|Aug 9, 2019|
|1 year 2 months ||NEW FILTER: FilterMaths to evaluate arithmetic expressions with
meteo data available. It's quite powerful! Also listing Kalman and
particle filter in the processing elements because all the
disclaimers are there.
|Aug 8, 2019|
|1 year 2 months ||Clang warnings regarding uninitialized values (hopefully) fixed,
please confirm. ZRXP changes for character encoding and parameter
units to make the plugin a little bit less, but the toolchain much
|1 year 2 months ||Minor casting issues found out by CLang
|Aug 7, 2019|
|1 year 2 months ||Enabling default CSV delimiter as per the documentation.
|Aug 5, 2019|
|1 year 2 months ||Redid the RasPi RNG tests and benchmarks on a strong laptop. Minor
cleanup and doc enhancements for PF & KF. Some unimportant typos.
|Jul 30, 2019|
|1 year 2 months ||A C99 function slipped in the particle filter undetected...
|1 year 2 months ||Handling another special case in the ZRXP plugin to better transport
untouched nodata values. Forgotten doc fix for TIMELOOP, slight
filters doc enhancement, small typos.
|Jul 29, 2019|
|1 year 2 months ||Adding example program an data for the statistical filters.
|1 year 2 months ||Finalizing the documentation for KF & PF with two more examples.
KF: output error estimation as standard deviation, alias for
substitution '1st'. PF: memory bugfix for model fitting, less
particles as default.
|Jul 25, 2019|
|1 year 2 months ||Documentation for the new filters. It is reachable via 'Modules' but
not linked in the processing list yet. KF: filter only the param it
runs on by default, better nodata handling, system matrix can also
read a vector, cleanup. PF: bugfix when using…
|Jul 18, 2019|
|1 year 3 months ||PF: Reintroducing previously kicked idea to use data points as model
by means of fitting a curve. Alternative choice for most likely
path. System and observation model's substitutions variables don't
have to match anymore. KF: Output of evolution of…
|Jul 15, 2019|
|1 year 3 months ||The time Loop filter was using a different name than in its
documentation. It really is TimeLoop!
|1 year 3 months ||Slightly improved documentation and smarter warning when duplicate
timestamps are made of nodata values. And by the way, I did see the
previous commit :-)
|1 year 3 months ||Quick removal of duplicate code from the last commit before anyone
|1 year 3 months ||'Trim' overload to return strings. KF: Should now be well-defined,
first results have been achieved. Input matrices can hold brackets
and be built from diagonal vectors, initial state can be set
automatically from meteo data, better parse errors. PF…
|1 year 3 months ||Adding a missing documentation entry
|Jul 12, 2019|
|1 year 3 months ||Oups, my previous commit was a mess...
|1 year 3 months ||Added the possibility to add custom compiler flags (this should help
for issue 740) and to turn LTO off when compiling with optimizations
(for systems with little available RAM). Both are marked as
"advanced" and thus hidden from view for the normal…
|Jul 11, 2019|
|1 year 3 months ||PF & KF: nodata handling and adaptive timestepping; some
const-ness. KF: input of time-dependent system matrix, core fix (but
still not usable very well).
|Jul 10, 2019|
|1 year 3 months ||PF: Possibility to dump internal states to file system and read it
at next startup --> on successive calls the past does not have to
be recalculated and observations can be aggregated online. Smaller
Eigen include, parse_args fix. KF: Core algorithm…