Accuracy & Testing¶

Testing Strategy¶

vedic-calc uses a multi-layer testing approach to ensure calculation accuracy across all 14 modules.

Layer 1: Cross-Validation Against Raw Swiss Ephemeris¶

The most important accuracy test. We compute planetary positions two independent ways:

1. Through calculate_chart() (our full code path)
2. Through raw swe.calc_ut() calls (bypassing our abstraction)

If both produce the same result (within 0.01 degrees), our code on top of pyswisseph is correct.

# From test_accuracy.py
def test_all_planets_match_raw_swisseph(self):
    """Every planet's longitude should exactly match raw pyswisseph."""
    # Compute via raw pyswisseph
    result = swe.calc_ut(jd, swe_planet_id)
    raw_sidereal = (result[0][0] - ayanamsa) % 360.0

    # Compare against our calculate_chart()
    chart = calculate_chart(...)
    our_longitude = chart.planets[planet].longitude

    assert abs(raw_sidereal - our_longitude) < 0.01

This catches bugs in:

• Ayanamsa subtraction
• Ketu derivation (Rahu + 180 degrees)
• Ascendant calculation
• Any sign/nakshatra arithmetic errors

Layer 2: Astronomical Fact Checks¶

Verify against known astronomical events:

These are facts any astrologer or astronomer would recognize -- if they fail, something fundamental is wrong.

Layer 3: Arithmetic Consistency¶

For multiple birth dates across different years, verify:

• sign == floor(longitude / 30) + 1
• nakshatra == floor(longitude / 13.333) + 1
• degree_in_sign == longitude % 30

This catches off-by-one errors, rounding issues, and edge cases at sign/nakshatra boundaries.

Layer 4: Cross-Validation Against PyJHora¶

Comparison tests validate our results against PyJHora (an independent Vedic astrology implementation) for:

Layer 5: Feature-Specific Unit Tests¶

Every module has dedicated tests covering:

• Edge cases -- sign/nakshatra boundaries, zero-degree placements, retrograde transitions
• Known charts -- verified against published examples in classical texts
• Round-trip consistency -- e.g., dasha periods cover exactly the expected lifespan
• Structural validation -- all returned models have valid field ranges

Why Swiss Ephemeris?¶

The Swiss Ephemeris is the industry standard for astronomical calculations:

• Accuracy: < 0.001 arcseconds (far beyond what astrology needs)
• Based on NASA JPL Development Ephemeris (DE431)
• Used by professional observatories and all major astrology software
• Maintained since 1997 by Astrodienst (Zurich)
• The same engine powers JHora, AstroSage, and most commercial Vedic astrology software

What About Ayanamsa Accuracy?¶

The ayanamsa (precession correction) is computed by pyswisseph using the same algorithms as the Swiss Ephemeris C library. For Lahiri specifically:

• Defined by the Indian Calendar Reform Committee (1956)
• Based on the position of the star Spica (Chitra) at the autumnal equinox
• The official ayanamsa of the Indian government
• pyswisseph's implementation matches the standard precisely

Different ayanamsa modes (Lahiri, Raman, KP) will produce slightly different sign placements for planets near sign boundaries. This is expected and correct -- it is a matter of which ayanamsa standard you follow, not a calculation error.

Running Tests¶

# Run all tests
uv run pytest -v

# Run only accuracy tests
uv run pytest tests/test_accuracy.py -v

# Run comparison tests (cross-validation against PyJHora)
uv run pytest tests/comparison/ -v

# Run with coverage
uv run pytest --cov=vedic_calc --cov-report=term-missing

Current Test Count¶