Theses messages apply only to specific point(s) for which a y-value cannot be generated. They will not appear unless you ask for them using the Warnings option and then they appear silently.

[Please note that all warning messages which refer to the variables 'x' or 'y' will actually be 't' or 'r' when you are dealing with a polar equation.]

"Overflow at x=#.##."

Some function or operation generated a number too large to fit into a 8-byte floating point variable. The point at x=#.## was not graphed. This error is not fatal, so the graphing process is continued, but if the message is repeated and no image is graphed, you may need to abort graphing and look at your equation again.

"Division by zero at x=#.##."

At x=#.## your equation included division by zero so that point was skipped. Unless you get this error repeatedly, there's no real problem.

"Can't raise a negative number to a fractional power. [x=#.##]"

Due to the possibility of getting an even root of a negative number (like -16^(1/2) which actually equals the square root of -16), the C Library pow() function refuses to process any arguments like these. This is not a fatal error, and the portion of your graph (if any) where the base is not negative or the power is not fractional should be graphed perfectly. This error also occurs when you try to take the square root of a negative number with the "sqr" function.

"Can't find the logarithm of a negative number. [x=#.##]"

The natural logarithm (ln) and base 10 logarithm (log) functions are defined only on x greater than zero.

"Inverse of abs() not defined on negative numbers. [x=#.##]"

This reminder warning occurs when you graph a curve like abs(y)=x without restricting the domain of the expression equal to abs(y) to be positive.

"Domain error: asin/acos are defined only on -1<=x<=1, acsc/asec on x < -1 or x > 1. [x=#.##]"

The arcsine (asin) and arc cosine (acos) functions are only defined between -1 and 1 (the range of the sin and cos functions). The asec and acsc functions are only defined outside this range.

"Lost accuracy in approximation before reaching value. Solution might not exist."

When processing ODE or Newton's method approximations, Graphmatica sets a limit on the number of iterations to avoid going into an infinite loop in a poorly-behaved (e.g. discontinuous) part of the curve. When this happens you will see this message.

"Curve-fit did not converge within #### iterations; these results might not be accurate. You might try increasing the maximum number of iterations or decreasing the order of the equation."

The curve-fitting process stops automatically when Graphmatica detects that the chi^2 value (variance) of the data set from the fitted curve has stopped decreasing significantly from one iteration to the next. If the maximum number of iterations is reached before the chi^2 value stabilizes (which can happen with some data sets, especially if you are fitting to a polynomial and the maximum order is set rather high), this warning message is displayed to alert you to the fact that there might be a better solution available if you are willing to increase the maximum number of iterations and wait a bit longer for it to be computed. The fitted curve will still be displayed, though.

See also Fatal Error Messages

Help contents

Back to kSoft homepage...

kSoft, Inc. ksoft@graphmatica.com Last updated: Sun 11 Jun 2017