A question about a variable in the algorithm described in the paper

[hongruyang]

Data:
    Input variables : inVars, Input specifications : inSpecs,      
    Program :P, Program with injected inaccuracy checks :P’, 
    Time budget :tBudget, Timeout parameters : τ0, τ1
Result:
    Maximum error, the error-inducing input

tStart = time();
/* Random search for an error-inducing input. */
randomErrMax = 0;
inBase = NULL;    
rStart = time();
while time() - rStart < τ0 do   
    generate a random input in the input domain : in;      
    err = compute - error(in, P);      
    if err > eMax then        
        randomErrMax = err;  
        inBase = in;      
    end 
end

/* Separate the input variables into operands. */
op1Vars,op2Vars = partition - variables(inVars, inSpecs);

/* Initilization. */
errMax = randomErrMax;
errInput = inBase;
concVars = op1Vars;    
symbVars = op2Vars;     
while time() - tStart < tBudget do 
    /* Concretize convVars using base values. */
    concretize the variables in convVars using inBase;
    symbolize the variables in symbVars;
    STAT, sInputs = symbolic - execution(convVars, symbVars, P’, τ1);     
    if STAT = timeout then 
        sNum = length(symbVars);
        s1Vars, s2Vars = random - divide(symbVars, sNum / 2, sNum / 2);   
        convVars = convVars + s1Vars;       
        symbVars = s2Vars;
    else
        for sInput in sInputs do
            err = compute-error(join-input(sInput, inBase), P);
            if err > errMax then 
                errMax = err;
                errInput ← join - input(sInput, inBase);
            end 
        end 
        sNum = length(symbVars);
        lop2 = length(op2Vars);
        s1Vars, s2Vars = random - divide(op2Vars, lop2 - sNum, sNum);
        convVars = op1Vars + s1Vars;
        symbVars = s2Vars;
     end
end
return errMax,errInput

Hello, I'm reading your paper recently.

I don't understand the variable InBase and sInput in the algorithm mentioned in your paper.

Taking x+y as an example, does InBase generate the values of x and y, or x or y.

My understanding is that the value in x or y is taken as the specific variable.

In addition, could you please explain the algorithm more?I just entered this field, and I don't know many things.Thanks a lot.