tag:blogger.com,1999:blog-1798883676384076486.post1628121469036296903..comments2019-05-10T02:46:44.414-07:00Comments on Metalight: Smooth Minimum and MaximumUnknownnoreply@blogger.comBlogger3125tag:blogger.com,1999:blog-1798883676384076486.post-26369695215927442932017-10-23T02:27:30.054-07:002017-10-23T02:27:30.054-07:00In case x values where 0 the gradient will be divi...In case x values where 0 the gradient will be divided by 0. how this could be solved without adding constant value. Feras Almasrihttps://www.blogger.com/profile/07545017104504679863noreply@blogger.comtag:blogger.com,1999:blog-1798883676384076486.post-89105718058935150762016-12-08T00:47:00.496-08:002016-12-08T00:47:00.496-08:00Your Function is flawed consider b^0 = 1 therefore...Your Function is flawed consider b^0 = 1 therefore say you have f1(x) = 0, f2(x) = 0 and f3(x) = 0 your function reduced to (1^s+1^s+1^s)^(1/s) = 3^(1/s) != 1 therefor log_b(3^(1/s)) != 0, note this error is only significant near 0, i.e. as any single value -> inf the error -> 0,<br />if you set s=1, then it can be corrected by subtracting q-1 from the sumGlen Fletcherhttps://www.blogger.com/profile/04054892178328099575noreply@blogger.comtag:blogger.com,1999:blog-1798883676384076486.post-84373058341223419872016-12-08T00:37:37.996-08:002016-12-08T00:37:37.996-08:00This comment has been removed by the author.Glen Fletcherhttps://www.blogger.com/profile/04054892178328099575noreply@blogger.com