Technology Behind MyAstrology

The VSOP87-controlled engine

The system has a single master: the VSOP87 analytical engine. Everything else is either a cache to make it faster or a reference to check it against.

                      VSOP87 engine
                     (master control)
                            │
        ┌───────────────────┼───────────────────┐
        ▼                   ▼                   ▼
   Apr 2025–Dec 2031    Before Apr 2025     After Dec 2031
   JPL DE441             VSOP87             VSOP87
   lookup table          direct             direct
   (panjika only)        (fallback)         (fallback)

Why this design

NASA JPL's DE441 ephemeris gives the highest-precision planetary positions available, but a table of it covers only a limited date range. The design decision is:

• VSOP87 remains the master. It can compute any date from ~1800 AD to ~2200 AD independently. The system does not depend on the table to function.
• A DE441 table is used by the panjika for 1 April 2025 – 1 January 2032. This makes the daily panjika faster. The birth-chart and other calculators always use the VSOP87 engine.
• Fallback is automatic. If a date falls outside the cached range — or if the cache is ever lost — the VSOP87 engine computes it directly. The system continues to work even without manual intervention.
• The cache is extendable, not essential. More years of DE441 data can be added at any time. But the absence of that addition does not break anything.

The Bishuddha Siddhanta computation rules

Every value in the panjika follows the classical rules as printed in the panjika itself.

The JPL DE441 table

NASA JPL's DE441 ephemeris is a reference source for planetary positions in the modern era — the JPL DE series is used for spacecraft navigation. For a limited date range, the panjika reads a precomputed DE441 table instead of computing.

How the two cooperate

For a date inside the table's range, the panjika reads the graha positions from the table. For a date outside it — past or future — the VSOP87 engine computes them directly. The two are not blended or compared at run time.

How the engine is verified

An ephemeris engine is only as good as its verification. VSOP87 output is compared against three birth charts (Kolkata, 8:00 AM, in 1985, 2000 and 2026) computed by two independent astrology programs — Prokerala (online, two charts) and Parashara's Light (desktop, one chart). The verification is a standing task — npm run verify-ephemeris-accuracy — which reports any regression before deployment.

Largest difference per graha (22 positions, 3 charts; measured October 2026)

Comparison with Swiss Ephemeris (401 dates, 1950–2050)

Since October 2026, the kundali engine's sidereal longitudes are also compared with Swiss Ephemeris 2.10.03 on 401 dates between 1950 and 2050. This check runs automatically whenever the engine changes. The table above records the earlier chart comparison and is kept as it was.

"আমাদের দৃষ্টিতে জ্যোতিষ জগতে কোনো প্রতিযোগী নেই, কেবলই রয়েছেন গুরুজন। কারণ, এই বিশাল শাস্ত্রে প্রতিটি প্ল্যাটফর্ম, প্রতিটি ব্যক্তির কাছ থেকে আমরা কিছু না কিছু শিখেছি। একজন শিষ্যের একমাত্র আকাঙ্ক্ষা থাকে—গুরুর দেখানো পথকে আরও সুপ্রশস্ত করে ভবিষ্যৎ প্রজন্মের কাছে পৌঁছে দেওয়া। সেই পথেই আমরা এগিয়ে চলেছি।"

In our view there are no competitors in astrology — only teachers. We have learned something from every platform and every person in this vast discipline; the comparison below is a student's measurement, not a contest. — Dr. Prodyut Acharya

One chart, seven programs (compared by Dr. Prodyut Acharya)

2 October 2026, 11:50 AM IST, Kolkata (22°34′ N, 88°22′ E). Each program was set up for this birth and its sidereal positions were read off by hand; the difference from Swiss Ephemeris 2.10.03 is shown in arcseconds. This is a single chart — a program's own settings (ayanamsa variant, true or mean node) can account for part of a difference.

The full readings behind the summary above — every position as each program showed it, so anyone can check a difference for themselves. All values are sidereal (Lahiri), degrees within the rashi.

All references at a glance

Every program and printed source that the engine has actually been compared with, and the largest difference found. Software not listed here has not been compared yet.

Known limitations

No computation engine is perfect. Below are the limitations we are aware of, stated honestly.

• !
  Measured range: 1950–2050

  The full VSOP87 series are accurate to about one arcsecond over several thousand years (Bretagnon & Francou, 1988), but the engine uses truncated series for the Sun, Moon and Mercury, and our Swiss Ephemeris comparison covers 1950–2050 only. The panjika can be computed for 1200–2000 BS (1793–2593 AD); for years far outside 1950–2050 the errors have not been measured.

• !
  How close each planet is to Swiss Ephemeris

  Since October 2026, Venus, Mars, Jupiter and Saturn are computed from the full VSOP87D series (with light-time, aberration and nutation). Compared with Swiss Ephemeris 2.10.03 on 401 dates, they differ by no more than 6 arcseconds. The Sun differs by up to 8″, the Moon by up to 26″ and Mercury by up to 75″, because those three still use shorter series. Positions are computed in Terrestrial Time: the engine adds ΔT (about 69 seconds today) to clock time, as Swiss Ephemeris and NASA JPL do — before this, the Moon differed by up to 64″. Every sign and nakshatra matched. A planet sitting within about a minute of arc of a boundary can still fall on either side. This comparison is re-run automatically whenever the engine changes. (Before this update, Venus differed by up to 560″.)

• !
  Sunrise convention was different in earlier engine versions

  Before the sunrise convention was aligned with the Bishuddha Siddhanta tradition (upper limb, refraction, elevation), the engine's sunrise differed from the printed panjika by around 2 minutes. This is documented in the 2020 worked example above. Current versions match to the second.

• !
  Festival definitions follow the Bishuddha Siddhanta tradition

  Some panjika traditions (e.g., Drik Siddhanta) define festival dates by slightly different rules — particularly for festivals that depend on sunrise or midnight-crossing. Where these traditions differ, MyAstrology follows Bishuddha Siddhanta. Users of other traditions should verify against their own panjika.

• !
  Checked against independent references — open to outside audit

  The engine is compared automatically with Swiss Ephemeris (planets) and NASA JPL ephemeris points (panjika table), and against charts from independent astrology programs. These checks are run by us; no outside party has yet audited the code. We welcome researchers who would like to audit the engine and publish their findings.