I got this circuit from internet. I don't know why the capacitors are installed in this circuit. Can somebody tell me the reason of these capacitors..
suha says stabilizing the voltage, but C3 is actually for stabilizing the regulator's control loop. It's the control loop which causes a stable output voltage, not the capacitor. Most regulators, especially LDOs will need C3 to prevent oscillations. ESR (Equivalent Series Resistance) is crucial.
The graph from this document shows that for the given regulator a capacitor with an ESR of 1 Ω is needed; the document shows how oscillation occurs with a too low ESR capacitor at a 150 mA load.
The regulator's control loop makes that it has a certain response time, so a sudden load change may cause a short dip in the output voltage before the regulator reacts. C2 acts as a buffer to catch those quick changes.
A 100 \$\mu\$F C1 smooths the input voltage, which improves output ripple, but in general isn't required for the regulator's operation, though you will need a 1 \$\mu\$F capacitor again for stability, especially for LDOs.
Steven has explained the purpose of C3, but this circuit is missing the equivalent on the input side. The problem is that C1 and C2 are both large caps that probably have poor high speed response and some ESR (Equivalent Series Resistance). That is fine for bulk storage, but not so fine for providing a large sudden surge of current. Note that "sudden" in the time domain is the same as "high frequency" in the frequency domain.
Perhaps the 78L05 is stable with a high ESR input cap, but that is usually not a good idea. Most datasheets advise you to put a low ESR cap physically close to both the input and the output of regulators. Ceramic caps meet the criteria well, but don't come in the large sizes that electrolytic caps do. This is why you sometimes see a large polarized cap in parallel with a much smaller one, as with C2 and C3 in this circuit.
Nowadays, 100 nF is silly for the low ESR cap of something like a 78L05. Long ago, that was about the largest ceramic cap you could get without paying a lot more. Nowadays 1 µF and even 10 µF at low voltages are readily available are reasonable cost. I would put a 1 µF ceramic at both the input and output of the regulator, physically placed as close as possible with short and direct traces to the regulator pins.
100 nF still has a bit better frequency response than 1 µF, but even the 1 µF of today are better than the leaded 100 nF of 20 years ago that this circuit was probably designed for. When you get up over 100 MHz or so, you have to look at these things carefully. For example, I used a specific model of 100 pF cap in a RF application once because it had the lowest effective impedance at the RF frequency of a variety of caps of higher values. However, this is a specialty issue. For something like a 78L05 regulator, just use 1 µF ceramic and be done with it.
They are used for filtering the noise and stabilizing the voltage. C1 filters the input, C2 and C3 improve the stability and transient response.