It is not so much the wills and woulds as the "solves" that makes it unclear which eventualities are hypothetical. In the context of present-form narrative, the indicative implies an actual eventuality, which in turn implies that a) Mary perseveres in her determination to kill herself and b) the consequent murder is also actualized.
If that is the case, either will or would is OK. Irrealis would implies a momentary hesitation while Mary considers the possible consequences, while realis will signals that Mary's determination to kill herself does not flag and she regards the consequence as inevitable until she comes up with a "solution".
But your afterword and the original form of your question indicate that Mary's resolution is weakened by her realization. If that is the case, then you should signal that with irrealis would and should recast "solves" as a hypothetical:
She realizes that she would leave her daughter behind, who would (have to) grow up without a mother. She might solve this problem by going upstairs and murdering her daughter . . . But that's no solution . . .
Note that the have to does not affect the reality of the clause: it merely introduces a layer of epistemicity to the modality.
Note, too, that the -ing forms are not affected in either case: these are gerund-participles, non-finite forms which are not marked for either tense or modality.
I see now that you inserted (das is doch keine Lösung) in order to cast "solves" as hypothetical. I'm afraid that doesn't work in English†: the verbform/construction has to "agree" with the modality of the clause.
† I'm not real comfortable with it in German either; but that may be a deficiency in my understanding of German idiom.
