In the GRB system, we combine the three primary colors, red, green, and blue, to make some new colors.
It's easy to understand the production of yellow because the wavelength of yellow is between red and green. So as cyan and purple.
However, violet, which is notated as grb(127, 0, 255), how do you understand it?
Violet has the shortest wavelength, so how could you composite it by combining red(127) and blue(255)?
Colors are not wavelengths. Light composed of equal parts "red" (say 650 nm) and "green" light (say 530 nm) is exactly that: light containing both wavelengths 650 nm and 530 nm. There will not be any light of a "yellow" wavelength (around 570 nm). A "spectral yellow" light (composed of just one wavelength of light) is physically different from light composed by mixing two wavelengths. You can use a prism or a diffraction grating to split any light into its component wavelengths and then see the difference. Objects also generally reflect a red+green mixture differently from spectral yellow light: that's why RGB lights are good enough for screens (that people look at directly) but are not good for general purpose lighting. Objects will tend to be more vibrantly colored under a "real" white light (i.e. replicating the "full rainbow" we get from the Sun) than under white light constructed by mixing pure red, green, and blue lights, even if the colors of the lights themselves look identical to the eye. (The property of "showing colors properly" i.e. replicating the spectrum of the Sun is quantified by the Color Rendering Index, which you may find advertised on light bulbs.)
Your eye is not sophisticated enough to split light into its component wavelengths and tell you how the energy is distributed over the whole spectrum. The color you experience when looking at a certain light beam is only an extremely "summarized" version of the physical content of that light. Your eye can't tell red+green light apart from a spectral yellow, and it also can't tell blue+red light apart from a spectral violet. To properly describe the content of a light beam, you need one number (the intensity) for every possible wavelength (so infinitely many). Your eye only records three quantities: how well the light activates a certain red pigment, a certain green pigment, and a certain blue pigment. It is then logically necessary that lots of physically different light beams can produce the same sensations in an eye.
In more detail, the mechanism of how a red+blue mixture can look similar to light of a shorter wavelength than either component is that the pigment in your eye responsible for detecting red also has a weak sensitivity to much smaller "violet" wavelengths, overlapping with the blue pigment. Therefore, a spectral violet activates both the blue and red receptors of your eye. An RGB screen can replicate this by simply providing the eye with a mixture of blue and red.
As a closing note, I'll mention that most colors the eye can see do not correspond to any pure wavelength. Most colors can only be seen when there is a mixture of different wavelengths (examples: white, pink, brown, ...). This is particularly true of violets and purples: there is a small window towards the end of the visible light spectrum where light will look violet, but most of the purples a human can see must be made by mixing red and blue light and don't correspond to any single wavelength.
The RGB color system can't produce a violet color. The color sensing cells in your retina, the cone cells, can only see red, green, and blue. This means that a very particular combination of red, green, and blue intensities is interpreted by your brain as violet. As long as the RGB color system can replicate that combination of intensities, your brain will again interpret that combination as violet. For much more detail, see this.
RGB(0,0,255) doesn't mean it will simulate only blue cones. Instead it simulates mainly blue cones plus some green cones and negligible red cones. Something like (1%,10%, 100%). Some green cones also get simulated because the efficiency curves for the cone receptors taper off asymptotically to zero like a Gaussian curve.
So what does the violet light simulate? It simulates blue cones comparatively more and green cones comparatively less as compared to blue light. Something like (0,5%,100%).
How can we achieve the same effect with RGB? Add red to blue.
How will it work?
We add that amount of red to blue to create a mixture such that the red cones get simulated as much as green cones. In effect brain perceives violet color mixed with white light i.e. in a lighter shade. Something like (7%,11%,100%) = white(7%,7%,7%) + indigo(0,4%,93%)
