Working with Scripts

Scripts can be executed using the command:

exec filename

The file can be either a fityk script (usually with extension fit), or a Lua script (extension lua).


Fityk can save its state to a script (info state > It can also save all commands executed (directly or via GUI) in the session to a script (info history >

Since a Fityk script with all the data inside can be a large file, the files may be stored compressed and it is possible to directly read gzip-compressed fityk script (.fit.gz).

Embedded Lua interpreter can execute any program in Lua 5.1. One-liners can be run with command lua:

=-> lua print(_VERSION)
Lua 5.1
=-> lua print("Today is %A."))
Today is Thursday.
=-> lua for n,f in F:all_functions() do print(n, f, f:get_template_name()) end
0       %_1     Constant
1       %_2     Cycle

(The Lua print function in fityk is redefined to show the output in the GUI instead of writing to stdout).

Like in the Lua interpreter, = at the beginning of line can be used to save some typing:

=-> ="Today is %A.")
Today is Thursday.

Similarly, exec= also interprets the rest of line as Lua expressions, but this time the resulting string is executed as a fityk command:

=-> = string.format("fit @%d", math.random(0,5))
fit 17
=-> exec= string.format("fit @%d", math.random(0,5))
# fitting random dataset (useless example)

The Lua interpreter in Fityk has defined global object F which enables interaction with the program:

=-> = F:get_info("version")
Fityk 1.2.1

Now the first example that can be useful. For each dataset write output of the info peaks command to a file named after the data file, with appended ”.out”:

=-> @*: lua F:execute("info peaks >'%s.out'" % F:get_info("filename"))

This and other methods of F are documented in the next section.

Here is an example of Lua-Fityk interactions:

-- load data from files file01.dat, file02.dat, ... file13.dat
for i=1,13 do
    F:execute("@+ < file%02d.dat:0:1::" % i)

-- print some statistics about the loaded data
n = F:get_dataset_count()
print(n .. " datasets loaded.")

total_max_y = 0
for i=0, n-1 do
    max_y = F:calculate_expr("max(y)", i)
    if max_y > total_max_y then
        total_max_y = max_y
print("The largest y: " .. total_max_y)

If a fityk command executed from Lua script fails, the whole script is stopped, unless you catch the error:

-- wrap F:execute() in pcall to handle possible errors
status, err = pcall(function() F:execute("fit") end)
if status == false then
    print("Error: " .. err)

The Lua interpreter was added in ver. 1.0.3. If you have any questions about it, feel free to ask.

Older, but still supported way to automate Fityk is to prepare a stand-alone program that writes a valid fityk script to the standard output. To run such program and execute the output use command:

exec ! program [args...]

Fityk DSL

As was described in Command Line, each line has a syntax:

[[@...:] [with ...] command [”;” command]...] [#comment]

The datasets listed before the colon (:) make a foreach loop. Here is a silly example:

=-> $a=0
=-> @0 @0 @0: $a={$a+1}; print $a

Command that follows the colon is run for each specified dataset in the context of that dataset. This is to say that:

=-> @2 @4: guess Voigt

is equivalent to:

=-> use @2
=-> guess Voigt
=-> use @4
=-> guess Voigt

(except that the letter sets permenently default dataset to @4.

@* stands for all datasets, from @0 to the last one.

Usually, when working with multiple datasets, one executes a command either for a single dataset or for all of them:

=-> @3: guess Voigt  # just for @3
=-> @*: guess Voigt  # for all datasets

The whole line is parsed and partly validated before the execution. This may lead to unexpected errors when the line has multiple semicolon-separated commands:

=-> $a=4; print $a  # print gives unexpected error
Error: undefined variable: $a

=-> $b=2
=-> $b=4; print $b  # $b is already defined at the check time

Therefore, it is recommended to have one command in one line.


The grammar is expressed in EBNF-like notation:

  • (*this is a comment*)
  • A* means 0 or more occurrences of A.
  • A+ means 1 or more occurrences of A.
  • A % B means A (B A)* and the % operator has the highest precedence. For example: term % "+" comment is the same as term ("+" term)* comment.
  • The colon : in quoted string means that the string can be shortened, e.g. "del:ete" means that any of del, dele, delet and delete can be used.

