What’s new in version 3.0 ?

Version 3.0 brings an “expert” fit mode, with a fit command line that makes it possible to automate fitting or do parameter space exploration, running many fits with random initial starting positions to greatly increase the likelihood of finding the best parameters (along with many other new features for fits).

Another improvement in the fits is a new implicit fit, to fit an implicit equation to data.

This release also brings in a command to permanently record information about a file, such as the conditions in which the experiments were made (the sample used, the temperature, etc…), in a format that QSoas automatically reads when reading the file. This makes it possible to quickly browse through data to find the experiments you need.

Finally, this release brings in a whole new offline documentation system, with and index and search facilities, startup tips, cross-reference examples and the possibility to specify manually all the command arguments and options in a dialog box when running a command from the menus.

See the full list of changes below

What’s new in version 2.2 ?

Version 2.2 brings in a lot of new features and improvements, notably greatly improved memory use for massive multifits, a fit for linear (in)activation processes, a new way to transform “numbers” like peak position or stats into new datasets and SVG output ! It also finally brings back the peak area output in find-peaks and related commands !

See the full list of changes below

What’s new in version 2.1 ?

Version 2.1 brings in a number of new functionalities, such as a command to solve equations, and ways to reparametrize fits, along with numerous improvements of already existing commands.

See the full list of changes below

What’s new in version 2.0 ?

Version 2.0 brings a whole new set of functionalities for fits, namely a fit engine with enormous performance increases in massive multibuffer fits (easy to reach a 1000 times speed increase !), definition of fits with a distribution of parameters, and many new functions.

See the full list of changes below

Changes in version 3.0 (03.12.2020)

Fits

The fit system has been greatly improved, with an /expert=true mode featuring:

• a command-line interface with over 50 commands in the fit dialog to run fits non-interactively
• parameter space explorers, which run fits systematically with different starting parameters, to improve the chance of finding the best parameters
• possibility to run scripts, even as a fit- command-line option, which means one can now perform fully automatic fits
• the display, handling, reuse of all the previous fit attempts in a fit session (“fit trajectories”), including ways to highlight changes
• the possibility to set the fit dialog title bar
• a Particle Swarm Optimizer fit engine
• the GSL implementation of the simplex as fit engine
• a mode in sim- that just writes the residuals
• a mode in sim- to export subfunctions
• improvements to the parameters spreadsheet
• a new fit, implicit, that fits the results of equations to data, together with a define-implicit-fit to create named implicit fits
• the arb fit now has its corresponding sim-arb function

New commands

• a mem command to query for memory usage and set the size of the cache for loading datasets
• a record-meta command to permanently set meta-data to a given file
• a verify command that aborts the current script if a condition is not verified
• a define-distribution command to define distribution of parameters in fits

Improvements of previous commands

• apply-formula, eval and stats can now work on several buffers
• browse-stack, browse-stack, overlay-buffer now accept the /for-which= option to further select the buffers
• apply-formula now better handles mathematical errors (like sqrt(-1))
• load and related commands now take an /expected= option to abort with an error if the number of loaded datasets is not what one expects
• run, run-for-each and run-for-datasets now provide several ways to deal with errors in the execution of the script: ignore, abort or fail
• reglin can now be used to define a baseline to characterize peaks, and it can also show the exponential equivalent to the line
• cursor has new symmetry operations, and can work effectively with many curves
• set-perp can now take the values of perpendicular coordinates from a dataset row
• possibility to disable optimization in auto-filter-bs
• show-stack can also list the meta-data now
• Gauss-Kronrod integration for distribution of parameters and other multidimensional integrators
• help can now be run without argument to start the new documentation system. With a command argument, it will still take to the documentation of the command.

