Provides access to linearly interpolated values (and derivatives) from discrete data points. Values beyond the given range of data points are implicitly taken as zero. More...

#include <OpenMS/MATH/MISC/LinearInterpolation.h>

Public Types
Typedefs

typedef Value	value_type
typedef Key	key_type
typedef std::vector< value_type >	container_type
typedef value_type	ValueType
typedef key_type	KeyType
typedef container_type	ContainerType
Public Member Functions
	LinearInterpolation (KeyType scale=1., KeyType offset=0.)
	Constructors and destructor.
	LinearInterpolation (LinearInterpolation const &arg)
	Copy constructor.
LinearInterpolation &	operator= (LinearInterpolation const &arg)
	Assignment operator.
	~LinearInterpolation ()
	Destructor.
Interpolated data

ValueType	value (KeyType arg_pos) const
	Returns the interpolated value.
void	addValue (KeyType arg_pos, ValueType arg_value)
	Performs linear resampling. The arg_value is split up and added to the data points around arg_pos.
ValueType	derivative (KeyType arg_pos) const
	Returns the interpolated derivative.
Discrete (non-interpolated) data

ContainerType &	getData ()
	Returns the internal random access container from which interpolated values are being sampled.
ContainerType const &	getData () const
	Returns the internal random access container from which interpolated values are being sampled.
template<typename SourceContainer >
void	setData (SourceContainer const &data)
	Assigns data to the internal random access container from which interpolated values are being sampled.
bool	empty () const
	Returns `true` if getData() is empty.
Transformation

KeyType	key2index (KeyType pos) const
	The transformation from "outside" to "inside" coordinates.
KeyType	index2key (KeyType pos) const
	The transformation from "inside" to "outside" coordinates.
KeyType const &	getScale () const
	Accessor. "Scale" is the difference (in "outside" units) between consecutive entries in "Data".
void	setScale (KeyType const &scale)
	Accessor. "Scale" is the difference (in "outside" units) between consecutive entries in "Data".
KeyType const &	getOffset () const
	Accessor. "Offset" is the point (in "outside" units) which corresponds to "Data[0]".
void	setOffset (KeyType const &offset)
	Accessor. "Offset" is the point (in "outside" units) which corresponds to "Data[0]".
void	setMapping (KeyType const &scale, KeyType const &inside, KeyType const &outside)
	Specifies the mapping from "outside" to "inside" coordinates by the following data: `scale`: the difference in outside coordinates between consecutive values in the data vector. `inside` and `outside`: these x-axis positions are mapped onto each other.
void	setMapping (KeyType const &inside_low, KeyType const &outside_low, KeyType const &inside_high, KeyType const &outside_high)
	Specifies the mapping from "outside" to "inside" coordinates by the following data: `inside_low` and `outside_low`: these axis positions are mapped onto each other. `inside_high` and `outside_high`: these axis positions are mapped onto each other.
KeyType const &	getInsideReferencePoint () const
	Accessor. See setMapping().
KeyType const &	getOutsideReferencePoint () const
	Accessor. See setMapping().
KeyType	supportMin () const
	Lower boundary of the support, in "outside" coordinates.
KeyType	supportMax () const
	Upper boundary of the support, in "outside" coordinates.
Protected Attributes
KeyType	scale_
KeyType	offset_
KeyType	inside_
KeyType	outside_
ContainerType	data_

Detailed Description

template<typename Key = DoubleReal, typename Value = Key>
class OpenMS::Math::LinearInterpolation< Key, Value >

Provides access to linearly interpolated values (and derivatives) from discrete data points. Values beyond the given range of data points are implicitly taken as zero.

The input is just a vector of values ("Data"). These are interpreted as the y-coordinates at the x-coordinate positions 0,...,data_.size-1.

The interpolated data can also be scaled and shifted in the x-dimension by an affine mapping. That is, we have "inside" and "outside" x-coordinates. The affine mapping can be specified in two ways:

using setScale() and setOffset(),
using setMapping()

By default the identity mapping (scale=1, offset=0) is used.

Using the value() and derivative() methods you can sample linearly interpolated values for a given x-coordinate position of the data and the derivative of the data.

See also:: BilinearInterpolation

Member Typedef Documentation

typedef std::vector<value_type> container_type

typedef container_type ContainerType

typedef Key key_type

typedef key_type KeyType

typedef Value value_type

typedef value_type ValueType

Constructor & Destructor Documentation

LinearInterpolation	(	KeyType	scale = `1.`,
		KeyType	offset = `0.`
	)			`[inline]`

Constructors and destructor.

