GBNLPANEL(1)

💖 GBNLPANEL — non-linear panel regression 💖

okay bestie, so this one is a User Commands page for gbnlpanel, and it's giving serious stats-girl energy. Sit down, grab a boba, we're doing panel regression. 🫶

---

✨ SYNOPSIS ✨

gbnlpanel [options] <FUN>

---

🌸 DESCRIPTION 🌸

okay so picture this: you've got blocks, and every row inside a block is a single process realization, fr fr. Blocks are separated by two white spaces and each block = one model variable. Columns? Columns are the time variable, periodt. 🕰️

Input gets read in as (X1_[r,t],...,Xj_[r,t],... XN_[r,t]), where r is the row, t is the column, and j is the block. Keep up bestie, we're going somewhere.

The model assumes that

FUN(X1_[r,t],...,XN_[t,t]) - c_r = e_{r,t}

with e i.i.d and c a row specific element 💅 — which can be either fixed (that's your fixed effect) or a normal random variable (that's your random effect). Iconic behaviour honestly.

---

👑 OPTIONS 👑

• -M — type of model (default 0)
  • 0 — fixed effects
  • 1 — random effects
• -O — type of output (default 0)
  • 0 — parameters
  • 1 — parameters and errors
  • 2 — <variables> and panel statistics
  • 3 — parameters and variance matrix
• -V — variance matrix estimation (default 0)
  • 0 — < J^{-1} >, computed via fully-reduced log-likelihood
  • 1 — < H^{-1} >, computed via fully-reduced log-likelihood
  • 2 — < H^{-1} J H^{-1} >, computed via fully-reduced log-likelihood
  • 3 — < J^{-1} >, computed via non-reduced log-likelihood
  • 4 — < H^{-1} >, computed via non-reduced log-likelihood
  • 5 — < H^{-1} J H^{-1} >, computed via non-reduced log-likelihood
• -v — verbosity level (default 0)
  • 0 — just results
  • 1 — comment headers
  • 2 — summary statistics
  • 3 — covariance matrix
  • 4 — minimization steps
  • 5 — model definition
• -e — minimization tolerance (default 1e-6)
• -F — input fields separators (default " \t")
• -h — this help
• -A — comma separated MLL optimization options: step,tol,iter,eps,msize,algo. Use empty fields for default. (default 0.1,0.01,500,1e-6,1e-6,0)
  • step — initial step size of the searching algorithm
  • tol — line search tolerance iter: maximum number of iterations
  • eps — gradient tolerance : stopping criteria ||gradient||<eps
  • algo — optimization methods: 0 Fletcher-Reeves, 1 Polak-Ribiere, 2 Broyden-Fletcher-Goldfarb-Shanno, 3 Steepest descent (not recommended, lowkey don't 🙈), 4 simplex, 5 Broyden-Fletcher-Goldfarb-Shanno-2

bestie pls note: the -A list is step,tol,iter,eps,msize,algo and the descriptions cover step, tol, iter, eps and algo — I'm not gonna make up a definition for msize, so it stays mysterious ✨

---

🦋 EXAMPLES 🦋

gbnlpanel 'x1-a*x2-c,a=0,c=0' < data.dat

linear panel regression with one independent variable 🎀

---

🌈 AUTHOR 🌈

Written by Giulio Bottazzi 👑

---

💌 REPORTING BUGS 💌

Report bugs to <gbutils@googlegroups.com> — they wanna hear it, no cap 😭

Package home page <http://cafim.sssup.it/~giulio/software/gbutils/index.html>

---

🎀 COPYRIGHT 🎀

Copyright © 2001-2018 Giulio Bottazzi

This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License (version 2) as published by the Free Software Foundation;

This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.

fr fr, that warranty disclaimer is the whole vibe — no promises, just math 💀

---

and that's the whole page bestie!! this is gbnlpanel version 6.2, dated June 2021, slay 💅✨🌸🦋