The functions that can be used in p_expr and v_expr are available here and here, respectively. v_expr contains only a subset of functions from p_expr (partly, because we need to calculate symbolical derivatives of v_expr)

Line structure

line      ::=  [statement] [comment]
statement ::=  [Dataset+ ":"] [with_opts] command % ";"
with_opts ::=  "w:ith" (Lname "=" value) % ","
comment   ::=  "#" AllChars*


The kCmd* names in the comments correspond to constants in the code.

command ::=  (
             "deb:ug" RestOfLine              | (*kCmdDebug*)
             "def:ine" define                 | (*kCmdDefine*)
             "del:ete" delete                 | (*kCmdDelete*)
             "del:ete" delete_points          | (*kCmdDeleteP*)
             "e:xecute" exec                  | (*kCmdExec*)
             "f:it" fit                       | (*kCmdFit*)
             "g:uess" guess                   | (*kCmdGuess*)
             "i:nfo" info_arg % "," [redir]   | (*kCmdInfo*)
             "l:ua" RestOfLine                | (*kCmdLua*)
             "=" RestOfLine                   | (*kCmdLua*)
             "pl:ot" [range] [range] Dataset* [redir] | (*kCmdPlot*)
             "p:rint" print_args [redir]      | (*kCmdPrint*)
             "quit"                           | (*kCmdQuit*)
             "reset"                          | (*kCmdReset*)
             "s:et" (Lname "=" value) % ","   | (*kCmdSet*)
             "sleep" expr                     | (*kCmdSleep*)
             "title" "=" filename             | (*kCmdTitle*)
             "undef:ine" Uname % ","          | (*kCmdUndef*)
             "use" Dataset                    | (*kCmdUse*)
             "!" RestOfLine                   | (*kCmdShell*)
             Dataset "<" load_arg             | (*kCmdLoad*)
             Dataset "=" dataset_expr         | (*kCmdDatasetTr*)
             Funcname "=" func_rhs            | (*kCmdNameFunc*)
             param_lhs "=" v_expr             | (*kCmdAssignParam*)
             Varname "=" v_expr               | (*kCmdNameVar*)
             Varname "=" "copy" "(" var_id ")" | (*kCmdNameVar*)
             model_id ("="|"+=") model_rhs    | (*kCmdChangeModel*)
             (p_attr "[" expr "]" "=" p_expr) % "," | (*kCmdPointTr*)
             (p_attr "=" p_expr) % ","        | (*kCmdAllPointsTr*)
             "M" "=" expr                     ) (*kCmdResizeP*)

Other rules

define         ::=  Uname "(" (Lname [ "=" v_expr]) % "," ")" "="
                       ( v_expr |
                         component_func % "+" |
                         "x" "<" v_expr "?" component_func ":" component_func
component_func ::=  Uname "(" v_expr % "," ")"
delete         ::=  (Varname | func_id | Dataset | "file" filename) % ","
delete_points  ::=  "(" p_expr ")"
exec           ::=  filename |
                    "!" RestOfLine |
                    "=" RestOfLine
fit            ::=  [Number] [Dataset*] |
                    "undo" |
                    "redo" |
                    "history" Number |
guess          ::=  [Funcname "="] Uname ["(" (Lname "=" v_expr) % "," ")"] [range]
info_arg       ::=  ...TODO
print_args     ::=  [("all" | ("if" p_expr ":")]
                    (p_expr | QuotedString | "title" | "filename") % ","
redir          ::=  (">"|">>") filename
value          ::=  (Lname | QuotedString | expr) (*value type depends on the option*)
model_rhs      ::=  "0" |
                    func_id |
                    func_rhs |
                    model_id |
                    "copy" "(" model_id ")"
func_rhs       ::=  Uname "(" ([Lname "="] v_expr) % "," ")" |
                    "copy" "(" func_id ")"
load_arg       ::=  filename Lname* |
p_attr         ::=  ("X" | "Y" | "S" | "A")
model_id       ::=  [Dataset "."] ("F"|"Z")
func_id        ::=  Funcname |
                    model_id "[" Number "]"
param_lhs      ::=  Funcname "." Lname |
                    model_id "[" (Number | "*") "]" "." Lname
var_id         ::=  Varname |
                    func_id "." Lname
range          ::=  "[" [expr] ":" [expr] "]"
filename       ::=  QuotedString | NonblankString