The first argument is the scale which is applied to the arguments of value() and derivative() before looking up the interpolated values in the container. The second argument is the offset, which is subtracted before everything else.

LinearInterpolation ( LinearInterpolation< Key, Value > const & arg ) [inline]

Copy constructor.

~LinearInterpolation ( ) [inline]

Destructor.

Member Function Documentation

void addValue	(	KeyType	arg_pos,
		ValueType	arg_value
	)			`[inline]`

Performs linear resampling. The arg_value is split up and added to the data points around arg_pos.

ValueType derivative ( KeyType arg_pos ) const [inline]

Returns the interpolated derivative.

Please drop me (= the maintainer) a message if you are using this.

bool empty ( ) const [inline]

Returns true if getData() is empty.

Referenced by LinearInterpolation< DoubleReal >::supportMin().

ContainerType& getData ( ) [inline]

Returns the internal random access container from which interpolated values are being sampled.

ContainerType const& getData ( ) const [inline]

Returns the internal random access container from which interpolated values are being sampled.

KeyType const& getInsideReferencePoint ( ) const [inline]

Accessor. See setMapping().

KeyType const& getOffset ( ) const [inline]

Accessor. "Offset" is the point (in "outside" units) which corresponds to "Data[0]".

Referenced by MaxLikeliFitter1D::fitOffset_().

KeyType const& getOutsideReferencePoint ( ) const [inline]

Accessor. See setMapping().

KeyType const& getScale ( ) const [inline]

Accessor. "Scale" is the difference (in "outside" units) between consecutive entries in "Data".

KeyType index2key ( KeyType pos ) const [inline]

The transformation from "inside" to "outside" coordinates.

Referenced by LinearInterpolation< DoubleReal >::supportMax(), and LinearInterpolation< DoubleReal >::supportMin().

KeyType key2index ( KeyType pos ) const [inline]

The transformation from "outside" to "inside" coordinates.

Referenced by LinearInterpolation< DoubleReal >::addValue(), LinearInterpolation< DoubleReal >::derivative(), and LinearInterpolation< DoubleReal >::value().

LinearInterpolation& operator= ( LinearInterpolation< Key, Value > const & arg ) [inline]

Assignment operator.

void setData ( SourceContainer const & data ) [inline]

Assigns data to the internal random access container from which interpolated values are being sampled.

SourceContainer must be assignable to ContainerType.

void setMapping	(	KeyType const &	inside_low,
		KeyType const &	outside_low,
		KeyType const &	inside_high,
		KeyType const &	outside_high
	)			`[inline]`

Specifies the mapping from "outside" to "inside" coordinates by the following data:

inside_low and outside_low: these axis positions are mapped onto each other.
inside_high and outside_high: these axis positions are mapped onto each other.

This four argument version is just a convenience overload for the three argument version, which see.

void setMapping	(	KeyType const &	scale,
		KeyType const &	inside,
		KeyType const &	outside
	)			`[inline]`

Specifies the mapping from "outside" to "inside" coordinates by the following data:

scale: the difference in outside coordinates between consecutive values in the data vector.
inside and outside: these x-axis positions are mapped onto each other.

For example, when you have a complicated probability distribution which is in fact centered around zero (but you cannot have negative indices in the data vector), then you can arrange things such that inside is the mean of the pre-computed, shifted density values of that distribution and outside is the centroid position of, say, a peak in the real world which you want to model by a scaled and shifted version of the probability distribution.

Referenced by LinearInterpolation< DoubleReal >::setMapping().

void setOffset ( KeyType const & offset ) [inline]

Accessor. "Offset" is the point (in "outside" units) which corresponds to "Data[0]".

Note: Using this invalidates the inside and outside reference points.

void setScale ( KeyType const & scale ) [inline]

Accessor. "Scale" is the difference (in "outside" units) between consecutive entries in "Data".

Note: Using this invalidates the inside and outside reference points.

KeyType supportMax ( ) const [inline]

Upper boundary of the support, in "outside" coordinates.

KeyType supportMin ( ) const [inline]

Lower boundary of the support, in "outside" coordinates.

Referenced by MaxLikeliFitter1D::fitOffset_().