Code covered by the BSD License

### Highlights from Free-knot spline approximation

4.5

4.5 | 2 ratings Rate this file 30 Downloads (last 30 days) File Size: 102 KB File ID: #25872

# Free-knot spline approximation

17 Nov 2009 (Updated 06 Jul 2012)

Least squares approximation of 1D data using free-knots spline

File Information
Description

The purpose of this function is to provide a flexible and robust fit to one-dimensional data using free-knot splines. The knots are free and able to cope with rapid change in the underlying model. Knot removal strategy is used to fit with only a small number of knots.

Optional L2-regularization on the derivative of the spline function can be used to enforce the smoothness.

Shape preserving approximation can be enforced by specifying the lower and upper bounds of the derivative(s) of the spline function on sub-intervals. Furthermore specific values of the spline function and its derivative can be specified on a set of discrete data points.

I did not test QUADPROG engine, but I have implemented it. Any feedback is welcome.

Acknowledgements

Pseudo Inverse, Multiple Same Size Linear Solver, and Min/Max Filter inspired this file.

MATLAB release MATLAB 7.9 (R2009b)
Other requirements optimization toolbox or QP solvers available at: http://sigpromu.org/quadprog/index.html (QPC) http://www.mat.univie.ac.at/~neum/software/minq/ QPC solver is strongly recommended.
05 Dec 2013

Hi,

I´m always getting the following warning message:

Warning: Options LargeScale = 'off' and Algorithm = 'trust-region-reflective'
conflict. Ignoring Algorithm and running active-set algorithm. To run
trust-region-reflective, set LargeScale = 'on'. To run active-set without this
warning, set Algorithm = 'active-set'.

What am I doing wrong?

05 Jul 2012

Richard,

not exactly like you want but you can enforce the y-value between two knots to be zero:

x=linspace(0,2*pi,100);
y = sin(x);
y = y + 0.1*randn(size(y));

nknots = 5;
lo = -inf(1,nknots);
up = +inf(1,nknots);
lo(3) = 0;
up(3) = 0;
shape = struct('p',0,'lo',lo,'up',up);
options = struct('shape', shape,'animation', 1, 'knotremoval','none');
BSFK(x,y,4,nknots,[],options);

03 Jul 2012

Brilliant package Bruno. Quick question.

I know you can fix any given knot, but is it possible to fix a given not to a y-value but let the least squares find the best x-value for it?

I'm using 4 knots/3 lines to represent my data but I would always like the 3rd knot to have y=0. While this condition is met some of the time (by chance) ideally I would like to enforce it.

26 Jan 2012

Nima, it is not rotational invariant. Because the fit is carried out using the least-squares to the ordinate data (y) only.

23 Jan 2012

Hey Bruno thanks for the awesome package, it works great!
I just have one problem, is this approximation rotation invariant? Since when I intentionally rotate my data points the knots are completely different with increased fit error compared to the baseline profile! please let me know if I am doing sth wrong

23 Jun 2011

Hi Bruno,

Nice package, many great features!

I am having a problem with the 'startingknots' parameter.

When I type this:

test=bsfk(x,y,2,[],[],struct('display',1,'animation',1,'startingknots',polyX))

I get this:

BSFK starts
??? Error using ==> sparse
Sparse matrix sizes must be non-negative integers less than MAXSIZE as defined by
COMPUTER. Use HELP COMPUTER for more details.

Error in ==> BSFK>BuildDineqMat at 1623
D = sparse(row,col,val,m,n);

Error in ==> BSFK>UpdateConstraints at 2039
[D LU X] = BuildDineqMat(t, knotidx, k, shape);

Error in ==> BSFK>InitPenalization at 2077
smoothing = UpdateConstraints(smoothing, t, shape, pntcon, periodic);

Error in ==> BSFK at 393
smoothing = InitPenalization(y, t, k, d, lambda, p, regmethod, ...

However, if I use 'chebyschev' instead of a vector of starting knots, it works. But it misses some important knots.

Any ideas?

Thanks,

Doug

03 Oct 2010

Michael, I'll reply in the appropriate place (minmaxfilt)

30 Sep 2010

Does it handle NaN data?

ePeriod = 3;
minmaxfilt(eData, ePeriod, 'max', 'valid')];

Actual output:
[5;3;3;NaN;8;3;3]

If we take NaN as a empty data, the expected output is:
[5;3;8;8;8;3;3]

08 Jun 2010

Just discover an issue with continuous regularization. In the mean time, please use the discrete regularization

01 Jun 2010

Periodic spline is now available

09 Jan 2010
18 Nov 2009

Update description, more options added to control the fit, discrete regularization

19 Nov 2009

Remove NaN data before fitting, change TRY/CATCH ME syntax for better compatibility (tested under 2006B), estimate automatic of the noise standard deviation

28 Nov 2009

Change title and description

04 Dec 2009

A major enhancement with shape preserving splines

08 Dec 2009

Point-wise constraints. Discover an error of the Jacobian formula in [Schutze/Schwetlick 97] paper, modify the calculation accordingly. This concern only the constrained fitting.

09 Dec 2009

Correct another bug in the Jacobian calculation (constrained case)

09 Dec 2009

Change the description.

10 Dec 2009

Singular constraints will issue a warning (instead of an error). Refine the Gauss-Newton direction. Fix few minor bugs.

14 Dec 2009

Correct a bug in UpdateConstraints that did not update the knot positions. Precasting data to double. Update more frequently the scaling matrix. Reduce the Lagrange's tolerance to detect active set of QPC solver

18 Mar 2010

fixed small bug when calling QP engine minqdef

10 Apr 2010

fix a bug with parsing k and nknots
Spline order can be as low as k=1 (piecewise constant fit)

31 May 2010

New feature: Periodic spline

01 Jun 2010

Remove some redundant code, modify test program

07 Jun 2010

A more robust conversion in pp form is implemented

07 Jun 2010

Fix a small bug (eigs with 'sa' option requires true symmetric matrix, which is now always the case by symmetrizing)

10 Jun 2010

Fix the bug for continuous regularization

05 Jul 2011

Fix the bug when starting knots are provided

06 Jul 2012

Fix a bug when checking for knot collision