Hi Stephen (or current maintainer). Thanks for making this code publicly available!
I think there might be a mistake here:
|
cdf = quad(@(t)LBA_n1PDF(t,A,A+b,v,sv),1,100000); |
Actually maybe two mistakes. First, the integral should begin at 0 not 1 (I assume you have used ms in which case this wouldn't matter, but since others might use seconds, it's probably worth fixing unless there's some additional factor I am missing). Second, the PDF should be called with just A as an argument, not A+b. So I think the line should read:
cdf = quad(@(t)LBA_n1PDF(t,A,A,v,sv),0,100000);
I can't confirm this in matlab (I'm implementing it in Julia and don't have access to matlab) but I found that the current code gives slightly incorrect results when you compare LBA_n1CDF to the empirical choice frequency from simulations from LBA_trial. With the changes I suggested, the discrepancy vanishes.
Hi Stephen (or current maintainer). Thanks for making this code publicly available!
I think there might be a mistake here:
LBA/LBA_n1CDF.m
Line 9 in 0778969
Actually maybe two mistakes. First, the integral should begin at 0 not 1 (I assume you have used ms in which case this wouldn't matter, but since others might use seconds, it's probably worth fixing unless there's some additional factor I am missing). Second, the PDF should be called with just A as an argument, not A+b. So I think the line should read:
cdf = quad(@(t)LBA_n1PDF(t,A,A,v,sv),0,100000);
I can't confirm this in matlab (I'm implementing it in Julia and don't have access to matlab) but I found that the current code gives slightly incorrect results when you compare
LBA_n1CDFto the empirical choice frequency from simulations fromLBA_trial. With the changes I suggested, the discrepancy vanishes.