thesis: improve conclusion

thesis/main.gls

-\glossarysection[\glossarytoctitle]{\glossarytitle}\glossarypreamble
-\begin{theglossary}\glossaryheader
-\glsgroupheading{A}\relax \glsresetentrylist %
-\glossentry{automaticdifferentiation}{\glossaryentrynumbers{\relax 
-		\setentrycounter[]{page}\glsnumberformat{27}}}\glsgroupskip
-\glsgroupheading{B}\relax \glsresetentrylist %
-\glossentry{blackbox}{\glossaryentrynumbers{\relax 
-		\setentrycounter[]{page}\glsnumberformat{21\delimN 22}}}%
-\glossentry{burnup}{\glossaryentrynumbers{\relax 
-		\setentrycounter[]{page}\glsnumberformat{6}\delimN 
-		\setentrycounter[]{page}\glsnumberformat{47}}}\glsgroupskip
-\glsgroupheading{C}\relax \glsresetentrylist %
-\glossentry{clustermanager}{\glossaryentrynumbers{\relax 
-		\setentrycounter[]{page}\glsnumberformat{28}}}%
-\glossentry{cycletime}{\glossaryentrynumbers{\relax 
-		\setentrycounter[]{page}\glsnumberformat{6}}}\glsgroupskip
-\glsgroupheading{E}\relax \glsresetentrylist %
-\glossentry{enrichmentgrade}{\glossaryentrynumbers{\relax 
-		\setentrycounter[]{page}\glsnumberformat{3}}}\glsgroupskip
-\glsgroupheading{F}\relax \glsresetentrylist %
-\glossentry{filesystem}{\glossaryentrynumbers{\relax 
-		\setentrycounter[]{page}\glsnumberformat{23}}}%
-\glossentry{fissile}{\glossaryentrynumbers{\relax 
-		\setentrycounter[]{page}\glsnumberformat{3\delimN 4}\delimN 
-		\setentrycounter[]{page}\glsnumberformat{7\delimR 9}}}\glsgroupskip
-\glsgroupheading{G}\relax \glsresetentrylist %
-\glossentry{groundtruth}{\glossaryentrynumbers{\relax 
-		\setentrycounter[]{page}\glsnumberformat{21}\delimN 
-		\setentrycounter[]{page}\glsnumberformat{35}\delimN 
-		\setentrycounter[]{page}\glsnumberformat{38}\delimN 
-		\setentrycounter[]{page}\glsnumberformat{40}\delimN 
-		\setentrycounter[]{page}\glsnumberformat{42\delimN 43}\delimN 
-		\setentrycounter[]{page}\glsnumberformat{45}}}\glsgroupskip
-\glsgroupheading{H}\relax \glsresetentrylist %
-\glossentry{hyperparameter}{\glossaryentrynumbers{\relax 
-		\setentrycounter[]{page}\glsnumberformat{22}\delimN 
-		\setentrycounter[]{page}\glsnumberformat{33}}}\glsgroupskip
-\glsgroupheading{I}\relax \glsresetentrylist %
-\glossentry{inference}{\glossaryentrynumbers{\relax 
-		\setentrycounter[]{page}\glsnumberformat{11}\delimN 
-		\setentrycounter[]{page}\glsnumberformat{13}\delimN 
-		\setentrycounter[]{page}\glsnumberformat{16\delimN 17}}}%
-\glossentry{irradiationtime}{\glossaryentrynumbers{\relax 
-		\setentrycounter[]{page}\glsnumberformat{6}}}%
-\glossentry{isotope}{\glossaryentrynumbers{\relax 
-		\setentrycounter[]{page}\glsnumberformat{3}}}\glsgroupskip
-\glsgroupheading{N}\relax \glsresetentrylist %
-\glossentry{netcdf}{\glossaryentrynumbers{\relax 
-		\setentrycounter[]{page}\glsnumberformat{28}}}%
-\glossentry{nuclearweapon}{\glossaryentrynumbers{\relax 
-		\setentrycounter[]{page}\glsnumberformat{1}}}\glsgroupskip
-\glsgroupheading{P}\relax \glsresetentrylist %
-\glossentry{parameter}{\glossaryentrynumbers{\relax 
-		\setentrycounter[]{page}\glsnumberformat{1}}}\glsgroupskip
-\glsgroupheading{R}\relax \glsresetentrylist %
-\glossentry{randomseed}{\glossaryentrynumbers{\relax 
-		\setentrycounter[]{page}\glsnumberformat{24}}}\glsgroupskip
-\glsgroupheading{S}\relax \glsresetentrylist %
-\glossentry{SQLite}{\glossaryentrynumbers{\relax 
-		\setentrycounter[]{page}\glsnumberformat{23}}}\glsgroupskip
-\glsgroupheading{T}\relax \glsresetentrylist %
-\glossentry{thermalneutron}{\glossaryentrynumbers{\relax 
-		\setentrycounter[]{page}\glsnumberformat{5}}}\glsgroupskip
-\glsgroupheading{W}\relax \glsresetentrylist %
-\glossentry{weaponsgradePu}{\glossaryentrynumbers{\relax 
-		\setentrycounter[]{page}\glsnumberformat{3}\delimN 
-		\setentrycounter[]{page}\glsnumberformat{6}\delimN 
-		\setentrycounter[]{page}\glsnumberformat{9}}}%
-\end{theglossary}\glossarypostamble