Mathematical expressions

expr        ::=  expr_or ? expr_or : expr_or
expr_or     ::=  expr_and % "or"
expr_and    ::=  expr_not % "and"
expr_not    ::=  "not" expr_not | comparison
comparison  ::=  arith % ("<"|">"|"=="|">="|"<="|"!=")
arith       ::=  term % ("+"|"-")
term        ::=  factor % ("*"|"/")
factor      ::=  ('+'|'-') factor | power
power       ::=  atom ['**' factor]
atom        ::=  Number | "true" | "false" | "pi" |
                 math_func | braced_expr | ?others?
math_func   ::=  "sqrt" "(" expr ")" |
                 "gamma" "(" expr ")" |
braced_expr ::=  "{" [Dataset+ ":"] p_expr "}"

The atom rule also accepts some fityk expressions, such as $variable, %function.parameter, %function(expr), etc.

p_expr and v_expr are similar to expr, but they use additional variables in the atom rule.

p_expr recognizes n, M, x, y, s, a, X, Y, S and A. All of them but n and M can be indexed (e.g. x[4]). Example: (x+x[n-1])/2.

v_expr uses all unknown names (Lname) as variables (example: a+b*x^2). Only a subset of functions (math_func) from expr is supported. The tilde (~) can be used to create simple-variables (~5), optionally with a domain in square brackets (~5[1:6]).

Since v_expr is used to define variables and user-defined functions, the program calculates symbolically derivatives of v_expr. That is why not all the function from expr are supported (they may be added in the future).

dataset_expr supports very limited set of operators and a few functions that take Dataset token as argument (example: @0 - shirley_bg(@0)).


Below, some of the tokens produced by the fityk lexer are defined.

The lexer is context-dependend: NonblankString and RestOfLine are produced only when they are expected in the grammar.

Uname is used only for function types (Gaussian) and pseudo-parameters (%f.Area).

Dataset        ::=  "@"(Digit+|"+"|"*")
Varname        ::=  "$" Lname
Funcname       ::=  "%" Lname
QuotedString   ::=  "'" (AllChars - "'")* "'"
Lname          ::=  (LowerCase | "_") (LowerCase | Digit | "_")*
Uname          ::=  UpperCase AlphaNum+
Number         ::=  ?number read by strtod()?
NonblankString ::=  (AllChars - (WhiteSpace | ";" | "#" ))*
RestOfLine     ::=  AllChars*

Fityk library API

Fityk comes with embedded Lua interpreter and this language is used in this section. The API for other supported languages is similar. Lua communicates with Fityk using object F of type Fityk. To call the methods listed below use F:method(), for example F:get_dof() (not Fityk.get_dof()).


Other supported languages include C++, C, Python, Perl, Ruby and Java. Except for C, all APIs are similar.

Unlike in built-in Lua, in other cases it is necessary to create an instance of the Fityk class first. Then you use this object in the same way as F is used below.

The fityk.h header file is the best reference. Additionally, C++ and Python have access to functions from the ui_api.h header. These functions are used in command line versions of fityk (cfityk or its equivalent – samples/

Examples of scripts in all the listed languages and in the samples directory.

Here is the most general function:


Executes a fityk command. Example: F:execute("fit").

The % operator for the string type is pre-set to support Python-like formatting:

= "%d pigs" % 3
= "%d %s" % {3, "pigs"}

Input / output


Query user. In the CLI user is asked for input in the command line, and in the GUI in a pop-up box. As a special case, if the prompt contains string “[y/n]” the GUI shows Yes/No buttons instead of text entry.

Example: TODO


Print string in the output area. The print() function in built-in Lua is redefined to do the same.


Fityk.set_option_as_string(opt, val)

Set option opt to value val. Equivalent of fityk command set opt=val.

Fityk.set_option_as_number(opt, val)

Set option opt to numeric value val.


Returns value of opt (string).


Returns value of opt (real number).


Fityk.load(spec[, d])

Load data to @*d* slot. The first argument is either a string with path or LoadSpec struct that apart from the path has also the following optional members: x_col, y_col, sig_col, blocks, format, options. The meaning of these parameters is the same as described in Loading Data.

Fityk.load_data(d, xx, yy, sigmas[, title])

Load data to @*d* slot. xx and yy must be numeric arrays of the same size, sigmas must either be empty or have the same size. title is an optional data title (string).

Fityk.add_point(x, y, sigma[, d])

Add one data point ((x, y) with std. dev. set to sigma) to an existing dataset d. If d is not specified, the default dataset is used.

Example: F:add_point(30, 7.5, 1).


Returns number of datasets (n >= 1).


Returns default dataset. Default dataset is set by the “use @n” command.


