)
        \item robust
      \end{itemize} 
    \end{column}
    \begin{column}{0.4\textwidth}
      Frequency Domain
      \begin{itemize}
        \item narrowband pulses
        \item resolve spectra by tuning OPAs directly
        \item slow (lots of motor motion)
        \item fragile
      \end{itemize}
    \end{column}
  \end{columns}
\end{frame}

\begin{frame}{Postage stamp}
  [FIGURE FROM LIT]
\end{frame}

\begin{frame}{Czech}
  [FIGURE FROM CZECH]
\end{frame}

\begin{frame}{Bandwidth}
  \adjincludegraphics[width=\textwidth]{opa/OPA_ranges}
\end{frame}

\begin{frame}{TOPAS-C}
  \includegraphics[width=\textwidth]{opa/TOPAS-C}
  Two ``stages'', each with two motorized optics.
\end{frame}

\begin{frame}{Tuning}
  % TODO: curve plot?
  Tuning curves---recorded correspondence between motor positions and output color.
  \vspace{\baselineskip} \\
  Exquisite sensitivity to alignment and lab conditions---tuning required roughly once a week.
  \vspace{\baselineskip} \\
  Manual tuning is difficult...
  \begin{itemize}
    \item high dimensional motor space
    \item difficult to asses overall quality
    \item several hours of work per OPA (sometimes, an entire day for one OPA)
  \end{itemize}
\end{frame}

\begin{frame}{Preamp}
  \includegraphics[width=\textwidth]{opa/preamp}
\end{frame}

\begin{frame}{Automation}
  \begin{columns}
    \begin{column}{0.5\textwidth}
      \adjincludegraphics[width=\textwidth]{opa/autotune_preamp}
    \end{column}
    \begin{column}{0.5\textwidth}
      Fully automated OPA tuning
      \begin{itemize}
        \item less than 1 hour per OPA
        \item can be scheduled for odd times
        \item high quality from global analysis 
        \item reproducible
        \item unambiguous representations automatically generated
      \end{itemize}
      \vspace{\baselineskip}
      Other calibration steps also automated.
    \end{column}
  \end{columns}
\end{frame}

\section{Acquisition}  % ==========================================================================

\begin{frame}{Acquisition}
  \centering \huge
  Control of the MR-CMDS \\
  Instrument
\end{frame}

\begin{frame}{The instrument}
  Many kinds of component hardware
  \begin{itemize}
    \item monochromators
    \item delay stages
    \item filters
    \item OPAs
  \end{itemize}
  $\sim10$ settable devices, $\sim25$ motors, multiple detectors.
\end{frame}

\begin{frame}{Acquisition}
  PyCMDS---unified software for controlling hardware and collecting data.
  \adjincludegraphics[width=\textwidth]{acquisition/screenshots/000}
\end{frame}

\begin{frame}{Central loop}
  At its core, PyCMDS does something very simple...
  \vspace{\baselineskip} \\ 
  Set, wait, read, wait, repeat.  % TODO: better figure
  \vspace{\baselineskip} \\
  Everything is multi-threaded (simultaneous motion, simultaneous read).
  \begin{itemize}
    \item decrease scan time by up to $\sim2\times$, more for complex experiments
  \end{itemize}
\end{frame}

\begin{frame}{Extensibility}
\end{frame}
  
\subsection{Hardware}  % --------------------------------------------------------------------------

\begin{frame}{Modular Hardware Model}
  [DARIEN ADDED AEROTECH IN ONE DAY]
  [I ADDED NEW OPA IN TWO DAYS]
\end{frame}

\subsection{Acquisitions}  % ----------------------------------------------------------------------

\begin{frame}{Acquisition Modules}
  [SUNDEN ADDED POYNTING TUNE IN SEVERAL DAYS]
\end{frame}
  
\subsection{Queue}  % -----------------------------------------------------------------------------

\begin{frame}{Queue}
  \adjincludegraphics[width=\textwidth]{acquisition/screenshots/004}
\end{frame}

\begin{frame}{Queue}
  This strategy can be incredibly productive!
  \begin{itemize}
    \item Soon after the queue was first implemented, we collected more pixels in two weeks than
      had been collected over the previous three years.
  \end{itemize}
\end{frame}

\section{Artifacts}  % ============================================================================

\begin{frame}{Acquisition}
  \centering \huge
  Artifact Rejection
\end{frame}

\begin{frame}{Shots Processing}
  [DIGITAL SHOTS PROCESSING---NO MORE BOXCARS]
\end{frame}
 