thesis/mainmatter/conclusion.tex

 \mychapter{Conclusion and Outlook}{conclusion}
 
 The method developed in this work has been shown to be feasible, and produce useful results. It confirms that Bayesian
-inference is useful in the context of nuclear archaeology; in doing this, it joins \cite{Antonio2019} and
-\cite{Gttsche:772987}. The reasons \emph{why} it works, and why ``conventional'' methods may not be able to achieve
-comparable results in the given scenario, have also been discussed in detail.
+inference is useful in the context of nuclear archaeology, which has been shown before in e.g.
+\cite{Antonio2019} and \cite{Gttsche:772987}. The reasons \emph{why} it works, and why ``conventional'' methods may not
+be able to achieve comparable results in the given scenario, have also been discussed in detail.
 
 The presented results for up to three independent fuel cycle parameters are promising when using synthetic
 measurements; and the implementation is ready to use real measurements instead of generating measurements from the same
-mechanism used for the reconstruction process. Thus, a logical next step is using the approach not only on such
-\emph{synthetic} measurements, but to attempt reconstructing an actual fuel cycle. This is obviously more complex than
-relying on an artificial ``perfect'' example, as was done here.
+mechanism used for the reconstruction process.
 
+Reconstructing more parameters is one goal, as this allows for reconstructing fuel cycles even if few details
+are known with good accuracy. It was shown that the existing implementation can already handle ambiguities if provided
+information is not sufficient. For improving both accuracy and inference run time, differentiating the fuel cycle
+simulation may prove worthwile; this has the potential of better informing the more modern sampling algorithms
+available.
+
+A next step bringing the method closer to actual applications is to use not only \emph{synthetic} measurements, but to
+attempt reconstructing an actual fuel cycle. This is obviously more complex than relying on an artificial ``perfect''
+example, as was done here.
 One way to go about this would be verifying a fuel cycle of limited scope: search for inconsistencies in reported data,
-by reconstructing reported (dubious) parameters from remaining data, or externally observed parameters. A main hurdle
-in such an endeavor is modeling the fuel cycle itself, a matter not treated very thoroughly in this work. In order to
-improve reconstruction quality, additional measurements (more isotopes, for example) can be taken into account for such
-an attempt. An evaluation like this will uncover weaknesses of the method, which was developed using synthetic
-measurements, and allow improvements to be made to the inference code. An interesting topic to emphasize is
-finding a more physically accurate way to model uncertainties in a way that produces sensible posterior probability
+by reconstructing reported (dubious) parameters from remaining data, or externally observed parameters. The likely
+biggest obstacle in doing so is modeling the fuel cycle itself, a matter not treated very thoroughly in this
+work. In order to improve reconstruction quality, additional measurements (more isotopes, for example) can be taken
+into account for such an attempt. An evaluation like this will uncover weaknesses of the method, which was developed
+using synthetic measurements, and allow improvements to be made to the inference code. An interesting topic to emphasize
+is finding a more physically accurate way to model uncertainties in a way that produces sensible posterior probability
 densities.

