I'm getting an unexpected result of big.Rat.Quo(). In trying to track it down, I ran into an unexpected result from big.Rat.SetFloat64(). So I have a few questions...
Code snippet below reproduces what I see in my application, and is also in https://play.golang.org/p/Eph67_1yd- I'm working with data that includes pairs of decimal numbers, and I need the ratio between each pair. Originally, I started implementing with float64s. But I ran into some rounding errors and something I read somewhere (maybe this list) led me to think that big.Rat would be a better choice. Using bigRat solved my problem at the time, but now in processing a particular list of numbers I'm noticing an error. The result I would like is for input two values, 118.5285714285714 and 8297.0, is an output ratio should be 0.01428571428571428. I get this on a calculator, but not code using bit.Rat. Using big.Rat, getting the 8297 denominator is straightforward. Getting the numerator is tricky. Note these two approaches get different results: numerator.SetFloat64(118.5285714285714) // becomes 118.528571428571396495499357115477 big.NewRat(592642857142857, 5000000000000) // becomes 118.528571428571400000000000000000 - more accurate than SetFloat64 above! So my first question is why is SetFloat64() less accurate than the ratio used in NewRat() above? Next question is why do neither of these numerators get the expected result from big.Rat.Quo()? See output of code below (shown in comments). And given these problems, should I be using some other package to perform this math? Or, am I making some mistake in how I use big.Rat? Whole code snippet: package main import ( "fmt" "math/big" ) func main() { var numerator, denominator, result big.Rat // These are two ways to get not quite the same big.Rat value: numerator.SetFloat64(118.5285714285714) // becomes 118.528571428571396495499357115477 numerator2 := big.NewRat(592642857142857, 5000000000000) // becomes 118.528571428571400000000000000000 - more accurate than SetFloat64 above! denominator.SetInt64(8297) // remains 8297.0000000... for _, num := range []*big.Rat{&numerator, numerator2} { result.Quo(num, &denominator) // Expected result: 0.01428571428571428 // Show results as fractions. fmt.Printf("%s over %s is %s\n", num.String(), denominator.String(), result.String()) // Show a lot of decimal places. fmt.Printf("%s over %s is %s\n\n", num.FloatString(30), denominator.FloatString(30), result.FloatString(30)) } // Output: // 8340706720601115/70368744177664 over 8297/1 is 8340706720601115/583849470442078208 // 118.528571428571396495499357115477 over 8297.000000000000000000000000000000 is 0.014285714285714281848318591915 // 592642857142857/5000000000000 over 8297/1 is 592642857142857/41485000000000000 // 118.528571428571400000000000000000 over 8297.000000000000000000000000000000 is 0.014285714285714282270700253104 } -- You received this message because you are subscribed to the Google Groups "golang-nuts" group. To unsubscribe from this group and stop receiving emails from it, send an email to golang-nuts+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