Other improvements

• new syntax for statistics and meta: one can now use $stats.y_max (etc.) in addition to $stats["y_max"] to refer to the y_max stats (and for all stats and meta too), which means that there is no need anymore for cumbersome double quoting like in eval '$stats["y_max"]'
• completion on the meta/stats names for the code
• a %{ruby code} expansion in scripts or command-line
• a set of functions (k_mhc and k_mhc_double) to Marcus-Hush-Chidsey rate constants (using the integral formula)
• a trumpet_bv function to compute the position of peaks in so-called “trumpet plots” when using Butler-Volmer kinetics
• several additional functions like ln_erfc(x), ln_gamma(x)
• a --log command-line option to choose the log file name
• a --version command-line option
• Gauss-Kronrod integrators for parameter distributions
• full support for Qt5 (support for Qt4 is dropped)
• running a command from the menu now spawns a full dialog box with the possibility to set options
• it is now possible to deduce rate constants from thermodynamic constraints in kinetic systems
• new statistics: y_a and y_b, which correspond to parameters of the linear regression of $y = f(x)$. Available for y2 and all the others, still as a function of x.
• a _sum statistic that corresponds to the sum of all elements (available for each column)
• a new comprehensive offline documentation system with index, search capacities, cross-referenced examples, and the possibility to execute commands directly from within it
• a series of startup tips
• now, using the menus to start a command will show a dialog box in which one can choose the arguments and options

Changes in version 2.2 (2018.03.21)

New commands

• a load-as-parameters load function to load saved fit parameters as normal datasets
• pop, which can be combined with the various /accumulate= options to create datasets on the fly, from the results of commands like stats or 1

Fits

• new fit: linear-kinetic-system, which is useful for fitting the time traces of potential step experiments with linear inactivation (like in Fourmond el al, Nat Chem 2014)
• new time-dependent parameter: ramps
• interpolation of parameters in the “Parameters Spreadsheet”
• chi-squared “goodness of fit”
• qsoas and multi a much more memory-efficient for massive multi-buffer fits
• arb has more “automatic” parameters
• all the sim- commands now also take the /style= and /set-meta= options

Improvements of previous commands

• new picking modes available for all commands that pick points: the local min (Ctrl+M) and max (Ctrl+Shift+M)
• possibility to track the point selected for commands that pick points
• an /exclusive=yes option to flag to remove all the other flags
• most data-producing commands now have an /accumulate option to generate buffers step-by-step, to combine with pop
• new stats: stddev, and quantiles: q10, q25, q75 and q90
• save-datasets can now avoid to overwrite files
• threshold option for strip-if (to avoid creating near-empty buffers)
• load-as-csv now correctly parses CSV files with quotes
• print can now export SVG files
• bin can now use a column for weights, and normalize the sum to 1.
• Finally bring back the peak area for find-peaks, 1 and 2 !
• expand can now group the Y columns 2 by 2, (or 3 by 3, etc.)
• Improvements of unwrap, including the possibility to reverse its effects

Other improvements

• possibility to load a saved stack by just double-clicking on it
• in kinetic systems, can now use either Butler-Volmer or Marcus-Hush-Chidsey kinetics for interfacial electron transfers
• now fully compatible with Qt 5, still not the default build
• switched to using embedded mruby rather than plain ruby, which finally fixes a hard crash on Windows
• new special functions: incomplete gamma functions
• many bug fixes and other minor improvements

Changes in version 2.1 (2017.01.02)

New commands

• solve, to solve equations
• reparametrize-fit, to manipulate the parameters of a fit

Fits

• Reparametrization of the wave shape fits for 1- and 2-electron reversible catalysts ecr-wave and eecr-wave
• Implementation of the corresponding fits for irreversible catalysts, eci-wave and eeci-wave
• The commands that define fits, reparametrize-fit, combine-fits, define-distribution-fit and define-derived-fit can now redefine existing fits with the option /redefine=true.
• The qsoas and multi fit engines can now use scaling, see the “Fit Options” dialog
• Improvements of the parameters spreadsheet