thesis/mainmatter/introduction.tex

@@ -11,10 +11,9 @@ One of the best known examples for mostly uncontrolled nuclear armament is North
 several nuclear warheads for use on medium-range missiles, and is likely working on intercontinental ballistic missile
 capabilities. Details about its nuclear program remain unclear, for example which reactors are in use, and how they
 contribute to the nuclear inventory. Its uranium enrichment program keeps a low profile, making it difficult to
-estimate progress in this area. Occasional inspections have taken place, but not by international
-bodies like the International Atomic Energy Agency (\textsc{IAEA}). There is no reliable information about the
-\emph{nuclear fuel cycle} -- the chain of processes producing and using nuclear materials for civil or military use --
-of North Korea. \cite{BASNorthKorea21}
+estimate progress in this area. Occasional inspections have taken place, but not by international bodies like the
+\gls{iaea}. There is no reliable information about the \emph{nuclear fuel cycle} -- the chain of processes producing and
+using nuclear materials for civil or military use -- of North Korea. \cite{BASNorthKorea21}
 
 More is known about Iran's nuclear program, as it is permitting inspections by IAEA. However, it continues
 enrichment of uranium beyond levels required for purely civilian use since the United States have withdrawn from the

thesis/mainmatter/methodology.tex

@@ -118,7 +118,7 @@ The \cd{blackbox} module is responsible for interfacing with \cyclus~and calcula
 parameter sample; the sample's observations are then used for calculating the likelihood.
 
 For interfacing with \cyclus, the \cd{CyclusCliModel} class was implemented. There are essentially two ways of
-interfacing: either using the Python library coming with \cyclus, or executing the \cd{cyclus} binary from the
+interfacing: either using the Python library coming with \cyclus, or executing the \cd{cyclus} \gls{binary} from the
 inference process. The former option was originally preferred, but doesn't work in some scenarios enforced by \pymc:
 for example, a parallel execution from the same process was difficult to achieve. In addition to that, simulation
 result files were often left on the file system, becoming a problem when tens of thousands of
@@ -362,8 +362,8 @@ parallel in each instance. The former method has proven to be more reliable, and
 experiments.}
 
 Generating one sample takes more than one simulation execution, depending on the sampling algorithm. The \textsc{Slice}
-employed here uses usually less than ten simulation executions when inferring two parameters, but this number increases
-quickly with additional parameter space dimensions.
+algorithm employed here usually needs to run less than ten simulation executions when inferring two parameters, but
+this number increases with additional parameter space dimensions.
 
 Each instance saves the incomplete state every 100 to 200 samples to a \gls{netcdf} file on a shared cluster file
 system, where it can be inspected later. It was found that for 10 to 15 chains, even as few samples as 200 per chain

thesis/mainmatter/nfc_physics.tex

 \mychapter{Physics of Nuclear Fuel Cycles}{nfcphysics}
 
-A nuclear fuel cycle is the sequence of steps processing nuclear material. One goal of many fuel cycles is
+A \gls{nfc} is the sequence of steps processing nuclear material. One goal of many fuel cycles is
 producing electricity in nuclear reactors, but a fuel cycle can also result in \gls{fissile} material for use in nuclear
 weapons. A fuel cycle can either be ``once-through'', as in this work: in this case, uranium enters the fuel cycle, is
 processed through multiple steps, and electricity as well as fissile material are produced; final products are stored
@@ -21,13 +21,13 @@ reactor.
 \item[Reactor,] modeled after the american \emph{Savannah River Site} K-reactor \cite{NucWeapHandbook87}, which is
 mainly used for production of \gls{weaponsgradePu} (also simulated in this fuel cycle).
 \item[Reprocessing plant,] a reprocessing facility separating spent fuel chemically into uranium and plutonium as
-well as leftover waste \gls{isotope}s (fission and decay products).
+well as leftover waste \gls{isotope}s (\gls{fission} and decay products).
 \item[Plutonium storage,] receives the plutonium from the \emph{Separations} facility.
 \item[Reprocessing waste dump,] receives the separated waste from the \emph{Separations} plant.
 \item[Re-enrichment facility,] enriches the separated uranium from the spent fuel, which is produced by the
 reprocessing plant.
-\item[Depleted uranium dump,] receives depleted uranium from the re-enrichment facility.
-\item[Highly enriched uranium storage,] receives highly enriched uranium from the re-enrichment facility.
+\item[Depleted uranium dump,] receives \gls{depletedu} from the re-enrichment facility.
+\item[Highly enriched uranium storage,] receives \gls{heu} from the re-enrichment facility.
 \end{description}
 
 \begin{figure}[h]
@@ -47,7 +47,7 @@ the simulation are discussed.
 
