# How many forced sources?

**URL:** <https://www.rubin.community/t/how-many-forced-sources/8148>\
**Category:** Alerts & Brokers\
**Created:** [December 9, 2023, 10:02am UTC](https://www.rubin.community/t/how-many-forced-sources/8148 "2023-12-09T10:02:25Z")\
**Posts on this page:** 3\
**Page:** 1

<div class="post-metadata">

**Author:** ![roy](https://sea2.discourse-cdn.com/flex002/user_avatar/www.rubin.community/roy/32/1254_2.png) [@roy](https://www.rubin.community/u/roy)\
**Post date:** [December 9, 2023, 10:02am UTC](https://www.rubin.community/t/how-many-forced-sources/8148/1 "2023-12-09T10:02:26Z")

</div>

For DP0.2, the DiaObject table has 41 million objects, `DiaSource` has 162 million and `ForcedSourceOnDiaObject` 17 billion. Therefore each `DiaObject` has on average 4 `DiaSources` and 400 `ForcedSourceOnDiaObject`. The forced sources go back 4 years, so it comes to 100 per year per object.

When the real alerts start, brokers are expecting 10 million per night. If a year of forced sources is included for each alert, that 100 per alert, meaning a **billion** forced sources per night. Is that correct?

Can we just have the most recent forced sources? Those are the most useful.

---

<div class="post-metadata">

**Author:** ![MelissaGraham](https://sea2.discourse-cdn.com/flex002/user_avatar/www.rubin.community/melissagraham/32/2549_2.png) [@MelissaGraham](https://www.rubin.community/u/MelissaGraham)\
**Post date:** [December 11, 2023, 5:26pm UTC](https://www.rubin.community/t/how-many-forced-sources/8148/2 "2023-12-11T17:26:23Z")

</div>

Hi @roy, I agree that a twelve month history for any location in the sky covered with the wide-fast-deep strategy would include on average ~80 visits (over all filters). With the estimate of 10 million alerts per night, that would be 800 millions forced sources transmitted, so yes about a billion seems accurate to me too.

The option to remove histories from the alert packets was considered in Section 2 of Data Management Tech Note DMTN-248 ([Options for Alert Packets](https://dmtn-248.lsst.io/)), but was not recommended. The motivation for including historical records in alert packets is to enable brokers to assess the full time-domain event in order to make a robust classification (or prioritization) for follow-up, without the additional layer of complication and potential delay in having to query the Prompt Products Database for the history.

---

<div class="post-metadata">

**Author:** ![ryanlau](https://sea2.discourse-cdn.com/flex002/user_avatar/www.rubin.community/ryanlau/32/2837_2.png) [@ryanlau](https://www.rubin.community/u/ryanlau)\
**Post date:** [December 12, 2023, 5:34pm UTC](https://www.rubin.community/t/how-many-forced-sources/8148/3 "2023-12-12T17:34:41Z")

</div>

Hi @roy, I hope Melissa’s response addressed your question. But let us know if you have any follow-up questions. This is probably just different perspective, but I tend not to think of it as a billion forced sources per night, but rather that the light curve history (for a year) of each alert will be available in each alert packet. As Melissa mentioned, this is helpful for brokers to try to classify the object based on its past variability.