Improvements of previous commands

• echem-peaks now offers all the options of the other peak-finding commands, with the same meaning
• When several datasets are specified, commands that pick points cursor, baseline, etc. can switch between all displayed datasets (with the n and N keys)
• Commands that write to the output file can now also save the data as meta-data through the use of the /set-meta option.
• It is now possible to select the statistics one wants to display/save with the stats command
• diff can now compute an arbitrary derivative to an arbitrary order
• print can now overwrite output PDF file without asking via the /overwrite=true option
• most of the commands that can produce several datasets now take a /flag= option and a /style=, the exception being the sim- commands which still only support the /flag= option at this stage.

Other improvements

• Now compiles and runs with Ruby 2.3 and the most recent GSL releases
• The --stdout command-line option writes the content of the terminal to the standard output
• Sketch of an Emacs major mode, misc/qsoas-mode.el in the tarball
• A series of bug fixes

Changes in version 2.0 (2016.07.19)

New commands

• auto-flag, to automatically add flags to all newly created datasets
• auto-reglin, another way to compute the derivative of a signal
• define-distribution-fit, to define fits with distribution of values for parameters
• hide-buffer, to hide buffers from view
• mintegrate-formula, to integrate functions depending on one parameter
• reverse, to reverse the order of data points
• split-on-values, to split datasets on the values of certain colunms
• system, to run external commands
• timer, that measures the time between two calls
• add and multiply commands
• a debug function to save debug information in a directory, helpful for debugging (reproducible) crashes

Fits

• Load parameters using interpolation on Z values
• Shortcut to fix/unfix a parameter for all datasets (right-clicking)
• Spreadsheet mode for editing parameters
• A “Multi” fit engine that performs very well for large number of buffers
• Can now cancel the current fit using the shell command killall QSoas -SIGUSR1 on Linux and MacOS
• better convergence for the QSoas and Multi fit engines
• Generalize the use of “time-dependent parameters” (the /with=co:2,exp options to fit-kinetic-system) to fit-ode and fit-arb
• New types of time dependent parameters: rexp (exponential relaxation) and steps (discrete steps)
• The sim- commands can now compute the jacobian and reexport parameters (to the output file)
• Now, the fit dialog detects when there are NaNs in the data, which avoid hard-to-understand problems of convergence of fits
• a new pseudo-voigt fit
• a new polynomial fit
• Now, the commands that define fits can redefine existing (custom) fits if given the /redefine=true option
• One can now specify the fit engine from the command-line

Improvements of previous commands

• eval can now use statistics and meta-data of the current buffer
• Can now push the transform using uppercase T in filter-fft, and using the /transform option in auto-filter-fft
• save-datasets can now be used to name files (or buffers) based on meta-data; it can also simply rename without saving, and create directories if necessary (using /mkpath=true)
• run can change to the directory of a script when it is running it, using /cd-to-script=true
• flag can now clear all flags before adding new ones (with /set=true)
• generate-dataset can now generate several buffers at once when using the /number option
• More control in split-monotonic: can now keep only the first and/or last parts, and group several segments into datasets
• In apply-formula and strip-if, one can now use i_0 and x_0 to refer to the index and value of x of the first point of the segment
• strip-if is now aware of the metadata and stats, like apply-formula

Other improvements

• latest and latest:n specifications to refer to the datasets generated by the latest command (or the _n_th before)
• Faster startup time
• New special functions: Dawson integral, Debye functions, the clausen function, a couple of Fermi-Dirac integrals, multigamma functions, some hypergeometric functions, and an approximation of Marcus-Hush-Chidsey kinetics
• One can now use the --run and --exit-after-running command-line options to QSoas to run scripts from the command-line
• possibility to run “inline” Ruby code from within scripts and from QSoas itself
• Can now run QSoas commands from within Ruby code, for instance QSoas::load("file"). This is still experimental.
• Now supports GSL 2.0
• Quite a few bug fixes