 The \emph{dumps}, containing depleted uranium and reprocessing waste, are important as they will be the
 source of information for later reconstruction attempts. In a real-world scenario, these compositions may be
-available to inspectors, as opposed to stockpiles of highly-enriched uranium or plutonium. Therefore, this work is
+available to inspectors, as opposed to stockpiles of highly-\gls{enrichedu} or plutonium. Therefore, this work is
 mainly concerned with using the available materials, their composition, mass, etc. in order to make statements about the
 history of these materials and therefore the structure and characteristics of the fuel cycle that they have passed
 through.
@@ -158,7 +158,7 @@ material, and already partially burned during the time in the reactor.
 
 The lifecycle of fuel within a reactor can be characterized by (among other parameters) the \emph{specific
 \gls{burnup}}, often measured in \si{\mega\watt\day\per\kilogram} or \si{\mega\watt\day\per\tonne}. It specifies how
-much thermal energy was generated per unit mass of uranium. The specific burn-up is calculated from other parameters,
+much thermal energy was generated per unit mass of uranium. The \gls{sbu} is calculated from other parameters,
 notably the \gls{irradiationtime} (a synonym for \gls{cycletime}):
 
 \begin{equation}

thesis/preamble/glossary.tex

@@ -5,10 +5,11 @@
 \newacronym{hpc}{HPC}{High-Performance Computing}
 \newacronym{iaea}{IAEA}{International Atomic Energy Agency}
 \newacronym{leu}{LEU}{low-\gls{enrichedu}}
+\newacronym{nfc}{NFC}{Nuclear Fuel Cycle}
 \newacronym{nno}{n.n.o.}{not naturally occurring}
 \newacronym{npt}{NPT}{Nuclear Non-Proliferation Treaty}
 \newacronym{pdf}{PDF}{Probability Density Function}
-\newacronym{sbu}{SBU}{Specific burn-up}
+\newacronym{sbu}{SBU}{Specific \gls{burnup}, i.e. burn-up per unit mass}
 
 \newglossaryentry{automaticdifferentiation}{
 name=automatic differentiation,
@@ -50,13 +51,13 @@ description={Synonym of \emph{irradiation time}}
 }
 
 \newglossaryentry{depletedu}{
-name=depleted Uranium,
+name=depleted uranium,
 description={Uranium containing a lower concentration of \uranium{5} than natural Uranium; occurs as waste (or tails)
 in an enrichment process},
 }
 
 \newglossaryentry{enrichedu}{
-name=enriched Uranium,
+name=enriched uranium,
 description={Uranium that has a higher concentration of \uranium{5} than natural Uranium, which contains about
 \SI{0.72}{\percent} of \uranium{5}}
 }
@@ -68,14 +69,20 @@ description={The concentration of \uranium{5} in a sample of Uranium}
 
 \newglossaryentry{filesystem}{
 name=file system,
-description={A software used for storing files on a storage medium, such as a hard drive or a USB pen drive}
+description={A software driver used by an operating system for storing files on a storage medium, such as a hard drive
+or a USB pen drive}
 }
 
 \newglossaryentry{fissile}{
 name=fissile,
-description={The property of an isotope of both being able to undergo fission (being fissionable), and producing
-enough secondary neutrons to sustain a chain reaction. Usually this definition implies that the isotope can be
-fissioned by thermal neutrons}
+description={The property of an isotope of both being able to undergo fission (being fissionable) after capturing a
+thermal neutron. Being fissile usually also implies that the fission reaction produces enough secondary neutrons to
+sustain a chain reaction}
+}
+
+\newglossaryentry{fissionable}{
+name=fissionable,
+description={The ability of a nucleus to fission after colliding with a low or high energy neutron}
 }
 
 \newglossaryentry{fission}{