You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
This repository was archived by the owner on Feb 12, 2024. It is now read-only.
Overall, our project now consists of lines of Python code and lines of C++ code.
221
225
222
226
\section{Changes on the Design}
227
+
223
228
\begin{itemize}
229
+
224
230
\item We didn't implemented a matrix class, because we realized it is not useful to handle every matrix by its own.
225
231
Instead we decided to handle it in numpy arrays.
232
+
233
+
\item We decided to leave out the density because it wasn't
234
+
226
235
\item We changed the command interface to an abstract class to reduce redundancy of the code.
236
+
237
+
\item We decided to let out the possibility to select a density, because we couldn't guarantee that the fetched and cut matrices from Suite Sparse have the wanted density.
238
+
227
239
\end{itemize}
228
240
229
241
%collector class?
@@ -262,69 +274,114 @@ \subsection{Following Requirements are accomplished}
262
274
263
275
\subsection{Following Requirements were accomplished as far as possible}
264
276
\begin{itemize}
277
+
265
278
\item All mandatory requirements were as far as possible accomplished despite the cross-platformed compatibility is not fully given.
266
279
This was not possible, because some used entities (ssget, gingko) where not Windows compatible.
267
280
Compared to the specification sheet there is just the possibility to fetch and cut Suite Sparse matrices yet.
268
281
As we figured out our design we realized that generating random matrices is not that easy with our knowing.
269
282
So as already realized in the design document, because in Suite Sparse it is very rare to have same sized matrices from Suite Sparse, it was not necessary to just fetch Suite Sparse matrices in one size.
270
283
Instead we just implemented a generator that fetches and cuts Suite Sparse matrices.
271
-
\item We decided to let out the possibility to select a density, because we couldn't guarantee that the fetched and cut matrices from Suite Sparse have the wanted density.
284
+
285
+
\item We decided to let out the possibility to select a density as explained in the changes of design
286
+
272
287
\end{itemize}
273
288
274
289
275
290
\subsection{Following Requirements were not accomplished}
291
+
\begin{itemize}
292
+
\item A web interface to the software that is able to select a single, a set or all matrices of an uploadable le for prediction by the neural network. The web interface may also be able to visualize the contained matrices, annotated labels as well as prediction results.
293
+
294
+
\item The system must support at least five iterative solver algorithms.
295
+
296
+
\item The software must be able to utilize GPU accelerators for the training and prediction capabilities of the neural network.
297
+
298
+
\item Scalability of the workflow including matrix generation, training, prediction in that multiple processors may be used in parallel.
276
299
300
+
\end{itemize}
277
301
278
302
\section{Unittests}
303
+
279
304
\subsection{Controller}
280
-
\subsubsection{test_command_parser}
305
+
\subsubsection{test command parser}
281
306
This test checks the functionality of the command parser.
0 commit comments