# Efficient division for two probilities
**URL:** https://discourse.julialang.org/t/efficient-division-for-two-probilities/96579
**Category:** New to Julia
**Tags:** question
**Created:** [March 24, 2023, 9:34pm UTC](https://discourse.julialang.org/t/efficient-division-for-two-probilities/96579 "2023-03-24T21:34:59Z")
**Posts on this page:** 4
**Page:** 1
### Author:  [@X\_Chen](https://discourse.julialang.org/u/X_Chen)
#### Post date: [March 24, 2023, 9:34pm UTC](https://discourse.julialang.org/t/efficient-division-for-two-probilities/96579/1 "2023-03-24T21:34:59Z")
Hi everyone,
I have a very large dataframe (1 million observations) that I need to do calculation for two probabilities (p1 and p2): log(p1/(1-p2)).
And my code is:
log.(p1)/ (-p2.+1)
I think my code is so inefficient that it will run out of memory. So I am wondering if there is any way to get rid of the dot operator?
Thanks a lot!
---
### Author:  [@nilshg](https://discourse.julialang.org/u/nilshg)
#### Post date: [March 24, 2023, 9:50pm UTC](https://discourse.julialang.org/t/efficient-division-for-two-probilities/96579/2 "2023-03-24T21:50:13Z")
1 million observations shouldn’t be large for a reasonably new laptop. It seems that you are missing a dot and your parens are off, shouldn’t it be `log.(p1./ (1 .- p2)) `?
---
### Author:  [@blackeneth](https://discourse.julialang.org/u/blackeneth)
#### Post date: [March 24, 2023, 10:27pm UTC](https://discourse.julialang.org/t/efficient-division-for-two-probilities/96579/3 "2023-03-24T22:27:43Z")
If you p1 can be zero, or p2 can be 1, and you don’t want +Inf and -Inf in the results, one trick is to add 0.000001 to the top and bottom of the ration. Then the max and min values will be +13.8 and -13.8.
One million rows will be no problem for this calculation. Begin to worry between 100 - 300 million rows.
---
### Author:  [@X\_Chen](https://discourse.julialang.org/u/X_Chen)
#### Post date: [March 24, 2023, 11:04pm UTC](https://discourse.julialang.org/t/efficient-division-for-two-probilities/96579/5 "2023-03-24T23:04:32Z")
Thank you so much! That solves the problem.