diff options
Diffstat (limited to 'host/include/uhd/utils/math.hpp')
| -rw-r--r-- | host/include/uhd/utils/math.hpp | 4 |
1 files changed, 2 insertions, 2 deletions
diff --git a/host/include/uhd/utils/math.hpp b/host/include/uhd/utils/math.hpp index 6c8fceae9..6ee46e98a 100644 --- a/host/include/uhd/utils/math.hpp +++ b/host/include/uhd/utils/math.hpp @@ -85,10 +85,10 @@ public: * There are obviously a lot of strategies for defining floating point * equality, and in the end it all comes down to the domain at hand. UHD's * floating-point-with-epsilon comparison algorithm is based on the method - * presented in Knuth's "The Art of Computer Science" called "very close + * presented in Knuth's "The Art of Computer Science" called "close enough * with tolerance epsilon". * - * [(|u - v| / |u|) <= e] && [(|u - v| / |v|) <= e] + * [(|u - v| / |u|) <= e] || [(|u - v| / |v|) <= e] * * UHD's modification to this algorithm is using the denominator's epsilon * value (since each float_t object has its own epsilon) for each |