× ⚠️ You are reading the development version documentation.
Read here the documentation of the last stable release.

NDDataset creation and plotting example

In this example, we create a 3D NDDataset from scratch, and then we plot one section (a 2D plane)

Creation

Now we will create a 3D NDDataset from scratch

import numpy as np

As usual, we start by loading the spectrochempy library

import spectrochempy as scp

Coordinates

The Coord object allows creating an array of coordinates directly with Coord.linspace, attaching metadata (units, labels, title) in a single step — no separate numpy array needed.

coord0 = scp.Coord.linspace(
    200.0,
    300.0,
    3,
    labels=["cold", "normal", "hot"],
    units="K",
    title="temperature",
)
coord1 = scp.Coord.linspace(0.0, 60.0, 100, units="minutes", title="time-on-stream")
coord2 = scp.Coord.linspace(4000.0, 1000.0, 100, units="cm^-1", title="wavenumber")

Labels can be useful for instance for indexing

a = coord0["normal"]
print(a)

Coord: [float64] K (size: 1)

Data and nd-Dataset

scp.fromfunction builds an NDDataset directly from a Python function. The function receives the coordinate arrays and returns the intensity values.

def synth_func(temperature, time, wavenumber):
    return np.sin(2.0 * np.pi * wavenumber / 4000.0) * np.exp(-time / 60) * temperature


mydataset = scp.fromfunction(
    synth_func,
    coordset=[coord0, coord1, coord2],
    title="Absorbance",
    units="absorbance",
)
mydataset.description = """Dataset example created for this tutorial.
It's a 3-D dataset (with dimensionless intensity: absorbance )"""
mydataset.name = "An example from scratch"
mydataset.author = "Blake and Mortimer"
print(mydataset)

NDDataset: [float64] a.u. (shape: (z:3, y:100, x:100))

In a Jupyter notebook, the NDDataset is displayed as follows (click on the arrow on the left to expand the metadata):

mydataset

NDDataset [An example from scratch] — float64, shape: (z:3, y:100, x:100), a.u.

name

:

An example from scratch

author

:

Blake and Mortimer

created

:

2026-07-10 21:24:48+00:00

description

:

Dataset example created for this tutorial.
It's a 3-D dataset (with dimensionless intensity: absorbance )

history

:

2026-07-10 21:24:48+00:00> Created using method : fromfunction

Data

title

:

Absorbance

values

:

[[[ 1236 1227 ... 318.4 309]
[ 1224 1214 ... 315.2 305.9]
...
[ 459.3 455.9 ... 118.3 114.8]
[ 454.7 451.3 ... 117.1 113.7]]

[[ 1531 1519 ... 394.3 382.7]
[ 1515 1504 ... 390.3 378.8]
...
[ 568.8 564.5 ... 146.5 142.2]
[ 563.1 558.9 ... 145 140.8]]

[[ 1816 1802 ... 467.7 454]
[ 1798 1784 ... 463 449.4]
...
[ 674.8 669.7 ... 173.8 168.7]
[ 668.1 663 ... 172.1 167]]] a.u.

shape

:

(z:3, y:100, x:100)

Dimension `x`

size

:

100

title

:

wavenumber

coordinates

:

[ 4000 3970 ... 1030 1000] cm⁻¹

Dimension `y`

size

:

100

title

:

time-on-stream

coordinates

:

[ 0 0.606 ... 59.39 60] min

Dimension `z`

size

:

3

title

:

temperature

coordinates

:

[ 200 250 300] K

labels

:

[ cold normal hot]

We want to plot a section of this 3D NDDataset:

NDDataset can be sliced like conventional numpy-array…

new = mydataset[..., 0]

or maybe more conveniently in this case, using an axis labels:

new = mydataset["hot"]

To plot a dataset, use the plot method (generic plot). As the section NDDataset is 2D, a lines plot is displayed by default. As you can see, the x-axis is in wavenumber and the ordinate axis is in absorbance. Note that in this case, the default NDDataset.plot() command is equivalent to plot_lines().

_ = new.plot()

Note also that a colormap (‘viridis’) has been automatically set for the lines. This is because the y-dimension of the dataset has float coordinates (they correspond to a time). In such a case it is easy to add a colorbar explaining the colors <-> time value correspondence:

_ = new.plot(colorbar=True)

If the y-dimension had no coordinates or consecutive integer coordinates starting by 0`or `1, a categorical color map would have been chosen. The default behavior can be overridden by explicitly passing a colormap. For instance, if we want to use a categorical colormap instead of a sequential one, we can do:

_ = new[:, 0:20].plot(cmap="tab20")

But it is possible to display image plot instead (note that the x-axis is in wavenumber and the y-axis is in time-on-stream)

_ = new.plot_image()

or contour plot (note that

_ = new.plot_map()

This ends the example ! The following line can be uncommented if no plot shows when running the .py script with python

# scp.show()