\section{Processing}  % ===========================================================================

\begin{frame}{Processing}
  \centering \huge
  Data Processing
\end{frame}

\begin{frame}{Dimensionality}
  \begin{columns}
    \begin{column}{0.4\textwidth}
      \adjincludegraphics[width=\textwidth]{presentation/TOC}
    \end{column}
    \begin{column}{0.6\textwidth}
      TEST
    \end{column}
  \end{columns}
\end{frame}

\begin{frame}{Flexible data model}
  Flexibility to transform into any desired ``projection'' on component variables.
  \adjincludegraphics[width=\textwidth]{processing/fringes_transform}
  % mention: including expressions
\end{frame}

\section{Conclusion}  % ===========================================================================

\begin{frame}{Conclusion}
  [CONCLUSION]
\end{frame}

\begin{frame}{Acknowledgments}
  \begin{columns}
    \begin{column}{0.5\textwidth}
      Wright Group
      \begin{itemize}
        \item Kyle Sunden
        \item Darien Morrow
        \item Jonathan Handali
        \item Nathan Neff-Mallon
        \item Kyle Czech         
        \item Dan Kohler
        \item Erin Boyle
        \item Paul Hebert
        \item Skye Kain
        \item John
        \item (and more...)
      \end{itemize}
    \end{column}
    \begin{column}{0.5\textwidth}
      Committee
      \begin{itemize}
        \item Kyoung-Shin Choi
        \item Randall Goldsmith
        \item Tim Bertram
      \end{itemize}
      \vspace{\baselineskip}
      UW-Madison Chemistry Department
      \begin{itemize}
        \item Rob McClain
        \item Pam Doolittle
        \item Bill Goebel
        \item Steve Myers
      \end{itemize}
      \vspace{\baselineskip}  
      You, the audience!
      \hl{Questions?}
    \end{column}
  \end{columns}
\end{frame}
  
\section{Supplement}  % ===========================================================================
 
\begin{frame}{Modular hardware model}
  \adjincludegraphics[scale=0.25]{acquisition/hardware_inheritance}
\end{frame}

\begin{frame}{Modular sensor model}
  Can have as many sensors as needed.
  \vspace{\baselineskip} \\
  Each sensor contributes one or more channels.
  \vspace{\baselineskip} \\
  Sensors with size contribute new variables (dimensions).
\end{frame}

\begin{frame}{Universal format}
  WrightTools defines a \emph{universal file format} for CMDS.
  \begin{itemize}
    \item store multiple multidimensional arrays
    \item metadata
  \end{itemize}
  Import data from a variety of sources.
  \begin{itemize}
    \item previous Wright Group acquisition software
    \item commercial instruments (JASCO, Shimadzu, Ocean Optics)
  \end{itemize}
\end{frame}

\begin{frame}{Domains of CMDS}
  CMDS can be collected in two domains:
  \begin{itemize}
    \item time domain
    \item frequency domain
  \end{itemize}
\end{frame}

\begin{frame}{Time domain}
  Multiple broadband pulses are scanned in \emph{time} to collect a multidimensional interferogram
  (analogous to FTIR, NMR).
  \vspace{\baselineskip} \\
  A local oscillator must be used to measure the \emph{phase} of the output.
  \vspace{\baselineskip} \\
  This technique is...
  \begin{itemize}
    \item fast (even single shot)
    \item robust
  \end{itemize}
  pulse shapers have made time-domain CMDS (2DIR) almost routine.
\end{frame}

\begin{frame}{Frequency domain}
  In the Wright Group, we focus on \emph{frequency} domain ``Multi-Resonant'' (MR)-CMDS.
  \vspace{\baselineskip} \\
  Automated Optical Parametric Amplifiers (OPAs) are used to produce relatively narrow-band pulses.
  Multidimensional spectra are collected ``directly'' by scanning OPAs against each-other.
  \vspace{\baselineskip} \\
  This strategy is...
  \begin{itemize}
    \item slow (must directly visit each pixel)
    \item fragile (many crucial moving pieces)
  \end{itemize}
  but! It is incredibly flexible.  
\end{frame}

\begin{frame}{Selection rules}
  MR-CMDS can easily collect data without an external local oscillator.
  \vspace{\baselineskip} \\
  This means... [BOYLE]
\end{frame}

\begin{frame}{MR-CMDS theory}
\end{frame}

\begin{frame}{Mixed domain}
  [FIGURES FROM DAN'S PAPER]
\end{frame}

\end{document}