Back to the main page.

Bug 292 - ft_freqanalysis does not work with variable length trials and method = 'mtmfft'

Status	CLOSED FIXED
Reported	2010-12-10 11:55:00 +0100
Modified	2011-01-14 09:55:30 +0100
Product:	FieldTrip
Component:	core
Version:	unspecified
Hardware:	PC
Operating System:	Mac OS
Importance:	P1 major
Assigned to:	Roemer van der Meij
URL:
Tags:
Depends on:
Blocks:
See also:

---

Jan-Mathijs Schoffelen - 2010-12-10 11:55:37 +0100

this is because the cumtapcnt field is incorrectly created. This is an important bug because subsequent code relies on this field

---

Jan-Mathijs Schoffelen - 2010-12-10 12:06:07 +0100

I used cfg.output = 'fourier', in combination with cfg.method = 'mtmfft'; I have variable length trials

---

Roemer van der Meij - 2010-12-13 17:02:30 +0100

Hi JM, I thought I had it correctly implemented based on the old code, but I must have missed something. What are supposed to be the correct dimensions of the field? Then I'll re-implement it in a proper way.

---

Roemer van der Meij - 2010-12-14 16:16:01 +0100

It should now produce a correct cumtapcnt. For the moment, it's a split solution for mtmconvol (ntrialsXnfoi) and mtmfft (ntrialsX1). Cumtapcnt is not created for the other methods. Perhaps it would be best to think of a solution/representation that is equal for all of the freq methods (which can only be done if the other functions are 'specested').