Returns points for dataset d.

  • in C++ – returns vector<Point>
  • in Lua – userdata with array-like methods, indexed from 0.

Each point has 4 properties: x, y, sigma (real numbers) and is_active (bool).


points = F:get_data()
for i = 0, #points-1 do
    p = points[i]
    if p.is_active then
        print(i, p.x, p.y, p.sigma)
1       4.24    1.06    1
2       6.73    1.39    1
3       8.8     1.61    1

General Info

Fityk.get_info(s[, d])

Returns output of the fityk info command as a string. If d is not specified, the default dataset is used (the dataset is relevant for few arguments of the info command).

Example: F:get_info("history") – returns a multiline string containing all fityk commands issued in this session.

Fityk.calculate_expr(s[, d])

Returns output of the fityk print command as a number. If d is not specified, the default dataset is used.

Example: F:calculate_expr("argmax(y)", 0).


Get coordinates of the plotted rectangle, which is set by the plot command. Return numeric value corresponding to given side, which should be a letter L(eft), R(ight), T(op) or B(ottom).

Model info


Returns number of simple-variables (parameters that can be fitted)


Returns array of simple-variables.

  • in C++ – vector<double>
  • in Lua – userdata with array-like methods, indexed from 0.

Returns array of all defined variables.

  • in C++ – vector<Var*>
  • in Lua – userdata with array-like methods, indexed from 0.


variables = F:all_variables()
for i = 0, #variables-1 do
    v = variables[i]
    print(i,, v:value(), v.domain.lo, v.domain.hi,
          v:gpos(), v:is_simple())

Var.is_simple() returns true for simple-variables.

Var.gpos() returns position of the variable in the global array of parameters (Fityk.all_parameters()), or -1 for compound variables.


Returns variable $name.


Returns array of functions.

  • in C++ – vector<Func*>
  • in Lua – userdata with array-like methods, indexed from 0.


f = F:all_functions()[0] -- first functions
print(, f:get_template_name())          -- _1        Gaussian
print(f:get_param(0), f:get_param(1))         -- height  center
print("$" .. f:var_name("height"))            -- $_4
print("center:", f:get_param_value("center")) -- center: 24.72235945525
print("f(25)=", f:value_at(25))               -- f(25)=  4386.95533969

Return the function with given name, or NULL if there is no such function.


f = F:get_function("_1")
print("f(25)=", f:value_at(25))               -- f(25)=  4386.95533969
Fityk.get_components(d[, fz])

Returns %functions used in dataset d. If fz is Z, returns zero-shift functions.


func = F:get_components(1)[3] -- get 4th (index 3) function in @1
print(func)                   -- <Func %_6>
vname = func:var_name("hwhm")
print(vname)                  -- _21
v = get_variable(vname)
print(v, v:value())           -- <Var $_21>       0.1406587
Fityk.get_model_value(x[, d])

Returns the value of the model for dataset @d at x.

Fit statistics


Returns WSSR (weighted sum of squared residuals).


Returns SSR (sum of squared residuals).


Returns R-squared.


Returns the number of degrees of freedom (#points - #parameters).


Returns covariance matrix.

Examples in Lua

Show how the peak center moves between datasets:

-- file list-max.lua
prev_x = nil
for n = 0, F:get_dataset_count()-1 do
    local path = F:get_info("filename", n)
    local filename = string.match(path, "[^/\\]+$") or ""
    -- local x = F:calculate_expr("argmax(y)", n)
    local x = F:calculate_expr("F[0].center", n)
    s = string.format("%s: max at x=%.4f", filename, x)
    if prev_x ~= nil then
        s = s .. string.format("  (%+.4f)", x-prev_x)
    prev_x = x
=-> exec list-max.lua
frame-000.dat: max at x=-0.0197
frame-001.dat: max at x=-0.0209  (-0.0012)
frame-002.dat: max at x=-0.0216  (-0.0007)
frame-003.dat: max at x=-0.0224  (-0.0008)

Write to file values of the model F(x) at chosen x’s (in this example x = 0, 1.5, 3, ... 150):

-- file tabular-f.lua
file ="output.dat", "w")
for x = 0, 150, 1.5 do
    file:write(string.format("%g\t%g\n", x, F:get_model_value(x)))
=-> exec tabular-f.lua
=-> !head -5 output.dat
0       12.1761
1.5     12.3004
3       10.9096
4.5     9.12635
6       8.27044