Transform str expression into tree

[DenisSvirin]

Is there any function, that allows to convert expression in a string form (ex:(x+3) ) into a tree form, suitable for hall of fame?

or May be there is another way to insert some function into a hall of fame ?

[MilesCranmer]

Sorry, missed this one!

Yes:

from pysr import jl  # Julia Main namespace
from pysr import PySRRegressor
from pysr import SymbolicRegression as SR  # Backend library

# Create `.julia_options_`:
model = PySRRegressor(verbosity=0, max_evals=1)
model.fit([[1]], [1])
options = model.julia_options_

# Define operators for creating trees:
jl.options = options  # Store `options` in Julia namespace
jl.seval("@extend_operators options")  # Declare available operators

x1 = SR.Node(jl.Float32, feature=1) # If x is the first column (index 0 in Python)
# Store x1 in Julia namespace:
jl.x1 = x1
# Create equation from string:
eqn = jl.seval("x1 + 1")

The + operator here gets mapped into an integer based on the operators defined in options. You declare this mapping when you call @extend_operators options.

To verify this is the correct equation, you can look at:

SR.string_tree(eqn, options)

which you can see is x1 + 1.0.

To get the PopMember, you can create it as follows:

score = ...
loss = ...
SR.PopMember(eqn, jl.Float32(score), jl.Float32(loss), options)

which you can put in the hall of fame.